How we ranked and tested what moves each credit product
The data, the weights, the tests and the assumptions behind What moves each credit product.
General information, not legal or financial advice.
The piece ranks what moves each product by judgement, checks that ranking against what happened, and tests whether the official drivers would have predicted credit since 2017. Every figure below is computed from official, openly licensed data by code first tested on invented data with known answers (RBA; ABS; APRA).
The data
- Credit growth by product: RBA table D1, adjusted by the RBA for series breaks, to June quarter 2026; levels from D2 (RBA).
- Credit card balances: RBA table C1. It is not adjusted for breaks, so a year’s growth is left out when it spans a break the RBA lists (2002, 2008, 2013, 2018, 2024, 2025); the latest usable quarter is March 2024 (RBA).
- New car loans: ABS Lending Indicators, fixed-term personal loans for road vehicles. New lending, not a stock, so it swings by tens of points a year (ABS).
- Housing arrears: APRA’s housing loans non-performing, from March 2019, the only product APRA publishes by performance (APRA).
- Drivers: the cash rate and discounted variable mortgage rate (RBA F1.1, F5), unemployment, CPI, the wage price index (RBA H4), the terms of trade (RBA H1), population and net overseas migration (ABS), and house prices: the ABS eight-capital index to December 2021, when it was discontinued, then the ABS mean dwelling price (ABS). The two moved together at 0.97 over 2012–2021.
- Not available as open data: arrears on credit cards, personal loans, car loans, small loans or buy now pay later. Their rankings rest on the episodes and the RBA’s research, not on a fitted series.
The weights, with their evidence
Each product’s weights are ours, out of 100. They rank how much of the movement in lending, or in arrears, each driver has explained in the episodes and research cited (RBA D1, D2, C1; FSR October 2026).
| Product and driver | Lending | Arrears | Strongest evidence |
|---|---|---|---|
| Owner-occupier: interest rates and serviceability | 25 | 25 | Growth fell from 9.2% (December 2021) to 5.4% (June 2023) through the rate rises; arrears rose with repayments to income |
| Owner-occupier: house prices and equity | 20 | 15 | Loan sizes track prices; a 20% fall would put about 5% of borrowers in negative equity (FSR October 2026) |
| Owner-occupier: population and migration | 15 | — | Household formation; migrants rent first, buy later |
| Owner-occupier: jobs and income | 10 | 35 | Arrears start with lost income; losses need negative equity as well (RDP 2020-03) |
| Owner-occupier: buyer schemes, macroprudential, inflation | 18 | 20 | 5% deposit scheme widened October 2025; debt-to-income limit not binding (3.7% against 20%) |
| Investor: tax settings | 25 | — | New commitments fell “driven by investors” after the May 2026 Budget (RBA, August 2026) |
| Investor: interest rates | 20 | 25 | Higher debt to income than owner-occupiers |
| Investor: expected price growth | 20 | 15 | Growth peaked at 10.8% (2015) in a boom, fell to −0.7% (July 2020) |
| Investor: macroprudential limits | 15 | — | −6.2 points after the 2014 benchmark; 0.5% by June 2019 after the 2017 interest-only limit |
| Investor: rents and vacancies; unemployment | 10 | 40 | Yields and tenants’ ability to pay |
| Credit cards: substitutes | 30 | — | Debit is about half of consumer payments, cards about a quarter (RBA payments review, June 2026) |
| Credit cards: lending rules and caution | 20 | 15 | Balances paying interest fell from $31.8bn (2018) to $28.7bn (2019), before COVID |
| Credit cards: spending; cash flow | 35 | 35 | Purchases rose by half since 2019 while balances paying interest stayed flat |
| Credit cards: unemployment | — | 40 | Unsecured: losses follow the first trigger |
| Personal loans: substitutes | 25 | — | Home equity and redraw replaced personal loans (FSR April 2018) |
| Personal loans: confidence; rules; rates; competition | 65 | 30 | −13.0% in the year to October 2020 (monthly), the lowest on record; still below the July 2019 peak |
| Personal loans: unemployment; real incomes | — | 60 | Arrears around pre-pandemic levels (FSR March 2026) |
| Car loans: supply and prices | 25 | 15 | Chip shortages lifted car prices 7.4% in a year to June 2021 (RBA) |
| Car loans: tax settings for leases | 20 | — | About 39% of electric car sales used the fringe benefits tax exemption (Treasury review, May 2026) |
| Car loans: incomes, rates, fuel; unemployment and standards | 40 | 75 | New car loans: −18.3% at the GFC low |
| Small and medium loans: regulation | 35 | 20 | After June 2023, small loans fell and medium loans rose from about 10% to over 30% of lending (ASIC REP 805) |
| Small and medium loans: stress, income support, substitutes | 50 | 65 | Renters’ deficits about 15% against 4% for mortgage holders |
| Buy now pay later: spending; regulation; merchants | 70 | 30 | $3.1bn (2017–18) to about $19bn (2022–23); credit law from 10 June 2025 |
| Buy now pay later: young households leaving cards, stress; cost of living, youth jobs and stacking | 20 | 55 | 21% of surveyed users missed a payment in a year (ASIC REP 672) |
| Small business: demand cycle | 25 | 30 | Business credit −6.0% (December 2009); 11.0% now |
| Small business: lenders’ appetite, rates, property, tax; input costs | 60 | 60 | Lower capital for small business loans from 2023; most small business lending secured on homes (RBA Bulletin, October 2025) |
Rows that group drivers carry their combined weight; the remainder of each product’s 100 is “other”: competition, disasters, insurance, ill health and divorce.
