Political regime and COVID 19 death rate: Efficient, biasing or simply different autocracies?An econometric analysis.

The difference in COVID 19 death rates across political regimes has caught a lot of attention. The "efficient autocracy" view suggests that autocracies may be more efficient at putting in place policies that contain COVID 19 spread. On the other hand, the "biasing autocracy" view underlines that autocracies may be under reporting their COVID 19 data. We use fixed effect panel regression methods to discriminate between the two sides of the debate. Our results present a more nuanced picture: once pre-determined characteristics of countries are accounted for, COVID 19 death rates equalize across political regimes during the first months of the pandemic, but remain largely different a year into the pandemic. This emphasizes that early differences across political regimes were mainly due to omitted variable bias, whereas later differences are likely due to data manipulation by autocracies. A year into the pandemic, we estimate that this data manipulation may have hidden approximately 400,000 deaths worldwide.

Introduction

While democratic countries have previously been shown to overperfom compared to autocracies with respect to health outcomes (Besley & Kudamatsu, 2006; Bollyky et al., 2019; Franco et al., 2004; Kudamatsu, 2012; Pieters et al., 2016), data shows that, in the specific case of the COVID 19 pandemic, democratic countries may be fairing much worse (Sorci, Faivre, & Morand, 2020). Fig. 1 presents the evolution of the cumulative COVID 19 death rate across political regimes. It can be seen that the divergence between autocracies and democracies took place in two distinctive moments. A “first wave” of divergence happened in the first 25–50 days of the pandemic, during which democracies strongly diverged from autocracies. Then, the gap between the regimes only slowly increased over time until a “second wave” took place roughly 8 months into the pandemic, when democracies' divergence from autocracies accelerated drastically. Therefore, a year after the beginning of the pandemic,1 democratic countries’ COVID 19 death rate is on average larger than that of non democratic countries by approximately 42 per 100,000 inhabitants. That is, a year into the pandemic, the fatality rate in a democracy is on average 3.7 times larger than in an autocracy.

Fig. 1

Evolution of COVID 19 data reporting by political regime, time since first 0.4 cases per 100,000.

During the first wave of divergence, a debate (Ang, 2020) has emerged trying to unpack the reasons behind such wide differences across political regimes: a priori, all other things equal, the political regime should not be related to the spread of a disease. We distinguish three main hypotheses to explain this difference.

A first interpretation relates to the relative efficiency of social distancing measures in democracies and autocracies. Some have argued that democracies may be less well equipped to implement and enforce social distancing policies (Cepaluni et al., 2020; Narita & Sudo, 2021; Sorci et al., 2020), or that they may be implementing them with a suboptimal timing (Cheibub et al., 2020; Karabulut et al., 2021; Sebhatu et al., 2020). That is, in this view, autocracies are more able to implement social distancing measures. We will refer to this interpretation as the efficient autocracy hypothesis.

A second interpretation is that there may be voluntary misreporting of COVID 19 data, in particular by non democratic countries. For example, Tuite, Ng, et al. (2020) report that Egypt may have underreported its number of cases, Tuite, Sherbo, et al. (2020) report that Iran may also have underreported its number of cases, while Kavanagh (2020) discusses that China's political regime may have hindered its initial response to the pandemic. Adiguzel et al. (2020), Kapoor et al. (2020) and Badman et al. (2021) document that autocracies' COVID 19 data present signs of manipulation, while Annaka (2021) show that correlates of data manipulation contribute to explaining the difference in death rate between democracies and autocracies. In this view, there are systematic differences between the real and the reported death rate. When these differences are voluntary, they are systematically linked to the type of political regime. We will refer to this interpretation as the biasing autocracy hypothesis.

A third interpretation has caught less attention (Ashraf, 2020): democracies and autocracies tend to have systematically different characteristics apart from their political regimes. These differences, once accounted for, may in fact be sufficient to explain the difference in both the real and reported death rate. This would leave the contributions due to voluntary under-reporting or differences in policies to matter only marginally. An example of such differences would be that autocracies tend to have much younger populations (and therefore, a much smaller real death rate, all other things equal) but also a lower ability to test (and therefore, a much smaller reported death rate, all other things equal). We refer to this interpretation as the simply different autocracy hypothesis.

