Data: COVID-19 Trends Across Countries of Varying “Regime Types”

At the start of the pandemic — when things were *really* quiet — I relearned Python through some fantastic Udemy courses, with a focus on Pandas. This was one of my key first steps toward dabbling with data science, an ongoing journey that I am just starting. I found one of the best ways to practice these emerging skills was through COVID-19 data, specifically from the Covid19API. I mainly practiced on comparing daily statistics across various countries, or US states, but in this article I wanted to present some exploratory data analysis comparing overall COVID stats by countries…with a few other parameters.

Before I lose you with prose, here is a fun graphic to start us off:

Figure 1: This graphic shows the overall confirmed COVID cases per capita, binned by “regime type” as defined by the Economist Intelligence Unit (EIU). Each dot represents a country. The democracies (as defined by the EIU) appear to have a greater distribution of confirmed cases per capita, while the “authoritarian” and “hybrid regime” countries appear to show some clustering at the lower end of the spectrum (implying they have fewer cases of COVID per capita). There are numerous ways to interpret these findings, addressed later.

OK, let’s back this up.

I was curious to see how countries’ COVID stats compared to each other, factoring in population, GDP, and its type of government (democracy vs autocracy). So when we first pull the data from the API, the result looks like:

Then we bring in population, GDP, and “regime type” as defined by the Economist Intelligence Unit, and we get:

To bring in the above variables, I didn’t do a merge between tables using the Country name, because that may have been spelled differently across the tables. I used the fuzzywuzzy Python library but had to hardcode in some country equivalencies between datasets because of false matches, and especially because the EIU did not provide a democracy rank or regime type label for numerous countries. See end of this post for more details.

Of course, to compare how “well” countries fared during the COVID pandemic (in terms of confirmed cases and deaths), one has to factor in the vastly different population sizes. So I focused the analysis on the per capita rates. Some fun with functions brings us to:

Now we can start playing with the data. But some obligatory caveats:

1. The data of course do not account for all cases. With the rise of at-home testing (especially during the omicron wave), many cases are not being reported.
2. Some countries may have particularly flawed reporting, and countries have widely varying infrastructure to account for COVID cases.

Let’s start off by looking at the potential relationship between GDP Per Capita and the death rate, or percentage of confirmed cases that resulted in deaths ((deaths/confirmed) * 100):

Figure 2: As GDP per capita goes up, death rate goes down.

There is a lot going on in this graphic. Each dot represents a country; the color of the dots correspond to the regime type; the size of the dots correspond to where that country ranks in overall democracy per 2020 rankings (1 is the highest form of democracy — Norway — and the democracy gets worse in ascending order). The clearest finding in the above is that as GDP per capita increases, the death rate decreases. In addition, the higher ranked democracies (the smaller the dot, the better the democracy ranking) also have higher GDP per capita, and those countries also tend to fall into the “full democracy” (green) category. The “flawed democracy” (blue) category of countries tends to follow the same trend as the “full democracy.” The “authoritarian” (red) and “hybrid regime” (orange) countries fare the worst — they may have higher death rates and lower GDP per capita.

Let’s take a look at the potential relationship between confirmed per capita and deaths per capita:

Figure 3: As confirmed per capita cases go up, deaths per capita go up, unsurprisingly.

This is perhaps unsurprising. As confirmed per capita cases increase, deaths per capita increase. Notably, the spread of confirmed per capita and death per capita seems dominated by “full democracies” (green) and “flawed democracies” (blue), while “authoritarian” countries show a higher density toward 0,0. Let’s choose another view of these same data:

Figure 4: Jointplot view of the previous graphic. The distributions of “authoritarian” and “hybrid regime” countries appears to be more concentrated at the lower end of the confirmed per capita and death per capita spectra, certainly relative to the distributions of “full democracies” and “flawed democracies.”

The above findings seem to demonstrate that “authoritarian” and “hybrid regime” countries are more concentrated at the lower end of the spectrum, for both confirmed per capita and deaths per capita. The implication here is that the worse the democracy, the lower the confirmed per capita and deaths per capita.

