Today, we showed a guest post, the author is Jeffrey Frankel, Professor Happel of Harvard University’s Kennedy School of Government, and a former member of the White House Council of Economic Advisors.A shorter version appeared in Project Syndicate. Basic econometrics is in “Viruses, vaccination, and voting: an econometric analysis. “Thanks to Randy Kotti for his excellent research assistance.
Since the 1960s, we have been hearing the cliché: “If they can send people to the moon, why can’t they? X? ” Where X Usually some goals, such as eliminating hunger-technically simpler than the scientific miracle of space flight, but more difficult to achieve in practice because it involves human behavior. By 2021, the outstanding question is, “If we can create the scientific miracle of a vaccine that can end the Covid-19 pandemic that killed millions of people, why can’t we persuade enough people to get the vaccine?”
exist Low income nationThe jab is usually limited by the availability of vaccines. But for a lucky country like the United States, this is not the case. The main problem in the United States is vaccine hesitation and even complete vaccine hostility.
Awareness of the Two Americas
Many of us believe that the benefits of vaccination far outweigh the disadvantages-not only for society as a whole, but also for individuals. Therefore, selfishness cannot explain those who do not. What explains the mystery of widespread vaccine hesitation? [Personally, I have not had the opportunity to discuss this with anyone in that category.] In the words of Dr. Anthony Fauci, there is Two Americas. Their views on vaccination are separated by a wall.
We learned that people in other Americans are rarely persuaded by the expertise of remote authorities or the logic of scientific methods, such as Testing and approval by the Food and Drug Administration Three vaccines in response to the covid-19 emergency. In other words, Moderna, Pfizer and Johnson & Johnson. (AstraZeneca has been found in the British population and has proven Equally effective.)
Skeptic requirements evidence It is more tangible, closer to home.
Negative correlation between vaccination and virus victims
The latest data from counties in the United States shows that there is a strong negative correlation between vaccination rates and infection rates, hospitalization rates, or death rates.exist The week ending June 22Counties where 30% or less of the residents were vaccinated had 5.6 deaths from the new coronavirus per 100,000 people, while counties where more than 60% of the residents were vaccinated had less than half of the deaths, only 2.1 per 100,000 people. people. This seems to be more practical evidence than the FDA test.
In all these studies, the standard for the cause of death was whether the doctor or coroner entered covid-19 on the death certificate. As international research on excess mortality strongly suggests, this may underestimate the true number of deaths caused by covid-19. [See this post on US reported deaths, estimated excess deaths through July 10.]
exist update dataAs of June 9, the percentage of adults (and adolescents) fully vaccinated in a county has increased by 1 percentage pointday Related to covid-19 mortality in the following 30 days (until July 9)day) Lower 0.06 per 100,000 residents. This accounts for 2% of the total monthly deaths related to the new coronavirus. It is inferred that the obvious statistical effect from the current average vaccination rate to 100% vaccination will bring the number of deaths related to the new coronavirus close to zero.
figure 1: County-level deaths related to the new coronavirus are negatively correlated with the vaccination rate a month ago (showing counties that have reported at least one death from the new coronavirus during this period). source: Figure 2 Frankel (2021).
But, as they said, correlation does not need to prove causality.
Perhaps the obvious beneficial effect of vaccination is actually the illusory result of some third factor, such as the county’s poverty rate (an overlooked variable). In other words, perhaps low-income people are more likely to live in crowded environments and therefore become victims of the new coronavirus, and they are less likely to be vaccinated. The beauty of econometrics is that a third factor such as poverty rate or local temperature can be controlled to statistically isolate the impact of vaccination rates.
Or maybe it’s a simple observed correlation between vaccination and mortality Understatement Due to the endogenous nature of vaccination, the real impact of the former on the latter: In places where the risk of coronavirus is greater (for example, because it is close to a major airport or other transportation hub), people are more likely to see their neighbors become victims of the virus , And respond by deciding to vaccinate yourself. This “reverse causality” may contribute to the obvious positive correlation between vaccination and mortality.
This may help explain why the recent early research conducted in early June did not find a significant negative correlation.Not until recently Beneficial effect The intensity of vaccination is sufficient to dictate statistical correlation. The reason may be that the Delta variant poses an increasing challenge to the health of unvaccinated people.
Establish a causal relationship from voting to vaccination to the virus
The way to unravel causality is to examine the impact of changes in vaccination rates no It is related to the variation of disease transmission, but to some unrelated factors (exogenous tools). Partisanship or voting mode is the obvious choice. Even before the vaccine appeared, Red——state Governor It is found that measures such as mandatory wearing of masks are unlikely to be taken in 2020 to combat the coronavirus.
As widely reported, Republicans and those who voted for Donald Trump in the 2020 presidential election are less likely to be vaccinated.Republicans are less likely to receive vaccinations than Democrats 45% vs. 73%. One April 17 New York Times article It was found that in counties where Trump won by 50 percentage points or more, the vaccination rate was less than 25%. (And “even after accounting for income, race and age demographics, population density, and the infection rate and death rate in a county, the partisan gap still exists.”) Continue to expand in July.
Controlling the poverty rate and other related variables (especially age and temperature), we have Established As of June 9, the proportion of residents over 12 years of age who received the vaccine has increased by 1 percentage point.day Related to covid-19 mortality in the following 30 days (until July 9)day ) Reduced by 0.05 per 100,000 inhabitants. Therefore, controlling poverty and other variables will slightly reduce the estimated coefficients, but not significantly.
However, as mentioned above, even if there is a control, if the vaccination decision is affected by the prevalence of covid-19, the estimate will be biased.Using changes in vaccination decisions that are entirely attributable to Trump’s affinity, we have Established As of June 9, the vaccination rate in one county has increased by 1 percentage pointday Controlling the poverty rate and other variables, the covid-19 mortality rate was reduced by 0.04 per 100,000 residents in the following 30 days (to July 9).
As CDC Director said Recently, “this is becoming an unvaccinated epidemic.” Over 99% Of covid deaths are now occurring among those who have not been vaccinated.
The reason for considering the role of the voting model is to improve the estimate of the effectiveness of the vaccine for anyone (regardless of the political party). But maybe some skeptics will notice a higher casualty rate in their group and will change their minds. One can hope.
This article is written by Jeffrey Frankel.