The three aforementionned hypotheses are not mutually exclusive, and simple reduced form econometric methods can help measuring how much each of them contributes to explaining the differences observed across political regimes. Take the case where the econometrician only observes a reported death rate rather than the real death rate but can observe the variables determining COVID 19 real and reported death rate. Also assume that there are two such types of variables: fixed characteristics2 (say, the share of the population aged 65 or older who would determine real death rates or the number of hospital beds per capita who would determine both real and reported death rates) and policy response. Under the efficient autocracy hypothesis, regressing the reported death rate on a measure of democracy and controlling for all fixed parameters would lead to a positive and significant coefficient on democracy. However, further controlling for policy response in the regression should bring the coefficient on democracy close to zero and render it non significant. That is, all the differences observed between democracies and autocracies in their reported death rate, once fixed characteristics are accounted for, would be due to the difference in policy response across these two types of regime. In this case, there may be a difference between the real and the reported death rate, but this difference is not systematically linked to the political regime. In fact, these results would indicate that the policy response of autocracies is better than that of democracies, from the perspective of COVID 19 death rate.

Under the biasing autocracy hypothesis, in a regression of reported death rates on a measure of democracy and all relevant controls (including policy response), the coefficient on democracy should be positive and significant. That is, despite controlling for all relevant characteristics and policy response, there still is a systematic difference between democratic and non democratic countries which is not accounted for. In that case, the only reason why a difference may remain would be due to systematic underreporting of casualties by non democratic regimes. This would be due to the fact that the difference between the real and the reported death rate is always larger for autocracies.3 While the real death rate would be identical once all confounding factors are accounted for, the reported death rates remain different even when controlling for the characteristics influencing non voluntary under reporting.

We use daily level data of COVID 19 death rates of 137 countries for the first year of the epidemic and resort to simple reduced form econometric methods using the panel structure of the data. First we start by looking into the evolution of daily total death rates across political regimes, using a regression with no controls except country fixed effects (Regression 1). We then include controls for fixed characteristics of countries that are likely to determine the real COVID 19 death rate and allow them to matter differently across time (Regression 2). Finally, we also include controls for the stringency of social distancing measures and allow these to matter differently across time (Regression 3). Comparing Regression 2 to Regression 3 addresses the efficient autocracy hypothesis: any difference between the coefficient on democratic regime between Regressions 2 and 3 would be due to the differential in policy response across political regimes. An increase would indicate that autocracies implement more stringent social distancing measures that are successful in decreasing the death rate. Comparing Regression 1 to Regression 3 addresses the biasing autocracy and the simply different autocracy hypothesis: once all controls for both fixed characteristics and policy response are accounted for, does the difference between autocratic and democratic regimes remain (biasing autocracy hypothesis) or vanish (simply different autocracy hypothesis)?

Our results indicate that the inclusion of controls for country characteristics and policy response is in fact enough to remove almost all cross regime difference in COVID 19 mortality rates during the first wave of divergence. However, these controls are not sufficient to account for the second wave of divergence. Therefore, while the initial differences across political regimes were due to the fact that democracies and autocracies are simply different, we find evidence for the biasing autocracy hypothesis during the second wave of divergence. A year into the pandemic, we estimate that around 400,000 deaths - or 13% of total deaths - may have been hidden due to data manipulation. We find little support for the efficient autocracy hypothesis.

Section 2 of this paper presents the data. Section 3 elaborates on the methodology used to test our hypotheses. Section 4, 5 present our main results. Finally, Section 6 provides a discussion of our main findings and Section 7 concludes.

Data

In order to investigate our hypotheses, we assemble a dataset that comprises information on daily cases and deaths in the first 352 days of the pandemic for 137 countries. Our dependent variable, the daily country-level total number of reported cases and reported deaths due to the COVID-19 virus is from Dong et al. (2020) .4 Our main variable of interest, the classification of political regimes along the autocratic-democratic scale, comes from the Polity 5 project (Center for Systemic Peace, 2018).

