天气和星期几对消费的影响：来自高频数据的证据

The Impact of Weather and Day of the Week onConsumption: Evidence from High-Frequency Data Chang Soo JungJunior Economist, Macro Forecasting & Research Team,Research Department, Bank of KoreaTel. 02-759-4254, E-mail.crystalj@bok.or.krCho Byoung SooSenior Economist, Macro Forecasting & ResearchTeam, Research Department, Bank of KoreaTel. 02-759-4202, E-mail.bs.cho@bok.or.kr ①Daily card spending, captured through high-frequency datasets designed to monitor short-term consumption trends, varies substantially depending on both weather conditions andday of the week.Offline activities are limited by adverse weather conditions such as heat wavesor heavy rainfall, resulting in a measurable reduction in consumption expenditure. Furthermore,since consumption patterns differ by day of the week, the impact of adverse weather also variesdepending on when during the week these conditions occur. This study examines the impact of weather on consumption by leveraging daily cardspending data alongside temperature and precipitation variables while also explicitlyconsidering interactions with the day of the week. Drawing on these analyses, we estimatethe effects of major adverse weather events that have occurred this year on overallconsumption. ②First,analysis of the relationship between weather and consumption revealed that (1) cardspending declined by 7% during heat waves, 3% during cold waves, and 6% during periodsof heavy rainfall—each a statistically significant reduction.These effects were most evidentin offline activities, particularly in sectors such as offline shopping and dining out. In contrast,online shopping showed stable consumption patterns regardless of weather conditions. (2) Subsequent analysis of consumption patterns by day of the week revealed that total cardspending was highest on Fridays. Offline consumption displayed a noticeable concentrationon Saturdays compared to other weekdays.On weekdays (Monday through Friday), the shareof online shopping was relatively high. Expenditures on automobiles, medical services, andeducation—which typically occur on business days—were also prominent on weekdays. (3) Furthermore, the interaction effect of weather deterioration on Fridays and Saturdays—when offline consumption is highest—was estimated. The results indicate that rainfall onthese days led to a larger decline in card spending associated with offline consumptioncompared to other days of the week. When weather cleared following periods of restrainedconsumptiondue to rainfall,consumption rebounded to higher-than-usual levels—aphenomenon known as the pent-up effect.Meanwhile, the impact of heat and cold wavesshowed little variation across days of the week. This pattern appears to reflect heightened seasonaldemand during summer and winter—such as vacations and purchases of cooling or heatingequipment—which partially offsets the negative effects of extreme weather conditions. ③Buildingon these analyses,the impact of major weather events on consumer spending thisyear was estimated. Results indicate that cold waves during the first quarter and summerheat waves (June through August) contributed to a 0.18-percentage-point decline in annualconsumptiongrowth.However,below-average rainfall during the same period boostedconsumption by approximately 0.09 percentage points. Consequently, the net impact of weatherconditions on the private consumption growth rate for the year is estimated at approximately -0.09percentage points. ④This study provides valuable empirical evidence for the timely monitoring of consumerspending behavior using daily card spending and weather data. By quantifying temporarysources of variation in consumption, it is expected to enhance the accuracy of short-termconsumption forecasts. ■Disclaimer: The views expressed herein are those of the authors, and do not necessarily reflectthe official views of the Bank of Korea. When reporting or citing this paper, the authors’ namesshould be always explicitly stated.■We would like to express our gratitude to Professor Cho Seong-hoon of Yonsei University, LeeJiho(Director General,Research Department),Kim Minsik(Director,MacroeconomicForecasting Division), and Park Changhyeon (Head, Macro Forecasting & Research Team). Bank of Korea 1.Daily card spending, analyzed using high-frequency datasets designed for short-term consumption tracking,reveals substantial variation depending on weather factors and day of the week.The frequent occurrence of heatwaves and heavy rainfall, as during the last summer, constrains offline activities—particularly in-person and physicalstore transactions—leading to pronounced shifts in spending patterns. Moreover, daily consumption exhibitsconsiderable variation in both the value of card transactions and composition of purchase categories, depending on theday of the week. There are marked differences in consumer purchasing behavior between weekdays and weekends.For example, on rainy days, demand for delivery services tends to inc