UBS is going the big bear today. First from the always excellent Jonathon Mott:
Slowing mortgage credit growth The Australian housing market faces a number of substantial headwinds. We believe these are likely to lead to a steady slowdown in the market over coming years as the impact of record high levels of household debt weighs on the economy. As a result we expect housing credit growth to continue to slow over coming years. This is consistent with the end of the housing leveraging boom which has lasted since the 1940s.
We use a housing fund-flow model to forecast credit growth, similar to forecasting FUM for an asset manager. This is consistent with the disclosure provided by most of the majors. This has five moving parts: (1) New mortgage fundings or loans drawndown; (2) Redraws and accrued interest; (3) Principal repayments – both scheduled and in advance; (4) Property sales; and (5) External refinancing.
2. Tightening of lax mortgage underwriting standards
We have been concerned for some time about weak mortgage underwriting standards across the banks (refer to our recent reports UBS Evidence Lab – $500bn in Liar Loans and UBS Evidence Lab – Liar Loans #2: Interest Only). This is consistent with APRA’s ongoing focus on sound lending standards whereby it is paying close attention to two main areas: living expenses and knowledge of borrower’s financial commitments and total indebtedness.
With regard to living expenses APRA has reiterated its concern that lenders have become overly reliant on expense benchmarks, in particular the Household Expenditure Measure (HEM). Given the HEM benchmark only incorporates a very modest (frugal) level of living expenses APRA is concerned whether it provides a realistic assessment of the borrowers true living costs.
We believe it is very likely lenders will be required to incorporate much more substantial estimates of living expenses into affordability calculations going forward. This is likely to imply a lower Net Income Surplus (NIS) which reduces the amount the lenders are able to lend to borrowers.
In addition, the introduction of mandatory comprehensive credit reporting (CCR) from next year should enable the banks to see each borrower’s total debt position. This will enable the banks to assess each customer’s total Debt to Income (DTI) position as well as the Loan to Income (LTI). This is consistent with movement by regulators around the world.
We see these moves by APRA to continue to focus on sound lending standards as significant. If banks are required to collect more accurate estimates of borrowers’ income (requiring tax returns to be provided appears an obvious starting point) and use more realistic estimates of customers’ household expenditure beyond the HEM benchmark, this may lead to lower levels of estimated Net Income Surplus. A more complete view of customers’ borrowings is also essential.
The implication of this is that it is likely many borrowers will be offered less credit than they would have otherwise received in the past which will have an impact on total lending commitments, credit growth and prices in the housing market.
3. House prices falling in Sydney and peaking in Melbourne
The housing market in Sydney has now entered a period of correction as the housing bubble seen up until mid-2017 comes to an end. During the last three months house prices in Sydney have now begun to decline. According to the CoreLogic Home Value Index, Sydney house prices peaked in August/September and have fallen by 2% since this time. Sentiment in the housing market has turned negative with the Auction clearance rate now in the mid-to-low 50s while a substantial number of house are being withdrawn pre auction.
The housing market in Melbourne has been more resilient given greater affordability driving strong population growth and less reliance on investment property lending than in Sydney. However in recent weeks the auction clearance rate has fallen back to the mid 60s and house price growth has stalled. We believe this indicates a peaking of house prices in Melbourne.
Resulting in slowing of mortgage drawdowns (fundings)
We believe this combination of tightening lending conditions by the banks and softening sentiment in the housing market will lead to an ongoing slowdown in new mortgage drawdowns (fundings).
Using the data provided by the major banks (which is highly correlated with ABS housing commitments data) we expect mortgage fundings to continue to soften over the next few years as these factors take effect. However, we have incorporated a more prolonged slowdown in new mortgage fundings rather than a short, sharp correction in lending volumes which is usually associated with higher interest rates.
Accelerating repayments as customers’
Interest Only periods mature Alongside mortgage fundings (drawdowns) the key driver of credit growth is the run-off or amortisation rate across the system. The mortgage run-off rate has been relatively stable over the last few years, ranging from 12.6% in FY11 to 17.2% in FY15 (as a percentage of the opening book).