Under the simply different autocracy hypothesis, accounting for the differences in characteristics of countries would suffice to explain the difference in reported mortality rates across political regimes. We therefore collected an extensive array of country level variables. To proxy for income and health infrastructure differences, we gathered data on gross domestic product per capita in 2018 from the World economic outlook survey (IMF), and completed it with the World Factbook (CIA). Furthermore, information on the number of available hospital beds (per thousand inhabitants) is retrieved from the World Bank to account for differences in health infrastructure that may drive the mortality difference (actual and reported death rates).5

To capture differences in demographic characteristics which may explain the speed of the spread of the disease, we use data on countries' total population and density in 2019 from the World Bank, and data on countries' urbanization rate in 2019 from the World in Data website. To control for the effect of geographical characteristics, we collect data on the latitude and longitude of each country's capital from the World Cities Database, and classify each country according to its World Bank region.6 Finally, to control for population risk of mortality, we include the share of population aged 65 or older (from the World Bank) and, since air pollution has been shown to be associated with COVID 19 death rates (Zhu et al., 2020), we use summary exposure values to ambient ozone pollution and ambient particle matter pollution from the Global Burden Disease dataset (2017).

To test the efficient autocracy hypothesis, we use information on countries’ different COVID 19 containment policies from the “Variation in Government Responses to COVID-19” dataset (Hale et al., 2020). This dataset includes a daily policy stringency index based on the aggregation of 17 policy indicators.7

Given that the data on our dependent variable is at the daily level, this allows us to construct a panel dataset that comprises a total of 137 countries,8 classified as either democratic or non democratic, for which we have information on all the previously mentioned national characteristics. Therefore, our dataset displays information (by day and by country) on the total number of reported deaths due to the COVID-19 virus, on the stringency of policy measures taken by a given country, and on all other relevant characteristics of that country. We focus on the first year since the beginning of the pandemic in each country, which we define as having more than 0.4 cases per 100,000 inhabitants. 9

Methodology

Given the country-day panel structure of our data, we resort to fixed effect panel reduced form econometric methods to look into the differences in COVID 19 casualty rates across political regimes and time. This method allows us to remove the influence of all time invariant differences across countries by including countries fixed effects. This further allows us to control for an extensive set of countries’ pre-determined characteristics and for differences in containment policies across countries.

We specify the following regression equation, which we run using Ordinary Least Squares:

DeathRate is the inverse hyperbolic sine transformation of daily declared total deaths per 100,000 inhabitants in country c10, t days after the beginning of the pandemic in country c. democratic is the 2018 polity score of country c. time.from.start is a set of fixed effect for each day since the beginning of the pandemic. The interaction of democratic with time.from.start allows us to track day by day the evolution of the difference in death rates across political regimes, a standard approach in economics (see Duflo (2001) or Cassan (2019) among others).

X is a large set of controls for countries’ pre determined characteristics: GDP per capita, number of hospital beds per 1000, population, density, urbanization rate, share of population aged 65 or older, summary exposure value to particle matters pollution, summary exposure value to ambient ozone pollution, as well as for the square level of these variables, World Bank regions fixed effects, latitude and longitude. We interact all these variables with the time.from.start fixed effects to allow their effect to vary over time.

Y is a measure of country policy response to the pandemic. It is a stringency index of governmental response (as measured at t-15 to allow for lags in its effect: policy response affects the probability of infection on a given day, which affect mortality only two weeks after). We also include the square level of this variable to allow for non-linear effects. Furthermore, we interact these variables with the set of time.from.start fixed effects, to allow their effect to vary over time. δ is a set of country fixed effects. Finally, ω is a set of day of the week fixed effect interacted with time.from.start fixed effects, to control for variations in reporting across days of the week.

We perform this regression iteratively. First, we do not implement any of the X and Y controls (Regression 1). This allows us to see the evolution of the difference in casualty rates across political regimes when no confounding factors are accounted for. Then, we implement X but not Y (Regression 2). This will allow us to see how much of the difference across political regimes survives once the different pre-determined characteristics of countries are accounted for. Finally, we add the Y policy response controls (Regression 3).

This iterative procedure allows us to address the different sides of the debate on the role of political regime in fighting COVID 19. Comparing Regression 2 to Regression 3 addresses the efficient autocracy hypothesis: any difference between the β coefficients on democratic regime between Regressions 2 and 3 would be due to the differential in policy response across political regimes. An increase would indicate that autocracies implement more stringent social distancing measures that are successful in decreasing the COVID 19 death rate.

Comparing Regression 1 to Regression 3 addresses the biasing autocracy and simply different autocracy hypotheses: once all controls for both fixed characteristics and policy response are accounted for, do the β coefficients remain positive (biasing autocracy) or do they equalize to zero (simply different autocracy)? Note that these hypotheses are not mutually exclusive: autocracies may well be efficient, biasing and different at the same time. Our methodology allows to capture this possibility: if the β coefficients decrease but remain large and significant when passing from Regression 1 to Regression 3 and change but remain large and significant between Regression 2 and Regression 3, then this would support the simultaneous presence of the three hypotheses.

