Driving Forces of Changes in Behavior and the Economic Impact of COVID-19

As NYC starts (a slow and drawn out) reopening, and the United States begins dealing with the fallout of the decisions we have made, I think it is important to push back against the idea that the virus itself instead of the lockdowns caused the economic changes that are part of that aftermath.

Historically, pandemics have not been the cause of economic fallout. The brief Post-World War I Recession was due to hyperinflation and over production. The Depression of 1920–21 (which only occurred somewhat concurrently with Spanish Flu) was due to troops returning home from WWI. The Recession of 1958, which occurred concurrently with the Asian Flu pandemic, was due to monetary policy changes. The Recession of 1969–70, which occurred somewhat concurrently with the Hong Kong Flu, was due to fiscal tightening to close budgetary deficits from the Vietnam War along with simultaneous raising of interest rates.

None of those recessions have attributed their causes to the pandemics that have in many cases only tangentially even overlapped with them. People were likely more cautious during those pandemics in some instances for some people, but there has never been, in the history of the United States, such a wide government reaction to a pandemic to close businesses en masse as has occurred during the COVID-19 pandemic.

Some have previously suggested that people’s reactions and the economic aftermath resulting from that have not been driven by the lockdowns, but due to the virus itself. They’ve looked at mobility data relative to the date each state had shelter in place (SIP) orders go into effect, and when looking at that data, it does seem like the decline in mobility had already happened before state governments had enacted SIP orders (last order is always SIP for states that did SIP).

There are a few problems with this analysis however.

• One is that it makes it seem like people solely changed their behavior because they were worried about the virus without actually looking at those changes in light of the progression of the virus in different states.
• Another is that it makes it seem like SIP orders were the first government orders to have a major impact on people’s behavior.
• The third is that it assumes government orders went into effect the day they were announced, when in many cases (as in NY) they were announced 48 hours beforehand.

People changing their mobility around intuition of when the virus became a problem in their state seems reasonable at first. When looking at the trends in mobility relative to the date states first had more than one total deaths per million population, it’s a bit more complicated.

Mobility changed nearly completely before the date one in a million people for each state had died from the virus. It isn’t clear that people’s intuitions would act that much earlier on.

You can also note that there are overlapping factors in timing with when the virus had killed more than one person per million population in each state and when shelter in place orders went into effect. If you look at when states did enact SIP orders, it’s a roughly normal distribution roughly centered around just a few days ahead from the date where more than one total deaths per million population occurred.

Separately from showing that this means that they are at least somewhat confounded variables (which both took place after mobility had already mostly reached their max difference off from baseline mobility), depending on if you define sheltering in place (typically reserved for war) at around when just over one death per million population occurs to be reasonable timing to make those orders, it does seem most states were reasonable with their timing.

SIP orders were not the first orders that would dramatically change people’s lives — many states shut down schools (NY, which has been touted as one of the better examples of handling Spanish Flu never shut down schools then) and many shut down restaurants (which hasn’t happened before) significantly before SIP orders went into effect (39 states had first orders that were either school closures or recreational restrictions more impactful on day to day life than those just on large gatherings — like restrictions against dining in at restaurants and bars).

If you look at mobility changes when state governments’ first orders went into effect, it starts to seem like those orders may have been a lot more likely to have driven mobility changes. At the day those first orders went into effect, mobility was down just 9.9% from the average of states off baseline mobility compared to ~40%, and those changes had not fully taken place beforehand.

If you look backwards to account for that there was a delay from the announcement of orders that would impact people’s behavior to their going into effect, this becomes even more apparent.

If you assume announcements are what caused an impact, you would want to look at Day 0 Ahead First Order minus the Number of Days from Announcement for those Orders to go into Effect minus one (to account for that mobility changes would happen afterwards) when comparing to baseline.

If you look at Day -1 (one day before) when the first order went into effect (which assumes states on average had NO announcement prior to orders going into effect when comparing to baseline — when we would see the effects afterwards), the changes from baseline mobility are just 6.3% (and roughly the same at 4.9% for Day -2).

If you look at Day -3 when those first orders went into effect (which assumes Day -2 or 48 hours beforehand to be when they were announced and we would first see an impact), the change off from baseline is just 0.4%.

Is it possible that there would have been some degree of cautionary measures taken by some people if the government did not enact unprecedentedly restrictive orders during the COVID-19 pandemic? Yes.

Would that have certainly driven the changes in behavior to the degree we see, and caused the economic fallout resulting from those changes? No, and they haven’t historically.