Let’s look at yet another view to see more clearly how these different regime types compare to each other:

Figure 5: The democracies (green and blue) demonstrate much more distribution of overall confirmed per capita and deaths per capita, while the authoritarian and hybrid regime countries seem clustered to lower rates of both (with some spread).

The above graphics appear to show a much wider distribution of confirmed per capita and deaths per capita amongst the democracies, relative to the clustering shown in the “authoritarian” and “hybrid regime” categories at the lower end of the spectra. Some reasons that come to mind:

1. Authoritarian and hybrid regimes are societies with stricter government control, which can thus allow for harsher mitigation measures (stricter lockdowns, etc.) that decrease COVID rates. Democracies may have governments with more complicated dynamics at play, leading to less stringent lockdowns.
2. Some authoritarian and hybrid regimes may have less trustworthy reporting, and thus their numbers may appear falsely low. Open societies may report more reliable data.
3. Authoritarian and hybrid regimes may be countries with lower GDP per capita (as seen in Figure 2 above), and as such may lack the resources or infrastructure to appropriately gather and track the necessary statistics.

Finally, I’m interested in how the government’s “regime type” may affect the overall percentage of COVID cases that result in death. Do democracies experience fewer deaths relative to their confirmed case rate? Or vice versa, might the data show that more authoritarian regimes experience fewer deaths?

Overall, democracies — particularly “full democracies” — appear to experience a lower death rate from confirmed cases.

While “authoritarian” countries certainly show a cluster of cases under ~2.5% death rate, “full democracies” appear to fare better overall, experiencing fewer deaths out of all its confirmed COVID cases. But we certainly have a discrepancy in the number of countries that fit into each regime type category, so let’s look at the averages:

Indeed, full democracies display an average lower death rate (percentage of COVID cases that result in death) than other regime types.

“Full democracies” appear to have lower death rates, out of all the confirmed COVID cases. This matches one of the trends we saw above (Figure 2), comparing GDP Per Capita (on the X axis) vs. death rate (on the Y axis), in which full democracies (the green dots) appeared to show a lower death rate.

Let’s take a more granular look at just how much better full democracies fare with death rates. We’ll use the democracy ranks from the EIU:

As the democracy ranking gets worse (higher), the death rate (percentage of COVID confirmed cases that result in death) appears to get higher, or at least a greater distribution appears, with more countries experiencing a higher death rate.

Throughout these findings, the worse democracies (“hybrid regime” and “authoritarian” countries) appear to have lower confirmed per capita and deaths per capita statistics, relative to the democracies; and “flawed democracies” appear to have particularly higher confirmed/deaths per capita relative to the other countries. While one must question the reliability of the data coming from less democratic countries (“hybrid regime” and “authoritarian”) the above findings also demonstrate that the democracies have lower death rates — the percentage of confirmed COVID cases that result in deaths. This finding may also be tied to the relatively higher GDP per capita found amongst the democracies.

Caveats in the above analysis:

1. Data from the Andorra, Antigua and Barbuda, Bahamas, Barbados, Belize, Brunei, Dominica, Grenada, Kiribati, Kosovo, Liechtenstein, Maldives, Marshall Islands, Micronesia, Monaco, Palau, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and Grenadines, Samoa, San Marino, Sao Tome and Principe, Seychelles, Solomon Islands, Somalia, Tonga, Vanuatu, Vatican were included in the regime type analysis as “Unassigned” because they were omitted by the EIU’s assessment (and Vatican was left out of the analysis altogether because it was highly anomalous). These countries were also excluded from the above analysis that accounted for “democracy rank.”
2. The data used were from January 26, 2022.
3. Deaths lag confirmed COVID cases, so the above analysis does not take into account the subsequent deaths that may have resulted from the number of confirmed COVID cases on January 26, 2022 (when the omicron variant was hitting the world).