Results

Fig. 2 presents the β coefficients from Equation (1) for all three versions of the specification.11 The first panel presents the results of Regression 1, when no controls excepting country and day of the week fixed effects are included. The divergence between political regimes takes place in two waves. During the first two months of the pandemic, a first wave of divergence drives death rate between political regimes apart. From around day 60 to day 230, this wave slowly recedes, whereas a second wave of divergence emerges from day 230 onward.

Fig. 2

Evolution of COVID 19 deaths per 100,000 since 0.4 cases per 100,000.95% CI.

The second panel includes controls for pre-determined characteristics interacted with day fixed effects. The β coefficients become close to zero and statistically insignificant for the first wave, but start to increase slowly from day 30 onwards, to become significant roughly 200 days into the pandemic. That is, once countries’ differences in characteristics are taken into account, the difference in death rates across political regimes during the first wave does not survive. However, over the course of time, these characteristics alone are not sufficient to fully account for the difference.

The third panel adds controls for countries' policy response to the pandemic. Our coefficients of interest β remain virtually unaffected. That is, our results do not support the efficient autocracy hypothesis.12 Therefore, once systematic differences across countries’ characteristics and policy responses are taken into consideration, the differences in death rates apparent in Fig. 1 and in the first panel of Fig. 2 vanish for the first wave of divergence, but are still present for the second wave. The reason why reported COVID 19 death rates differ across political regimes is fully accounted for by factors which systematically differ between democracies and autocracies in the first months of the pandemic, but not over the long run. That is, our results indicate that the simply different autocracy hypothesis is prevailing during the first wave of divergence, but are also consistent with the biasing autocracy hypothesis in the longer run. Our results do not support the hypothesis that autocracies are more efficient at controlling the pandemic but show that they may be voluntarily under reporting casualty more often, even if the latter was probably not dominant in the first months of the pandemic.13

How many deaths are hidden?

The β coefficients of Regression 3, presented in the third panel of Fig. 2, are strongly suggestive that even once a very large set of characteristics and policies are taken into account, a year into the pandemic, there are differences in COVID 19 mortality rates across political regimes that are not accounted for. Under the assumption that our extensive set of controls captures the determinants of COVID 19 mortality rates, this suggests that autocracies may be manipulating their reported COVID 19 death rate.

In this section, we use the results of our estimations to answer the following question: how many COVID 19 deaths are hidden because of manipulations by autocracies? The β coefficients can be interpreted as the percentage change in deaths per capita due to an increase of one unit of the Polity score of a country. Therefore, we can compute the unbiased COVID 19 death rate as:Where DeathRate is the declared COVID 19 death rate of country c, t days after the beginning of the pandemic, beta the coefficient on democratic of Regression 3 and polity country c's polity score. This unbiased death rate tells us what the reported death rate of country c would have been if that country has had a polity score of 10,14 all other things equal. Based on this unbiased death rate, we compute each country's unbiased number of deaths. Fig. 3 presents the results of this exercise. It can be seen that a year into the pandemic, around 400,000 deaths have been hidden due to data manipulation by autocracies, which represents roughly 13% of total deaths in the world.

Fig. 3

Number and share of hidden deaths linked to lack of democracy.

Discussion

A few remarks are in order to help interpret our results. First, one should keep in mind that the variables that we consider pre-determined characteristics, such as the GDP per capita, are only pre-determined in the time horizon that we are considering. Over the long run, they are an outcome of the political regime. See for example Acemoglu et al. (2019), who show that democracy causes growth. In that sense, our results do not take into consideration the long term effect of political regimes on the variables that may determine COVID 19 death rates.

For instance, a better health care system will lead to both a lower real death rate (infected individuals are better treated) and a higher reported death rate (infected individuals’ death is better attributed to COVID 19). If, as has been argued in the literature (Besley & Kudamatsu, 2006; Bollyky et al., 2019; Franco et al., 2004; Kudamatsu, 2012; Pieters et al., 2016), democracies tend to have better health care policies; in the long run, the health care system (which we consider as pre-determined) will be better in democracies because of the political regime, which will causally affect both real and reported death rates across political regimes.