While higher levels of household debt and subdued income growth may suggest the paydown rate is likely to fall (leading to higher credit growth), we believe the maturity and active switching of Interest Only mortgages to Principal and Interest is likely to lead to a rise in the paydown rate over coming years. We estimate the level of Interest Only mortgages on the banks’ books will fall from around 40% to lower than 20% over the next five years as the banks implement APRA’s macro prudential tightening. However, the exact impact of switching is hard to estimate as many Interest Only customers have already actively paid down their loans.
Overall we have assumed the mortgage paydown rate rises by 1% to 16.7% over the forecast period.
Consumer leverage cycle coming to an end
One of the key concerns of regulators and investors in Australia is the very high level of household leverage which has developed over decades. According to the latest RBA data, Household Debt to Disposable Income reached 194% in June 2017.
Based on our new forecasts for household credit growth and our Economics Team’s estimates of growth in household disposable income, we estimate that household debt to disposable income will peak at 200% in mid-2018 before slowly deleveraging.
We see this as a change in a very long term trend with the Australian economy having leveraged up since the end of the 1940s, post WWII.
However, unless interest rates are cut further and bank underwriting standards loosened we believe the deleveraging cycle looks likely to begin.
Next, consumer smash from George Tharenou:
Households running out of savings – cutting the consumer further <mkt: RBA on hold until 2019
UBS has long held a below-market view that the consumer would slow ahead, rather than pick-up as consensus and the RBA expected. This has been a key driver of our dovish view of a downward wage-price spiral that would see the RBA keep the cash rate on hold until at least Q4-2018.
Indeed, our view has been strongly reinforced by the subsequent slump in nominal retail sales growth over recent months to under 2% y/y (Figure 2). Importantly, more recently this also spread to a ~record low trend of nominal consumption of 3½% y/y (Figure 3). Furthermore, despite weak inflation, real consumption volumes in Q3-17 also dropped to only 0.1% q/q, the worst quarter since the GFC. This dragged the y/y to 2.2%, which is only a tick above a ~4-year low (Figure 4).
That said, given business surveys indicate some rise in their wage bill (summing employment and wage rates, Figure 8), & hiring intentions remain ~consistent with ~2%+ y/y employment growth ahead (Figure 9), we already factored into our outlook a lift in WPI towards 2½% y/y next year, & a recovery in average earnings.
Household cash flow collapsed to a record low in 2017
For now, household cash flow in 2017 has been crunched down to a record low of only ~1½% y/y (Figure 10 and Figure 11). In addition to weak wages income, there have also been significant drags on household cash flow from the combination of higher taxes (i.e. mainly via ‘bracket creep’ of higher average tax rates), as well as a surge in both utilities and petrol (almost entirely driven by higher prices rather than volumes); while no additional RBA rate cuts meant a smaller fall in interest payments compared with prior years.
Discretionary/retail areas of spending are taking the hit
This weakness of household cash flow particularly hurt retail sales (Figure 10), & ‘discretionary’ categories within consumption. In the year to Q3, real consumption was weakest in hotels & restaurants (despite booming tourism, Figure 12); while nominal consumption was weakest mainly in ‘discretionary’ categories, especially clothing & footwear and household goods (Figure 13). That said, large retailers – which are likely mostly listed companies – are still growing at ~4% y/y, which is significantly faster than small retailers which slumped to -3% y/y (Figure 14).
Consumer outlook – first the positives, better global growth and strong business conditions + tax cuts?
Looking ahead, given strong business surveys and global growth, we expect some recovery of wages and average earnings. Indeed the Government indicated the prospect of household income tax cuts ahead, so we have also allowed for some modest tax cuts in 2018 & 2019 to avoid a repeat of the surge in taxes paid in 2017, albeit we don’t expect a material fiscal stimulus to lift the consumer outlook.
Also on the positive side, consumer sentiment bounced in recent months, with Dec-17 around the highest level in recent years, and marginally above average (Figure 15). At face value this suggests a decent pace of real consumption ahead.
Consumer outlook – return of the cautious consumer?
However, at the same time, there is some evidence of a return of ‘consumer caution’, with the proportion of respondents saying the wisest place for saving is repaying debt/mortgage trending up to a high share (Figure 16). This suggests limited scope for the household savings rate to drop sharply further ahead.
This concern about repaying debt likely reflects the surge of the household debt-income ratio to a record high of ~195%. While interest payments as a share of income are ‘manageable’ at ~8½% (Figure 17), this reflects a ~record low level of interest rates currently, and households have never been more sensitive to a hike.