Would the drivers have predicted it?
For each product, a model of its yearly growth (housing arrears: its level) on its own value a year earlier and up to four drivers. A driver entered only at a lag where it pushed the way economics says it should, fixed before fitting; near-duplicates were skipped and any that changed sign were dropped. Fitted to December 2016 and tested from March 2017, given the actual drivers in the test years, so it tests the relationships, not a forecast of the drivers. Housing arrears were fitted on 2019–2023 and tested on 2024–2026 (RBA; APRA). The last column compares the model’s errors with “the same as a year ago”: above zero is better.
| Product | Drivers kept, share of what they explain | Fit, to 2016 | Fit, 2017 on | Against a year ago |
|---|---|---|---|---|
| Owner-occupier housing credit | house prices 45%, terms of trade 29%, unemployment change 26% | 0.89 | −0.65 | 0.06 |
| Investor housing credit | house prices 95%, cash rate change 4%, terms of trade 2% | 0.79 | −0.03 | −0.22 |
| Other personal credit | house prices 40%, inflation 30%, unemployment change 26%, terms of trade 4% | 0.69 | −0.40 | −1.01 |
| Credit card balances | inflation 47%, terms of trade 32%, house prices 20% | 0.87 | −0.16 | 0.15 |
| New car-loan commitments | unemployment change 37%, real mortgage rate 34%, terms of trade 22%, cash rate change 7% | 0.61 | −0.41 | 0.44 |
| Business credit | wage growth 45%, unemployment change 29%, inflation 19%, terms of trade 7% | 0.41 | 0.42 | 0.55 |
| Housing loans non-performing | inflation 59%, unemployment change 41% | 0.95 | −6.10 | 0.60 |
- Investor and personal credit lost to “a year ago”. The years 2017–2026 hold APRA’s interest-only limit, the Royal Commission, COVID income support and deferrals, and the RBA’s term funding: none of them moves a driver, all of them moved credit.
- Fit from 2017 is below zero for six of seven. That measure compares with the test years’ own average, which no model knows in advance.
- Signs that look wrong are history, not cause. Over 2004–2026, migration peaked as credit growth slowed and rates rose in booms, so taken one at a time migration correlates negatively with credit growth and rates positively. The tested models refuse such drivers.
- The samples are short: about 20 independent years, one or two cycles; housing arrears have 30 quarters and no recession but COVID, when deferrals hid arrears.
The paths to 2031 and 2036
The base runs the RBA’s August 2026 forecasts to December 2028, then unemployment settling at 4.25%, inflation at 2.5%, population growth easing to 1.1% and net overseas migration near 235,000 a year (Population Statement 2025; Intergenerational Report 2026). The cash rate, house prices and the terms of trade are our assumptions: cash rate 3.35% by 2031, house prices 4.5% to 5% a year. The downside: unemployment to 7% by mid-2028, house prices −9% a year in 2027–28, rates cut to 1.6%, migration halved. The upside: unemployment near 4%, house prices 6% to 8% a year.
What the fitted models give on those paths, as average yearly growth 2027–2031 (housing arrears: December 2031):
| Product | Base | Downside | Upside |
|---|---|---|---|
| Owner-occupier housing credit | 6.3% | 4.0% | 6.9% |
| Investor housing credit | 7.3% | 2.8% | 8.8% |
| Other personal credit | 1.5% | −3.4% | 3.3% |
| Credit card balances | 1.5% | −1.8% | 3.1% |
| New car-loan commitments | 3.9% | 5.4% | 3.9% |
| Business credit | 6.4% | 5.1% | 7.3% |
| Housing loans non-performing | 0.99% | 0.97% | 0.99% |
We did not publish these as the ranges, for three reasons (RBA; FSR September 2009):
- They are too gentle in a recession. A 2.5-point rise in unemployment adds about 0.1 point to housing arrears; in the GFC, banks’ impaired and past-due loans rose from 0.53% to 2.19%. New car loans dip to about −4% against −18.3% in the GFC, then rebound above the base as rates fall.
- They cannot see policy. The investor model does not know the tax changes from 1 July 2027, so its base of 7.3% is too high; we set 3% to 5%.
- They lost to “a year ago” for two products, so their paths are a check, not an answer.
The published ranges take the base from these models and the RBA’s forecasts, adjusted for policy already announced; the downside from the depth of past episodes (business credit −6.0% and personal credit −13.0% at their lows) and the RBA’s severe case (FSR October 2026); and widen both to the size of past misses. We will compare each with the outcome every year.
Sources
- RBA statistical tables D1 and D2, C1 and C1.2, F1.1 and F5, H1 and H4, retrieved 10 October 2026. CC BY 4.0.
- ABS, Lending Indicators; National, state and territory population; Residential Property Price Indexes; Total Value of Dwellings; Consumer Price Index. CC BY 4.0.
- APRA, Quarterly ADI Property Exposures, June 2026.
- RBA, Statement on Monetary Policy, August 2026; Financial Stability Reviews of October 2026, March 2026, April 2018 (Box B) and September 2009; Bulletin, October 2025; Review of payments system regulation, June 2026; Statement on Monetary Policy, August 2021.
- Treasury, Electric Car Discount review, final report (May 2026); Intergenerational Report 2026; Centre for Population, Population Statement 2025.
- ASIC, REP 805 and REP 672; Parliamentary Library, Bills Digest on buy now pay later.