Weather forecasts are perhaps the most pervasive information intervention undertaken by governments. Do people use this information? Are forecasts accurate enough now, or should society devote additional resources to making them better? To investigate these questions, we analyzed daily data on deaths, weather conditions, and National Weather Service forecasts for every U.S. county from 2005 to 2017. By comparing mortality rates following accurate and inaccurate day-ahead forecasts, we determine the impact of people's responses to forecasts on mortality.
The effectiveness of more accurate forecasts in reducing mortality hinges on the interplay between excessively hot and excessively cold forecast errors. If a too-hot forecast for a hot day would save the same number of lives as a too-cold forecast on the same hot day, then mitigating both types of errors would not result in overall lives being saved. However, if mortality is asymmetrically influenced by forecast errors, where milder forecasts cause more deaths than the number of lives saved by extreme forecasts, improving forecast accuracy would save lives overall.
Our findings provide robust evidence of asymmetric responses to forecast errors. Specifically, a too-warm forecast on a cold day leads to additional deaths, while a too-cold forecast on a cold day does not have life-saving effects. Similarly, a too-mild forecast on a hot day results in increased deaths, whereas a too-hot forecast on a hot day does not save lives.
Across the temperature spectrum, excessively mild forecasts pose a greater threat during hot weather. This is partly due to the rapid fatality rates associated with hot days, such as hyperthermia, compared to the longer time span required for individuals to succumb to the effects of cold days, including subsequent respiratory infections. Consequently, forecast accuracy proves more valuable on hot days.
In our examination of people's responses to forecasts, we analyzed time-use diaries. We observed individuals spending more time on leisure and less time on work when the weather was forecasted to be milder than it turned out to be. Surprisingly, we also found that electricity usage responded to forecasts even after accounting for observed weather, indicating that people not only made immediate adjustments like changes in air conditioning but also incorporated weather forecasts into their long-term planning.
Recognizing that responses to forecasts entail some costs, we theoretically demonstrate how these costs can be incorporated into estimates of the public's willingness to pay for improved forecasts. Our findings reveal that across the U.S., the aggregate willingness to pay (based solely on reductions in mortality) for a day-ahead forecast that is 50% more accurate over the remaining century amounts to $112 billion. Approximately 20% of this value is driven by climate change, which leads to an increased frequency of hot days that particularly benefit from enhanced forecast accuracy. Importantly, the overall willingness to pay far exceeds the entire budget of the National Weather Service, approximately $1 billion per year. Thus, our results suggest that society stands to gain from the allocation of additional resources towards improving forecast accuracy.
© Jeffrey G. Shrader, Laura Bakkensen, and Derek Lemoine
Jeffrey G. Shrader is Assistant Professor at Columbia University’s School of International and Publica Affairs and IZA Research Affiliate
Laura Bakkensen is Associate Professor and Director of MPA and MPP Programs at University of Arizona
Derek Lemoine is Associate Professor of Economics at Eller College of Management, Research Associate for the NBER and Associate Fellow for CEPR
Please note:
We recognize that IZA World of Labor articles may prompt discussion and possibly controversy. Opinion pieces, such as the one above, capture ideas and debates concisely, and anchor them with real-world examples. Opinions stated here do not necessarily reflect those of the IZA.
Related IZA World of Labor content:
Does hot weather affect human fertility? by Alan Barreca
Climate change and the allocation of time by Marie Connolly
Temperature, productivity, and income by Olivier Deschenes