Abstract:
Public warning systems are policy tools whose consequences depend not only on the information they provide but also on how that information is delivered. This paper studies South Korea's Emergency Alert (EA) system, which made existing particulate matter (PM) warnings more immediate and salient through direct mobile notifications. Using daily data for eight city-level metropolitan jurisdictions from 2015-2019, I estimate whether the warning-specific influenza response is amplified when warning delivery shifts to the EA regime. The primary estimand is the post-EA change in the warning-specific component relative to lower-tier advisories. The baseline amplification is 0.100 additional influenza cases per 1,000 residents per day during the 60-day post-warning window, corresponding to about 8.6% of annualized post-EA incidence when scaled by observed post-warning exposure frequency. The estimate remains positive under exact calendar-date fixed effects and across alternative weighting, pollution, timing, and sample specifications. Dynamic estimates show no sustained positive pre-warning pattern and peak approximately 25-40 days after warning onset. Search behavior and heterogeneity by holiday, school status, and age are consistent with a short-lived change in activity environments preceding a longer transmission process. The findings identify an unintended health spillover associated with the amplification of warning responses under the EA-era delivery regime and show that warning-delivery design can shape outcomes outside the targeted policy domain.

Keywords: Air pollution; Behavioral response; Information policy; Influenza; Alert

JEL Codes: D83, Q53, I18, I12

Figure 1. Event Study: Dynamic Warning Effects Before and After the Emergency Alert Delivery Regime