Household cash flow doesn’t add up – we cut consumption
However, our detailed household cash-flow model just doesn’t add up to anything but a weaker consumer outlook. Even with steady interest rates, the rising and record stock of household debt means interest payments lift. Hence, our previous expectation of one RBA hike in 2018 would put too much pressure on disposable income. Indeed, the consensus view of faster real consumption would likely require an unrealistic further collapse in the household savings rate to fund the spending.
Hence, we cut our outlook for real consumption to only 2.0% y/y in 2018 & 2019 (see chart on page 2), to be further below consensus which has a bounce to 2.4% in 2018 & 2.5% in 2019. But with positive offsets from a likely ongoing domestic public (infrastructure) investment boom, and business investment turning up amid stronger global growth, we still expect Australian real GDP to recover modestly to 2.7% y/y in 2018 and 2019, albeit still below consensus of 2.8%/2.9%.
Although the RBA recently materially downgraded their growth outlook again, they also still expect consumption to “pick up gradually”, albeit the RBA now concede “significant uncertainty” for income growth. Hence, the RBA are likely to again downgrade their real GDP outlook in the Feb-18 SOMP, given they still expect a persistent boom of 3¼-3½% y/y from end-2018 through to end-2019.
Full impact of macroprudential policy tightening still ahead
Another key factor for our more negative consumer view is the lagged impact of macroprudential policy tightening. In Q3-17, the flow of home loans on interest-only (IO) terms as a share of total loans collapsed to a ~multi-decade low of 17%, which is now well below APRA’s cap of 30% announced in Mar-17 (Figure 18). Hence, there will likely be a consequent reduction in borrowing capacity from fewer interest-only home loans ahead. This will be compounded by APRA now starting to focus on tightening lending benchmarks in 2018, with APRA finding it “open to question” if “prominent” use of benchmarks for living expenses “always provide a realistic assessment” (Figure 19). This is ~consistent with UBS research.
‘Phase 3’ of macroprudential policy to come in 2018?
APRA also found the net income surplus (NIS) – or lender’s assessment of surplus income borrowers have left over after living expenses, debt repayments & some buffers – was “limited” for a “reasonable proportion of new borrowers”. Specifically ~17% of new lending occurred with a NIS of <$200/month, and ~3% of lending had a negative NIS (Figure 20). In addition, mandatory reporting of comprehensive credit reporting (CCR) from 2018 should enable lenders to see borrower’s total debt position or more accurately assess total debt-to-income.
Effectively, we think that APRA is flagging ‘Phase 3’ of macroprudential policy tightening in 2018. We think that APRA’s new focus on raising the floor of assumed minimum living expenses is likely to see lenders reduce the maximum borrowing capacity of some borrowers, relative to the last few years.
IO to P&I switching impact has been relatively small, but set to become much bigger over time
Notably our household cash flow model doesn’t include the direct impact of principal repayments of debt. Hence there is an additional headwind that will build over time from the lagged impact of macroprudential tightening, because there is now an ongoing lift in switching of mortgage repayments from IO to P&I.
APRA estimates the stock of ADI’s IO loans outstanding dropped by ~$36bn in the six months to Q3-17. We estimate that the additional household cash-flow hit from switching to P&I was only ~$1-2bn (given some ‘buffer’ from the rundown of mortgage offset accounts), or just 0.1%-0.2% of total household sector income.
But there is still likely to be a notable and increasingly negative impact over time. Those individuals switching from IO to P&I do face a material ‘shock’ with scheduled mortgage repayments jumping (effectively overnight) by 35%-50%. UBS banks team estimate the potential expiry of IO loans in coming years (assuming a 5-year maturity and no rollover to another IO term) will surge up to ~$105bn in FY18, ~$133bn in FY19, and peak at ~$159bn in FY20 (Figure 21).
Indeed, UBS Evidence Lab research found that the majority of borrowers under financial stress from higher interest rates would cut consumption, while a significant share would sell their property (Figure 22).
Falling home loans to drag credit growth and house prices
Given this backdrop of a likely tightening in lending standards ahead, detailed modelling by UBS banks teams shows that home loans (values) are likely to fall over coming years. This, combined with a pick-up of mortgage repayments as borrowers switch from expiring interest-only (IO) home loans to principal and interest (P&I), will see housing credit growth slow sharply from >6% y/y now, to a record low towards ~3% y/y in 2019 (Figure 23).