Second, it is important to remember that our methodology has some limitations. It can not account for all forms of misreporting of the data. Since our method is in essence comparative, we can only estimate differences in misreporting. So our results can only be interpreted as a lower bound of overall misreporting.15

Another limitation of our method is its residual approach: once confounding factors are taken into account via the extensive set of control variables that we include in our regressions, the remaining variation should not be correlated with the political regime in the absence of data manipulation. This is true if indeed we control for all confounding factors that are correlated both with political regime and COVID 19 death rate. It is however impossible to be certain that all such variables are controlled for. Therefore, our results can be interpreted as suggesting that during the second wave, autocracies manipulated COVID 19 data, but not as a definitive proof of such manipulation.

Third, our focus is only on COVID 19 death rates. Arguably, however, one may have wanted to study death rates from all causes rather than just from COVID 19. Even in times of pandemic, governments should aim at preserving the health of their citizens from all sources of harm, not from one specific cause only. In a time during which most of the attention is drawn towards COVID 19 death rates rather than towards death rates in general, a pro-democracy argument would be that while there does not seem to have been differences across political regimes for COVID 19 death rates during the first wave of divergence, this may hide the fact that autocracies have focused on decreasing COVID 19 death rate at the expense of deaths from other sources.

One could develop this idea even further and argue that democracies have higher COVID 19 mortality rates because they are better at preventing non COVID 19 deaths, leading to a population which is on average older and therefore more likely to die if infected by COVID 19. This question can unfortunately not be tackled with the available data, and we leave it to future research (when mortality data from all causes will be available for a sufficient number of countries), but note that the differences in countries' population's susceptibility to die from COVID 19 upon contamination seem to be one of the main drivers of the difference in COVID 19 mortality rates across political regimes during the first wave of divergence, all other things equal.

Fourth, our findings do not contradict previous studies on under reporting of COVID 19 data during the first wave of divergence, in particular country specific studies. Indeed, because of the statistical analysis used, our results do not imply that no single country underreported or manipulated its COVID 19 mortality data. However, our results do address the widespread idea that autocracies were systematically and willingly under reporting COVID 19 casualties during the first wave of divergence. What our results do indicate is that under reporting during the first wave of divergence (by any political regime) was primarily due to the different characteristics of countries that are correlated with the political regime rather than a direct causal effect of the political regime. This however, is not true for the second wave of divergence, for which we find support for the view that autocracies manipulated COVID 19 death rates.

That is, a plausible interpretation is that autocratic governments may well have been under reporting data during the first wave while not manipulating it. One could argue that even if autocracies were under reporting COVID 19 death rates, this may have been primarily driven by their overall incapacity to link death to its cause rather than to a direct attempt at data manipulation. The low reported COVID 19 death rate in autocracies may in part be due to the lower level of development of both the public health infrastructure and the statistical apparatus of autocracies. However, and this goes back to our first point, over the long run, public health infrastructure and statistical apparatus may well be determined by the political regime.

Conclusion

We investigated the COVID 19 death rate gap between democratic and autocratic countries. We uncovered that it widened in two phases that we call waves of divergence. We formulated three main hypotheses based on the previous literature to explain them: they can be due to the fact that autocracies are more efficient at implementing restricting policy measures; that autocracies are underreporting their COVID 19 data and that autocracies simply have different characteristics that can explain the death rate gap. Our analysis, relying on simple econometric tools, allows to make progress in the debate around the sources of the observed differences in COVID 19 death rates across political regimes.

We show that once pre-determined characteristics and policy responses are taken into account, COVID 19 death rates do not exhibit any difference across political regimes during the first wage of divergence: the coefficients on democracy become precisely estimated zeros. For this initial period, our results therefore do not show support neither for the efficient autocracy nor for the biasing autocracy hypotheses, as we do not find evidence that autocracies are neither systematically better at preventing COVID 19 death nor that they are more often under reporting casualties.

However, several months into the pandemic, a second wave of divergence between democracies and autocracies emerged. Our results indicate that this second wave can not be fully explained by differences of characteristics across political regimes. We therefore find support for both hypotheses of biasing autocracy and simply different autocracy. According to our estimates, had all the countries in our sample been fully democratic, the number of reported deaths would have increased by approximately 400,000 or 13% of the deaths at the time.

Author statement

Guilhem Cassan; Planning: original idea and implementation, Conduct: data analysis. Reporting: Writing.

