The impact of weather forecasts on small business performance: methods and evidence from atmospheric river (AR) forecasts in the american west lead principal investigator: abhishek nagaraj, national…

The impact of weather forecasts on small business performance: methods and evidence from atmospheric river (AR) forecasts in the american west lead principal investigator: abhishek nagaraj, national bureau of economic research (NBER) and UC berkeley unfunded collaborator: weilong wang, postdoctoral fellow, columbia university in january 2023, atmospheric river (AR) events caused widespread disruptions in california, with small businesses severely affected. Weather forecasts played a crucial role in helping businesses and communities prepare for and mitigate the impacts of these events by providing advance warning of the timing, intensity, and location of expected flooding and high winds. ARS are becoming increasingly important to understand and predict, particularly in the american west, as climate change drives more frequent and intense events. Forecasts are poised to be a key tool in mitigating the negative impacts of ARS on vulnerable communities and businesses by providing actionable information to inform emergency preparedness and response efforts. While the value of forecasts is widely acknowledged, the lack of standard methods for rigorously quantifying their benefits means that accounts of their impacts remain largely anecdotal. A systematic methodology is needed to quantify the economic value of weather forecasts, starting with short to medium-term precipitation forecasts such as those accompanying AR weather events in the west. The findings would help value improvements in forecasting accuracy and guide future investment in forecasting products targeted toward improving AR prediction in the american west. To address this need, we employ a novel methodology that uses the applied microeconomics toolkit of natural experiments to quantitatively estimate the impact of AR forecasts on small business performance. These methods approximate the thought experiment we would like to run: how would economic outcomes differ if we randomly improve forecasts on certain days but not others, keeping everything else, including the weather itself, constant? Specifically, our approach compares scenarios where a location experiences very similar weather conditions but was well warned by accurate forecasts in one case and not another. By looking at variations in business outcomes (such as differences in footfall or revenue) across the two scenarios, we can quantify the value of improved prediction over and above the effect of the weather itself. Our proof-of-concept work quantifying the effects of 1-day precipitation forecasts across a broad sample of places suggests that this approach could be promising. Our initial results suggest that relative to well-calibrated forecasts, errors in forecasts have real effects on small business performance. The effects depend on the amount and direction of error, the severity of the weather event, and the establishment's industry. We hope to refine and apply these methods to analyzing AR forecasts on different regions in the american west between the years 2018-2023. Doing so will allow US to understand (a) the average value (in dollars) of good precipitation forecasts and (B) the variation that drives this average by considering the conditions under which the value of such forecasts is higher versus lower, including separating the value by industry, location and weather conditions. Our work will contribute to the small but burgeoning literature by applying an economic lens to estimate the economic value of weather forecasts. We will be the first to look specifically at the role of weather forecasts around AR events to better understand how we might mitigate their adverse societal impact. More generally, we hope to shape investment at the national level in terms of improving forecasting performance.