Reflecting home loans provide a good lead indicator for house prices (Figure 24) – & investor loans are likely to retrace further after ~doubling in this cycle (Figure 25) – we now expect that house price growth will weaken further ahead to ~flat in 2018 (revised down from 0% to +3%), and probably fall in 2019 (0% to -3%).
While overall consumer sentiment is decent, we note that specific attitudes towards real estate as the ‘wisest place for saving’ remain weak, after collapsing to a record low share recently (Figure 26). Similarly, ‘time to buy a dwelling’ has edged up in recent months, but is still stuck around 20% below its long-run average (Figure 27).
Ongoing record housing completions drag on prices ahead
Given the persistent upside surprise of residential building approvals in recent months, we are revising up our forecast for dwelling commencements in 2017 to 210k (was 205k), but we still expect a sharp correction ahead to 185k in 2018 and 175k in 2019. Nonetheless, the record pipeline of approvals, commencements and dwellings under construction means that dwelling completions will still remain around a record high level of ~220k in the coming year, which is far above the long-run average of ~150k. This is likely to put some upward pressure on the residential rental vacancy rate, and hence lead to some easing of house prices ahead (Figure 28).
That said, despite the already record high level of completions, vacancy rates have surprisingly fallen over the last year – seemingly reflecting the incredible ‘people growth’ which boomed close to ~3% y/y (Figure 29), much faster than population growth which also re-accelerated to a strong 1.6% y/y. That said, this strong demand for housing has still not stopped house price growth weakening towards flat over the last 6 months.
Foreign demand for housing has peaked and started falling
Also underpinning our view of weaker house price growth ahead is that the peak of foreign demand for residential investment has likely already occurred (Figure 30) – with the largest source of total real estate investment coming from China (Figure 31). The value of FIRB approved investment in housing skyrocketed from $17bn in 12/13 to $72bn in 15/16. However, comments by Treasurer Morrison highlighted that the number of FIRB approvals slumped to around 15k this year, down sharply from 40k last year. The weaker trend of foreign demand for housing is likely to continue ahead given the lagged impact of prior tax hikes on foreigners, and the further lift for several States will only come into effect from 2018
Weaker house prices suggest consumption unlikely to lift
Hence, the overall implication is the prior large (indirect) boost to consumption growth from the ‘household wealth effect’ – via a collapsing household savings rate – is likely to fade ahead. This will see growth in nominal spending converge to nominal income. So unless there is a surprisingly strong surge in wages, or a further drop in inflation, then real consumer spending will likely slow ahead.
Slower consumption could put downward pressure on CPI
Importantly, our weaker consumer outlook suggests ongoing low inflation pressure ahead, and hence some downside risk to our underlying inflation forecast which ticks up to 2% y/y or just above in the coming year. However, the RBA’s underlying CPI forecasts are already quite dovish, actually easing to 1¾% y/y in the coming year, before returning to 2% y/y in 2019.
Interestingly, the consumption deflator – which is GDP-basis consumer prices weighted by real-time consumption shares (~equivalent to the US PCE) – was only 1.1% y/y in Q3-17. This remains far below headline CPI at 1.8% y/y (Figure 32). This suggests the technical impact of CPI re-weighting – which takes effect over the next year from Q4-17 CPI onwards – will actually drag on headline CPI by much more than our initial estimate of ~0.2%pts, and instead subtract around ~0.4%pts.
RBA to remain on hold for even longer… until 2019
So overall, we also now see the RBA on hold for even longer until Q1-2019 (delayed from our long-held view of Q4-18); before the RBA ‘lags the Fed’ with two rate hikes of 25bp each in 2019 – with UBS expecting that the Fed will hike rates by 25bp ~6 times by end-19. But we continue to worry that if the RBA hikes too much and/or too early, they risk turning an orderly housing correction/soft landing – as appears to be unfolding at the moment – into a housing collapse. In the latter case, we would expect the unemployment rate to lift towards 6%+ (rather than our base case of a modest further decline towards 5%), which could trigger a bad debt cycle (rather than our still very benign base case), and potentially lead to a self-reinforcing negative feedback loop between asset prices and the real economy.
I have only one thing to add. If mortgage growth falls to zero in the years ahead then rate cuts will flow and it will not matter.