Answer Happy

(2) (This example is loosely based on actual studies) Researchers have gathered daily data on new COVID infections from two states, which we'll call State A and State B. During the time period studied, State A imposed no restrictions related to COVID, while State B imposed a "shutdown" (closing bars and restaurants and limiting the size of gatherings) halfway through this time period, which remained in effect for the rest of the time included in the data. The researchers want to use a differences in differences research design to estimate the effect of shutdowns on new daily COVID infections. The graph below plots daily infections in the two states over the time period (40 days). The vertical line indicates that State B's shutdown went into effect starting on the 21st day. (Note that in reality we might expect the impact of such as shutdown to take a week or two to show up in the counts 1 of new infections, which are really new diagnoses of previous infections, but here let's assume we actually have the number of people who were newly infected on each date, not the number of people who were previously infected and diagnosed on a certain date).

who were previously infected and diagnosed on a certain date). State A State B 600 500 400 COVID cases 300 200 100 o 10 20 30 40 day

eans for (a) Explain what the parallel trends assumption for t Screen Shot 2021-07-08 at 9.09.38 PM this example and whether, based on the graph, it seems to be reasonable here. (b) Explain what you would estimate (roughly) was the effect of the restrictions on infections? (Hint: look at the y-axis labels to get an idea of what a given vertical distance on the graph means and compare the distance between the two states before treatment is switched on as well as after treatment is switched on for state B). Be sure to explain why you can make that conclusion given the data, under the assumptions of the differences in differences model.