“…The key novel point of our work relates to the analysis of the surprise effect rather than the effect of the weather per se, which was largely analysed in related literature (Brandes and Dover, 2022;Lee et al, 2014;Radic and Lück, 2018;Štumpf et al, 2021). Interestingly, by running a specification where only the actual weather (not the mismatch with prior expectations) is considered (Table 5, Column 1), we notice that the coefficients of sunny and rainy are both negative and significant (respectively -0.052 and -0.054, Column 1, reference category: cloudy).…”