The weather affects air conditioner purchases to fill the energy efficiency gap

Energy efficiency improvements are crucial for achieving Net Zero emissions, yet the energy efficiency gap—underinvestment in energy-saving technologies—hinders progress. This gap is partly due to behavioral anomalies; consumers struggle to accurately predict future utility when making investment decisions. Existing research suggests that consumers may heavily discount future fuel economy and pay limited attention to energy efficiency. This study explores whether short-term environmental changes, specifically temperature fluctuations, can influence consumer choices and potentially bridge this gap. Previous research indicates that weather affects various purchasing decisions (automobiles, solar systems, clothing, college enrollment), suggesting that current environmental conditions might shape predictions of future conditions. This study focuses on the unexplored question of whether short-term temperature changes can impact the purchase of energy-efficient air conditioners, specifically Energy Star models, thereby mitigating the energy efficiency gap.

The literature review highlights the energy efficiency gap and its behavioral underpinnings. Studies show consumers often underestimate the long-term benefits of energy-saving products, leading to underconsumption. Existing research explores the influence of weather on various purchasing decisions, including automobiles, solar systems, clothing, and college enrollment. Two primary mechanisms are proposed: projection bias (consumers expect future conditions to resemble the present) and salience (consumers pay more attention to immediately relevant environmental factors). However, the impact of short-term temperature fluctuations on energy-efficient appliance purchases remains underexplored.

The study uses transaction-level data on air conditioner purchases from 2006 to 2019. A generalized linear regression model analyzes the relationship between weekly average temperature deviations from the mean and the probability of purchasing an Energy Star air conditioner. The model includes control variables such as other meteorological indicators (precipitation, wind speed, relative humidity), price, and fixed effects (month, year) to account for confounding factors. Several robustness checks are performed to address potential issues like autocorrelation in weather data, geographical variations in Energy Star product availability, and potential harvesting behavior (consumers bringing forward purchases). The analysis also investigates heterogeneous effects, examining how the relationship between temperature and Energy Star purchases varies based on factors such as electricity prices, previous cooling degree days (CDDs) and heating degree days (HDDs), income, education, age, homeownership, electricity use for heating, and attitudes toward climate change. The study employs a GLS method due to challenges in converging with Probit or Logit models after including numerous fixed effects. A falsification test using telephone purchases is conducted to rule out spurious correlations. The specification is: l(ESR) = Σj Pu(tempj) + δWr + β2 pricer + μr + γt + εir, where ESR is a dummy variable for Energy Star purchases, Pu(tempj) represents temperature intervals, Wr includes other meteorological variables, pricer is the transaction price, μr represents county fixed effects, γt represents year fixed effects, and εir is the error term. CDDs and HDDs are also used as key variables.

The analysis reveals that the probability of purchasing an Energy Star air conditioner significantly increases as the weekly average temperature deviates from 20–22°C. This effect is more pronounced with fewer cooling degree days in previous years, higher electricity prices, higher income, higher education levels, older age, higher homeownership rates, more common use of electricity, and stronger concern about climate change. Robustness checks, including controlling for current week's weather and lagged weather data, confirm the findings. The results suggest that both projection bias and salience may contribute to this effect. Heterogeneous effects show that the response to temperature changes is larger in states with higher average electricity prices and areas with fewer background CDDs. The response also tends to increase with income and age, and is stronger in areas with a higher proportion of homeowners and less common use of electricity for heating. Stronger attitudes toward climate change are associated with a larger response. A 1°C increase from 21°C could increase Energy Star air conditioner purchases by 17,760 units, while a 1°C decrease could increase purchases by 8,580 units, based on an average of 4.44 million annual air conditioner shipments.

The findings address the research question by demonstrating that short-term temperature fluctuations can influence the purchase of Energy Star air conditioners, potentially mitigating the energy efficiency gap. The significant impact of temperature on purchasing decisions, coupled with the identified heterogeneous effects, suggests that targeted policy interventions could leverage these patterns to enhance energy efficiency. The results highlight the potential for using weather-based marketing strategies and nudges to promote energy-efficient appliances, particularly among populations identified as more responsive to temperature changes. The study's implications are relevant for policymakers, who can use this information to improve the design and effectiveness of energy efficiency programs, including non-price policies. Integrating consumer behavior into climate change impact assessments is crucial for developing effective mitigation strategies. Demand-side strategies, such as leveraging temperature fluctuations, could play a significant role in reducing energy consumption and promoting energy-efficient technologies.

This study demonstrates that temperature deviations significantly increase the purchase probability of Energy Star air conditioners relative to non-efficient models. This effect can be leveraged through targeted interventions like weather-based marketing and nudges. Future research could explore daily weather fluctuations, incorporate online sales data, analyze private vs. household purchasing decisions, and delve into more detailed consumer behavior patterns at a finer geographical scale.

The study uses weekly aggregated data, potentially missing daily weather fluctuations. The data may not encompass all air conditioner transactions (excluding online purchases). The retail scanner data provides store-level sales information but lacks individual consumer behavior details. These limitations could affect the interpretation and generalizability of results, specifically the ability to precisely quantify total electricity savings at more aggregated levels. The study also relies on county-level proxies for consumer characteristics which may not perfectly reflect individual consumer behavior.