Understanding how households adapt to rising temperatures is central to evaluating the economic costs of climate change, yet evidence on adaptation in food consumption and expenditure remains limited. This paper studies behavioral adaptation in food-at-home spending to climate change by distinguishing short-run responses to weather fluctuations from long-run responses to persistent warming in the United States. Using household-level grocery transaction data matched with high-resolution temperature records, we decompose realized temperature into a slow-moving climate norm and a short-run weather deviation and estimate behavioral responses on both the extensive margin, the decision to shop, and the intensive margin, expenditure conditional on shopping. We find that food-at-home spending declines with higher temperatures in both short and long-run, but the long-run climate response is much larger. A 1℃ increase in weekly temperature above the norm lowers expected weekly spending by about 13 cents per capita, while a 1℃ increase in the climate norm lowers weekly spending by about 50 cents. This implies that behavioral adaptation intensifies, rather than attenuate the short-run effect of heat. Responses vary across income groups, regions, and food categories, and differ between cold and heat exposure. These results highlight the importance of household consumption behavior for accurately measuring economic impact of climate change and the role of behavioral adaptation in adjusting long-run welfare cost.
