Why Does a Model Count a Home as ‘Uninsured’ When the Premium Reaches Four Weeks of Income?

Geoffrey Chen

When Numbers Start Making Decisions · Season Five, “Things That Have Not Happened Yet” · Article 6

1. A home with a policy becomes uninsured inside a model

In 2026, the Australian Prudential Regulation Authority published an Insurance Climate Vulnerability Assessment. Five large general insurers representing around 80% of the home insurance market participated. The stress test covered roughly ten million detached houses and modelled risk and premiums to 2050. APRA: Mind the Gap

To estimate the home insurance protection gap, the test divided the modelled annual premium for each property by estimated weekly household income. When the premium equalled or exceeded four weeks of income, insurance was treated as unaffordable and the property was modelled as uninsured.

Four weeks is about 7.7% of annual income. At 3.9 weeks, a property remains insured in the scenario; at 4.0, it changes class. A real household may struggle to renew above the line or abandon cover below it. Why does a future stress test need such a boundary, and what may that boundary legitimately decide?

2. The threshold does not measure an individual choice

APRA did not cancel anyone’s policy or instruct insurers to refuse cover. Four weeks is a modelling assumption that converts continuous affordability pressure into a system-wide estimate of financial vulnerability.

The stress test asks how severe but plausible changes in climate hazards, construction cost and incomes might reduce effective protection and affect households, mortgage collateral and financial stability. Without a behavioural rule, the model could show rising premiums but not translate them into a protection gap.

“Modelled as uninsured” must therefore remain an internal scenario classification, not a factual finding about a particular household. That limit is central to the number’s authority.

The four-week measure also has a history outside APRA. The Actuaries Institute uses it as an indicator of affordability stress. It creates a common metric across regions and incomes. Reuse makes comparisons possible, but it does not transform the threshold into an observed law of household behaviour.

3. The premium is already a prediction

Home insurance is not priced only from last year’s claims. Insurers estimate the frequency and severity of fire, flood, cyclone and other losses during the next policy period, then add reinsurance, rebuilding cost, expenses, tax and capital requirements.

Address, terrain, building characteristics and mitigation increasingly enter property-level pricing. The ACCC’s Northern Australia Insurance Inquiry documented a shift from broad postcode averages towards address and building-level assessments, particularly for cyclone and flood.

An annual premium is therefore a monetary expression of a future loss distribution. A household receives a total price without necessarily seeing how much the model attributes to flood, wind, rebuilding inflation or reinsurance. A high premium may reflect an input error, a missing resilience feature or a sound assessment of serious exposure; the customer may struggle to distinguish them.

APRA then divides one prediction by another. Hazard and rebuilding forecasts shape the numerator; future household income shapes the denominator. A clean four-week boundary sits above several layers of uncertainty.

4. Why four weeks rather than three or six?

Any affordability threshold trades individual variation for comparability. Four weeks has intuitive force: devoting roughly a month of gross household income to home insurance indicates extreme pressure. It permits comparisons across prices, regions and income groups.

Gross income is not disposable income. Two households with the same gross amount can have different mortgages, care, health and debt obligations. A low-income but asset-rich retiree differs from a family with no savings. One household may lose effective access at three weeks; another may retain insurance at five by sacrificing other essentials.

The proper use of the line is to identify systemic exposure, compare scenarios and locate areas needing policy attention. It cannot by itself decide subsidy eligibility, loan default or whether a specific household should rationally stop buying cover. If government later connects assistance to a four-week line, it must reconsider the income definition, assets, scope of cover and tapering around the boundary.

5. The model intentionally assumes policy does not rescue the outcome

APRA uses a static-policy assumption: no additional public policy or physical adaptation is introduced in the scenario. This does not predict that governments will do nothing until 2050. It isolates the drivers of risk and asks what could happen without new action.

That conditional design is easy to misread. A reader may treat a stress number as a point forecast. An advocate may say it proves one policy inevitable; an opponent may call it unrealistic. Both ignore the “if”: if the severe scenario develops and if no further adaptation changes it.

The purpose of a stress test is often to make reality diverge from its result. Retrofitting, planning, public flood protection, clearer risk data and better risk-sharing can change losses and premiums. The ACCC’s 2026 monitoring found that the cyclone reinsurance pool moderated premium increases for homes and small businesses at medium and high cyclone risk, although affordability concerns remained.

If policy works and the 2050 outcome fails to occur, that may demonstrate that the model helped alter the future rather than failed to describe it.

6. Who disappears at the threshold?

The four-week classification can obscure at least three groups. First are households just below the line with inadequate cover. They may reduce the sum insured, raise the excess or remove contents cover. The ratio improves while more risk returns to the household.

Second are households still paying a very high premium by sacrificing necessities. Modelling them as uninsured understates current expenditure stress, although it may reasonably indicate that the position is not sustainable.

Third are homes that cannot obtain a quote. Without a premium, no ratio can be calculated, yet absence of an offer can be more serious than a high price. Availability, affordability, underinsurance and non-insurance should be reported separately.

Geographic and social distribution matters. High hazard often overlaps with low incomes, remote communities, First Nations communities and older buildings. More granular risk pricing may be actuarially accurate while assigning the full cost of historical location and future climate to people with little ability to change either.

7. Explanation, review and action

Individual customers should receive the main factors driving a price, the recorded characteristics of the property and recognised mitigation measures. They should be able to correct errors about address, roof, materials and flood features and learn whether improvements will affect renewal. Commercial models may remain protected, but factual inputs and the basic reason for a decision cannot all become a black box.

For APRA’s test, reviewers need the scenarios, four-week threshold, income and premium estimates, property scope and the behavioural assumption equating unaffordability with non-insurance. Policymakers should examine sensitivity at three, four and six weeks rather than let one line monopolise the story.

Action should reduce physical risk as well as price. Subsidies can preserve near-term cover while shifting escalating loss to public budgets. Compulsory relocation can fracture communities. Staged resilience work, public infrastructure, lending support, risk disclosure and targeted assistance need to form one adaptation path.

8. Future risk returns to the present through price

Insurability is not a property possessed by the house alone. It forms among hazard, repair capacity, income, insurance capital, infrastructure and policy. The premium is a present interface for those relationships; the four-week line turns it into a system category.

Government delegates some knowledge of household risk to insurers’ models. Delegation creates visibility and the danger of authority without understanding. If policy repeats only the estimated number of uninsured homes without the static-policy and behavioural assumptions, the model receives more power than the evidence warrants.

Risk maps and premiums are not fate. They can direct retrofit and public investment. Institutional failure occurs when a modelled future vulnerability becomes a reason to deny action rather than organise it.

Conclusion: use the four-week line as an alarm, not a household verdict

Four weeks of gross income should remain a common indicator of severe home-insurance affordability pressure. It is intuitive, comparable and supported by actuarial work and APRA’s stress testing.

Every report must also state that reaching the line does not prove a household has stopped insuring and falling below it does not prove adequate protection. Subsidy, lending or relocation eligibility cannot import the threshold without a separate justification. No-quote properties, underinsurance and multiple sensitivity lines should be visible.

The threshold is a smoke alarm, not a judgment on one home. It shows that the relationship among risk, price and income is approaching an unsustainable condition. Once the alarm sounds, insurers, banks, governments and communities—not the decimal—must decide how the future can be changed.

Primary sources

Discover more from Geoffrey Chen

Subscribe to get the latest posts sent to your email.