In order to assess economic consequences of events past or future you need to understand the events. This goes for evaluating policy options as well. One factor confounding people is the volatility of daily positivity rates. One day the rate may seem high (by anyone’s subjective standard) and on another day it may seem low. This is one reason entities like the World Health Organization recommend a 14-day moving average comparison to thresholds. It is a way to smooth out the daily fluctuations and see the longer trend and patterns in the data. I looked at three different positivity rates, though I really only like two of them. These data are publicly available from the state. I simply aggregated over all counties for a given day.
As listed, the above graph is positive tests divided by susceptible test encounters. What is that denominator? It essentially is all tests for people that have not already contracted COVID-19, meaning it is excluding those that are testing because they tested positive before and are seeing if they are still contagious. I think the benefits of the MA calculation are pretty clear when you see how scattered the daily rates can be. I like this measure the best. I know some do not like to include serial testers but I see no reason to exclude the information they present in our calculation. The 14-day centered moving average is pretty clearly moving in the wrong direction, and has been for some time. It is also drawing closer to the critical value of 10%.
This measure is very similar to the one already listed. The difference here is the denominator includes all tests, even to those checking to see if they are no longer contagious. You could argue those taking the test to confirm that they are over the disease are more likely to test when they know they will come up with a negative. As a result this would present a downward bias to the number, that is, no increase to the numerator and an increase of one in the denominator. Even with this potential downward bias we see the moving average is still trending in the wrong direction and is again pulling close to the 10% level. Now we turn to my least favorite measure.
The above graph shows all positives divided by those receiving their first test. It is the most restrictive measure in the denominator but I am not convinced of its value when the positives in the numerator could come from people taking their first, second, third, or whatever number test. I do not think it accurately reflects the level of virus in the population at large and wold be of little value in terms of assessing larger economic consequences.
I will be tracking these data closely moving forward and looking to connect them with economic data to see if there are predictions to make or insights for policy alternatives.