Does cumulative thermal discomfort reshape heating priorities? Evidence of constrained resignation among French households

Household energy decisions are typically modeled as linear responses to climatic conditions, overlooking the behavioral responses to prolonged climatic stress. A cumulative Thermal Discomfort Index (TDI) is introduced to capture cold stress through an asymmetric, deviation-based design. Unlike Heating Degree Days, which proxy heating requirements from mean temperature alone, the TDI integrates multidimensional deviations from local climatic norms over three winters and yields the lowest AIC among eleven benchmark indices tested on the same specification. Drawing on a survey of 3874 French households and spatial econometric methods, the analysis examines stated heating priorities, a self-reported trade-off between thermal comfort and energy savings, rather than metered consumption. The results document a non-monotonic behavioral pattern: heightened cumulative discomfort does not strengthen the stated preference for warmth but is associated with a shift toward energy savings prioritization. Cumulative TDI enters with the expected signs, positive on savings prioritization and negative on comfort, but neither pooled coefficient is statistically distinguishable from zero ( p = 0.42 and p = 0.75 ), and a likelihood ratio test against a controls-only specification does not reject ( p = 0.133 ). No benchmark climatic index reaches conventional significance on this sample either. Individual effect sizes are small per unit and accumulate only across the full observed range of climatic exposure: from minimum to maximum TDI (77.9 units) the implied shift reaches approximately 7 percentage points toward savings, while between the 5th and 95th percentiles (28.5 units) it is 2.6 points. Sub-sample stratification points to structural concentration in point estimates: the TDI coefficient is roughly two and a half times as large in high-stress regions, although the difference between sub-samples does not reach conventional significance ( z = − 0.43 ). This regional contrast separates the proposed mechanism from a pure budget-constraint account, which would predict the sharper response in colder regions where heating needs and bills are highest; the larger point estimates arise instead where cold is unexpected and dwellings are structurally unprepared, a response to violated climatic expectations rather than to chronic heating costs or income constraints alone. Moran’s I diagnostics on the model residuals ( I = 0.015 , p ≈ 0.75 ) indicate that the behavioral shift is associated with household-level constraints, building inefficiency and income precariousness, rather than geographic contagion. These findings support reorienting energy policy from bill-based subsidies toward removing the structural barriers that lock vulnerable households into involuntary thermal deprivation. The TDI offers a diagnostic tool for identifying populations where climate adaptation may have become a source of behavioral constraint rather than resilience.