This study examines the heterogeneous causal effects of climate-driven cooling demand on U.S. household electricity consumption using 2020 Residential Energy Consumption Survey data and a causal machine learning framework. Estimates of conditional average treatment effect (CATE), conditional quantile treatment effect (CQTE), and conditional super quantile treatment effect (CSQTE) show substantial increases in electricity consumption under elevated cooling degree days (CDD), with average effects of 32–34% and disproportionate burdens concentrated among lower-consuming households. OLS projections of CATE, CQTE, and CSQTE estimates highlight significant subgroup variation: households with central AC and evaporative coolers, middle-income earners ($60k–$79k), and minority racial groups face the highest burdens, while single-family homes exhibit consistently stronger responses than apartments or mobile homes. Robustness checks using alternative treatment definitions (CDD above the 75th percentile and 30-year average CDD) confirm the stability of results. These findings underscore that climate-driven cooling demand exacerbates distributional inequalities in energy use, reinforcing the need for targeted efficiency and adaptation policies that account for subgroup vulnerabilities.
View Full Paper →This study examines the effects of climate-induced cooling demand on household electricity consumption and expenditure in the United States using an unconditional quantile regression framework. Drawing on data from the 2015 and 2020 Residential Energy Consumption Survey, it estimates heterogeneous and nonlinear responses to cooling degree days across low-, median-, and high-consuming households. Results show that a 1% increase in cooling degree days raises electricity consumption by 1.04%, 0.88%, and 0.68% at the 25th, 50th, and 75th quantiles, respectively, while expenditure increases by 0.98%, 0.73%, and 0.54%. Higher impacts are observed among households in hot and mixed-dry zones, those with central air conditioning, and lower-income or minority households at lower quantiles. Using long-run climatic exposure confirms the robustness of these results. These findings underscore the need for adaptive efficiency and affordability policies to address climate-driven electricity vulnerability.
This study investigates the spatial and socioeconomic determinants of renter cost burdens across U.S. census tracts. Focusing on the percentage of renters spending more than 30% of their income on housing, the study estimated a baseline OLS model followed by spatial econometric models – including Spatial Lag, Spatial Error, and Spatial Autocorrelation models to account for spatial dependencies. The results demonstrate that higher poverty and unemployment rates are robust predictors of increased rent burdens. Additionally, single-parent household prevalence significantly raises cost burdens, while educational attainment and mobility strongly influence affordability. Notably, limited mobility substantially reduces rent burdens, emphasizing the role of access in housing opportunities. The significance of spatial lag and error terms confirms that rent burden is spatially structured, with spillover effects across neighboring communities. These findings highlight the need for coordinated regional policies that integrate economic, educational, and transportation strategies to effectively address renter housing challenges.