The risk of energy hardship increases with extreme heat and cold in Australia

Anthropogenic climate change is altering temperature distributions, increasing average temperatures and the intensity, frequency, and duration of extreme heat events, and in the short term intensifying extremes of cold in some regions. Even small increases in average temperature can substantially raise the frequency of extremes. These changes disrupt human and ecological systems and have social and health consequences, including energy hardship. Energy hardship (or energy poverty) arises from high energy costs or constrained access, low household income, and poor housing quality. Historically, heating demand dominated energy costs; however, recent heatwaves have increased cooling demand and shifted attention to climate change as a distal driver of energy hardship. Difficulties in affording essential energy services affect health and social outcomes via thermal discomfort, financial stress, and trade-offs in basic living costs. Prior research has focused more on cold and heating, with limited and mixed evidence on how climate change affects energy hardship through both heat and cold, and with few studies using metrics that capture intensity, duration, and frequency of extremes or considering acclimatisation. Energy hardship vulnerability is unevenly distributed; temperature changes do not uniformly translate into hardship, underscoring the need to examine how social vulnerability and adaptive capacities shape unequal impacts across people, housing, and neighbourhoods.

Prior work often defines and measures fuel or energy poverty in terms of difficulties achieving adequate warmth, centering on cold exposure. Evidence linking temperature and energy hardship in a changing climate is limited and mixed. Studies using electricity disconnections as outcomes report that extreme heat increases disconnections among low-income households in California and remote communities in Northern Australia. Limited studies directly linking climate change to energy hardship report both increases in energy poverty with warmer temperatures and potential alleviation in some contexts. Few analyses incorporate extreme temperature metrics that capture intensity, duration, and frequency (e.g., heatwaves/coldwaves) and account for short- and long-term anomalies and acclimatisation. The literature also highlights uneven vulnerability across social groups and places, suggesting that climate change could reshape inequities in energy hardship. This study responds to these gaps by jointly analysing extreme heat and cold metrics, their non-linear and interacting relationships with hardship, and heterogeneity by individual, housing, and neighbourhood characteristics.

Data: Individual- and household-level data were drawn from the nationally representative longitudinal Household Income and Labour Dynamics in Australia (HILDA) Survey, 2005–2021. Over 11,000 people aged 15+ were followed annually, yielding 30,005 individuals and 269,500 person-year observations. Energy hardship measures included: (1) the proportion of equivalised household disposable income spent on equivalised annual energy expenditure (electricity, gas, other heating fuels); (2) an expenditure-based indicator for high energy cost share (>10% of income); (3) a consensual/relative Low-Income High-Costs (LIHC) indicator (energy costs above national median and residual income after energy and housing costs below 60% of the national median); and (4) self-reported utility bill payment arrears.

Extreme temperature metrics: Weather station records were used to construct annual measures aligned with Bureau of Meteorology (BoM) standards. Metrics captured (a) intensity via average annual maximum and minimum temperatures; (b) frequency via the annual number of heatwave and coldwave events; and (c) duration via the average length (days) of heatwaves and coldwaves per year. Heatwave and coldwave identification followed the Excess Heat Factor (EHF) and Excess Cold Factor (ECF) framework developed by the Centre for Australian Weather and Climate Research, incorporating both long-term anomaly indices (e.g., excess heat significance) and short-term acclimatisation indices. Positive EHF indicates heatwaves; negative ECF indicates sustained cold periods relative to local climate baselines.

Statistical analysis: The study estimated relationships between extreme temperature measures and annual energy hardship using multivariate fractional polynomial (FP) models to flexibly capture non-linearities. Linear regressions were used for continuous energy expense share; logistic regressions for risks of >10% expense, LIHC, and utility bill arrears. Models included FP terms for extreme heat and cold metrics and their interactions, year fixed effects, and clustered standard errors at the individual level. Effects were reported as average marginal effects at specified levels of heat/cold exposure.

Heterogeneity: Inequality in impacts was examined by stratifying models across: individual-level factors (age groups; household structure such as couples, lone parents, lone persons, group/multifamily); housing factors (housing quality; tenure—owner-occupied vs rental); and neighbourhood factors (greenspace coverage per km²: ≤10%, 11–40%, >40%; density of small-scale renewable energy installations per km²: ≤1, 2–40, >40; and climate zones per the National Construction Code). Three-dimensional heatmaps displayed joint distributions of hot/cold metrics and hardship outcomes.

Projections: Future changes in hardship risks were simulated by combining (1) estimated FP relationships between hardship and average annual max/min temperatures; (2) mean baseline risks from the sample; and (3) out-of-sample predictions using climate model projections (CSIRO and BoM) of average annual max/min temperatures for 20-year periods centered on 2030, 2050, 2070, and 2090. Two Representative Concentration Pathways (RCPs) were considered: RCP 4.5 (moderate emissions reductions) and RCP 8.5 (high emissions). Projections reported changes in risk for two hardship measures: >10% energy expenditure and utility bill arrears.

Overall relationship: Energy hardship risk increases with the intensity, frequency, and duration of both extreme heat and cold. Interactions indicate that the effect of higher average maximum temperatures is larger in colder regions, while the effect of lower average minimum temperatures is larger in warmer regions.

