Corona 2020: Store closures had a small effect, masks a large one

The study has been published on a preprint server, then went through a peer review process and has already been accepted for publication. According to the study, contact restrictions reduced the increase in Corona virus infections by almost 14 percentage points. In other words, if there had been an increase in cumulative infections of, say, 30 percent in a region in a given time period without this measure, infections actually increased by only about 16 percent thanks to distance requirements and other contact restrictions. Mandatory masks in buses, trains, supermarkets and other public places also proved effective in flattening the curve and preventing exponential spread of the disease: The duty reduced the increase by another 13.5 percentage points.

School and daycare closures had a smaller, but still notable, impact of about 5.5 percentage points. Restaurant closures had an effect of about 2 percentage points, and the closure of parks, zoos, museums or wellness facilities - as well as stores - had a barely detectable effect. "The effect of store closures was hardly detectable," notes the head of the study, Kassel-based statistician Prof. Dr. Reinhold Kosfeld. One reason could be that there is usually much more space available in the closed stores than in the system-relevant stores that remained open, such as supermarkets.

Kosfeld continued, "Contact restrictions and mandatory masks were the pillars of success in containing the pandemic. The effect of closing schools and daycare centers was significant, but much smaller. The calculated effect accounts for 14 percent of the total effect, or about one-seventh of the total success. The share drops to 6 percent for closures of restaurants, cafes, and bars. The shutdown of non-essential retail is even smaller in proportion, at 4.5 percent, and is also statistically uncertain. In general, one should keep in mind that the effects are influenced by the chronology of closures during the crisis. Additional measures taken relatively late often have a smaller effect than the first measures."

Kosfeld also emphasizes that he calculated the statistical effect in the study, but does not make any statements about policy reasons for specific measures, such as closing and opening educational institutions, nor does he describe any medical or epidemiological causalities. What's more, the measures were examined in spring 2020, when some underlying conditions may have been different, such as the rate of virus infection or the availability of rapid tests and vaccinations. "Nevertheless, it is worthwhile not only to speculate about the effect of school, business and restaurant closures or their opening, but to draw on robust experience from the first wave," the statistician appeals.

Kosfeld's research group used a difference-in-differences approach for the analysis: they evaluated the data for 401 German counties and cities. Since the measures were introduced in the federal states at different times, they were able to form groups of cities and counties for certain periods of time in which the measures were already in effect and control groups in which they were not (yet) in effect. They also projected disease progression curves for each territorial unit using data from the RKI. They calculated that the counties and cities were in different phases of the pandemic, i.e., the rate of spread in a Bavarian city may have been higher than in a Mecklenburg county. The study period was mid-March to late April, when initial relaxations took effect.

Kosfeld is an adjunct professor at the Institute of Economics at the Kassel University. Until last fall, he headed the Department of Statistics. In addition to the Kassel statistician, researchers from the universities of Mainz, Darmstadt and Southern Denmark were involved. The study was published on the "Preprint Server for Health Sciences" medrxiv.org. It has been accepted for publication for the Special Issue "Covid-19 and the Regional Economy" in the Journal of Regional Science, one of the leading international regional science journals.

Link to the study on the preprint server (peer review process has now taken place): https://www.medrxiv.org/content/10.1101/2020.12.15.20248173v2

https://doi.org/10.1101/2020.12.15.20248173

Contact:

Prof. Dr. Reinhold Kosfeld
Kassel University
Institute of Economics
E-Mail: rkosfeld[at]uni-kassel[dot]de
Phone: 0561 804-3084

Press contact:

Sebastian Mense
Kassel University
Communication, Press and Public Relations
Phone: +49 561 804-1961
E-mail: presse[at]uni-kassel[dot]de
www.uni-kassel.de