["Oxford Bulletin of Economics and Statistics, Volume 88, Issue 2, Page 173-197, April 2026. ", "\nABSTRACT\nThis study implements a geographical regression discontinuity design to estimate the impact of Japan's anti‐COVID‐19 policy, which imposed what is called “a state of emergency” (SOE). This policy can be considered a ‘mild lockdown’ without any geographical restriction. In January 2021, an SOE was declared in 11 out of Japan's 47 prefectures. We utilised the fact that the risk of infection and other factors were geographically continuous across the borders between targeted and nontargeted areas, while the SOE situation was completely discontinuous at each border. Consequently, those living barely outside the targeted areas were more alarmed by COVID‐19 than those living barely inside the targeted areas. We also found that this fear among those living barely outside the targeted areas contributed to behaviours such as refraining from going to bars or restaurants and to becoming more careful about practising COVID‐19 countermeasures. Our study is the first to simultaneously reveal the negative effects of the SOE in targeted areas and the positive effects in nontargeted areas. Thus, the results of this study highlight the importance of reconsidering the necessity of disease control measures that utilise the borders between administrative divisions, such as states or prefectures, to guide the implementation of a ‘mild lockdown.’\n"]