Intensity (annual average temperatures):

• Average annual maximum temperature (larger effect in colder areas): At average minimum temperature = 3 °C (colder), a 1 °C increase in average max temperature was associated with: +0.71 percentage points (pp) in risk of spending >10% of income on energy (95% CI: 0.35, 1.07); +3.13 pp in LIHC risk (95% CI: −0.26, 6.51); and +0.81 pp in utility bill arrears risk (95% CI: 0.27, 1.34). Effects attenuated and sometimes became insignificant in warmer areas (e.g., at min temp = 9 °C).
• Average annual minimum temperature (larger effect in warmer areas): At average maximum temperature = 30 °C (warmer), a 1 °C lower average min temperature (equivalently, a 1 °C increase in min temperature reduces risk by these amounts) corresponded to: +0.95 pp in >10% energy expense risk (magnitude from −0.95; 95% CI: −1.16, −0.73); +1.23 pp in LIHC risk (magnitude from −1.23; 95% CI: −1.80, −0.66); and +0.76 pp in arrears risk (magnitude from −0.76; 95% CI: −1.08, −0.45).

Frequency (number of events):

• Heatwaves (evaluated at coldwave events = 2, 3, 4): increases in the number of heatwave events were associated with higher hardship. For example, risk of >10% energy expense rose by about +0.09 pp (95% CI: 0.02, 0.16) at 2 coldwave events, +0.34 pp (95% CI: 0.24, 0.44) at 3, and +0.65 pp (95% CI: 0.48, 0.81) at 4. LIHC and arrears risks also increased with more heatwave events.

Duration (average length of events):

• Heatwave duration (evaluated at coldwave duration = 5, 6, 7 days): longer heatwaves increased hardship. For example, at coldwave duration = 7 days: +0.21 pp in >10% expense (95% CI: 0.01, 0.41); +0.33 pp in LIHC (95% CI: 0.16, 0.51); +0.40 pp in arrears (95% CI: 0.19, 0.60).
• Coldwave duration (evaluated at heatwave duration = 5, 6, 7 days): effects were mixed for expenditure share and >10% risk (small negative to positive), but LIHC and arrears risks generally rose with longer coldwaves (e.g., at heatwave duration = 7: +0.17 pp in >10% expense (95% CI: 0.01, 0.34); +0.25 pp in LIHC (95% CI: 0.10, 0.40); +0.57 pp in arrears (95% CI: 0.21, 0.93)).

Vulnerable groups and protective factors:

• Individuals: Older adults faced larger increases in the proportion of income spent on energy with extreme temperatures; middle-aged groups more often reported utility bill arrears.
• Households: Lone parents with children and lone-person households showed the largest increases in expenditure-based hardship; group or multi-family households and those in poorer housing quality had elevated consensual-based hardship. Renters were more susceptible than owner-occupiers.
• Housing/Neighbourhood: Good-quality and owner-occupied housing mitigated impacts. Areas with low greenspace coverage (≤10%) had higher hardship risks as heat intensified and became more frequent. Areas with low density of small-scale renewable energy installations (≤1 unit/km²) faced greater risks; higher prevalence of renewables was associated with lower impacts, consistent with improved energy security.

Geographic/climate interaction: The heat–hardship link was stronger in relatively colder areas, while the cold–hardship link was stronger in relatively warmer areas, suggesting heightened vulnerability where communities are less accustomed or adapted to specific temperature hazards.

Projections (temperature-driven changes):

• RCP 4.5: Short run (2020–2039) increases of about +0.45% to +0.56% in risk of >10% energy expenditure and +2.24% to +2.47% in arrears; long run (2080–2099) increases of +0.08% to +0.52% (>10% expense) and +2.20% to +2.56% (arrears).
• RCP 8.5: Short run (2020–2039) increases of +0.32% to +0.52% (>10% expense) and +2.22% to +2.38% (arrears); long run (2080–2099) increases of +0.62% to +3.26% (>10% expense) and +2.32% to +2.76% (arrears).

Overall, risks are projected to rise more under high-emissions scenarios, and despite warming, energy hardship remains a persistent concern, especially in colder regions less prepared for heat and among vulnerable populations without adequate housing and neighbourhood adaptations.

The study shows that climate change, through more intense, frequent, and prolonged extremes of heat and cold, increases energy hardship and that these effects are unevenly distributed. Contrary to assumptions that warming might reduce energy hardship by lowering heating demand, excess heat is an increasingly important driver. The stronger heat effect in colder regions and stronger cold effect in warmer regions indicate maladaptation where local infrastructure, housing stock, and behaviours are not aligned with newly emerging climate hazards. Vulnerable groups—older adults, middle-aged households (for arrears), lone parents, lone persons, renters, and those in poor-quality housing—face greater hardship due to higher energy needs, constrained resources, and less efficient or lower-quality homes. Neighbourhood features such as greenspace and distributed renewables provide protective effects by moderating microclimates and enhancing energy security. The findings suggest that improvements in housing quality and energy efficiency, expansion of renewables, and urban greening can build adaptive capacity and reduce susceptibility. Projections under RCP 4.5 and 8.5 indicate that without substantial mitigation and adaptation, energy hardship risks will rise, reinforcing the need for targeted policies and investments tailored to local climate exposure and social vulnerability.

Energy hardship will remain a concern as average temperatures rise, with future hardship increasingly driven by extreme heat. Risks grow with greater intensity, frequency, and duration of heat and cold, and are concentrated among older adults, lone-person and single-parent households, renters, and those in poor-quality housing, especially in colder regions less prepared for heat. Improving adaptive capacity through stronger energy-efficiency standards, housing quality upgrades, renter protections and rights to repairs, strategic landscaping and urban greening, and wider uptake of renewable energy can mitigate risks. Policy and planning should target vulnerable groups and areas lacking protective infrastructure while anticipating shifts in local climate hazards under different emissions pathways. Future research could expand to additional adaptation measures, dynamic housing and energy market changes, and fine-grained spatial analyses to guide equitable climate resilience strategies.