Mass gathering events and the spread of infectious diseases: Evidence from the early growth phase of COVID-19

Abstract

This paper studies the impact on reported coronavirus 2019 (COVID-19) cases and deaths in Spain resulting from large mass gatherings that occurred from March 6 to March 8, 2020. To study these outcomes, the geographic differences in the planned pre-pandemic major events that took place on these dates were exploited, which is a quasi-random source of variation for identification purposes. We collected daily and detailed information about the number of attendees at football (soccer) and basketball matches in addition to individuals participating in the Women's Day marches across Spain, which we merged with daily data on reported COVID-19 cases and deaths at the provincial level. Our results reveal evidence of non-negligible COVID-19 cases related to the differences in the percentage of attendees at these major events from March 6 to March 8. In a typical province, approximately 31% of the average daily reported COVID-19 cases per 100,000 inhabitants between mid-March and early April 2020 can be explained by the participation rate in those major events. A back-of-the-envelope calculation suggests that this implies almost five million euros (169,000 euros/day) of additional economic cost in the health system of a typical province with one million inhabitants in the period under consideration. Several mechanisms behind the spread of COVID-19 are also examined.

Keywords: COVID-19; Cases; Deaths; Mass gathering events; Spain.

Fig. 1

% Attendees to mass gathering events (March 6–8) over total population at the provincial level. Notes: Lighter colors correspond to lower % attendees to mass gathering events (March 6–8) in each province.

Fig. 2

Number of COVID-19 cases per 100,000 inhabitants at the provincial level. Notes: Fig. 2 shows the evolution of COVID-19 cases per 100,000 inhabitants at the province level. Lighter colors correspond to lower COVID-19 cases in each province and on each day.

Fig. 3

Identification, event study of COVID-19 cases. Notes: Robust standard errors are clustered at the province level. Significant at the 5% level; see Table B1.

Fig. 4

Identification, event study on COVID-19 deaths. Notes: Robust standard errors are clustered at the province level. Significant at the 5% level; see Table B1.

Fig. B2

Identification #2: Event study on excess mortality, Notes: Robust standard errors are clustered at the NUTS 2 region level. Significant at the 5% level.

Fig. B3

Identification #3: DID, from last period before treatment changes (t = −1) to t.

Fig. C1

Distribution of placebo estimates. This appendix shows the distribution of the estimated coefficients from 2000 placebo regressions using random attendance rates. Vertical black lines represent estimated significant coefficients using the actual attendance rates, obtained from Table 1, columns (1)–(4). The mass of density of the placebo estimated coefficients is far to the left of the estimated values using the actual attendance rates.