Milan Van Steenvoort; Conduct: data collection and analysis, Reporting: Writing.

Funding

We are grateful to Jeremie Decalf, Romain Lutaud, Glenn Magerman, Marc Sangnier and Vincenzo Verardi for helpful discussions and suggestions. We thank seminar participants at UNamur. Guilhem Cassan thanks CEPREMAP and the for financial support. Research on this project was financially supported by the Excellence of Science (EOS) Research project of O020918F. All errors remain our own.

Declaration of competing interest

None.

Table1

Data sources

Data	Source
COVID19 Death Rate	Dong et al. (2020)
COVID19 Cases	Dong et al. (2020)
Democratic	Polity 5 project (Center for Systemic Peace, 2018)
	Freedom in the world (Freedom House, 2021)
Stringency Index of Policy Response	Variation in Government Responses to COVID-19 (Hale et al., 2020)
Gross Domestic Product per capita	World Economic Outlook, IMF (2018) + World Factbook, CIA (2018)
Share of 65+	World Bank (2019)
Population Density	World Bank (2019)
Population	World Bank (2019)
Urbanization Rate	World in Data
Hospital Beds per 1000	World Bank
Summary Exposure Value to Air Pollution	Global Burden of Disease (2017)
Summary Exposure Value to Ambient Ozone Pollution	Global Burden of Disease (2017)
Latitude and Longitude	World Cities Database
World Regions	World Bank

Table2

Descriptive statistics

	Mean	sd	Min	Max
Polity score	4.51	6.13	−10.00	10.00
Democratic (polity>0)	0.76	0.43	0.00	1.00
Democratic (polity>6)	0.55	0.50	0.00	1.00
Free or Partly Free according to Freedom House	0.74	0.44	0.00	1.00
Total deaths per 100,00 - Inverse Hyperbolic Sine Transformation	1.72	1.48	0.00	5.96
Share of 65+	9.56	6.64	1.16	28.00
GDP Per Capita	23,632.97	23,539.24	727.17	132,886.39
Population in million	52.75	169.22	0.55	1397.71
Hospital beds per 1000	2.96	2.48	0.10	13.40
Population Density	206.56	777.60	2.06	8829.05
Urbanization Rate	61.96	21.96	13.37	100.00
Summary exposure value to ambient ozone pollution - Age standardized	36.25	12.55	2.22	52.87
Summary exposure value to ambient particulate matter pollution - Age standardized	32.35	17.27	6.80	90.02
Stringency t-15	60.66	21.07	0.00	100.00
Observations	48,351

Table 3

Polity score and Freedom classification, by country

Country	Polity Score	Freedom Classification
Afghanistan	−1	Not Free
Albania	9	Partially Free
Algeria	2	Not Free
Argentina	9	Free
Australia	10	Free
Austria	10	Free
Azerbaijan	−7	Not Free
Bahrain	−10	Not Free
Bangladesh	−6	Partially Free
Belarus	−7	Not Free
Belgium	8	Free
Benin	7	Partially Free
Bhutan	7	Partially Free
Bolivia	7	Partially Free
Botswana	8	Free
Brazil	8	Free
Bulgaria	9	Free
Burkina Freeaso	6	Partially Free
Burundi	−1	Not Free
Cabo Verde	10	Free
Cambodia	−4	Not Free
Cameroon	−4	Not Free
Canada	10	Free
Central African Republic	6	Not Free
Chile	10	Free
China	−7	Not Free
Colombia	7	Partially Free
Costa Rica	10	Free
Croatia	9	Free
Cuba	−5	Not Free
Cyprus	10	Free
Czech Republic	9	Free
Denmark	10	Free
Djibouti	3	Not Free
Dominican Republic	7	Partially Free
Ecuador	5	Partially Free
Egypt, Arab Rep.	−4	Not Free
Country	Polity Score	Freedom Classification
El Salvador	8	Partially Free
Estonia	9	Free
Eswatini	−9	Not Free
Ethiopia	1	Not Free
Fiji	2	Partially Free
Finland	10	Free
France	10	Free
Gabon	3	Not Free
Gambia, The	4	Partially Free
Georgia	7	Partially Free
Germany	10	Free
Ghana	8	Free
Greece	10	Free
Guatemala	8	Partially Free
Guinea	4	Partially Free
Guyana	7	Free
Haiti	5	Partially Free
Honduras	7	Partially Free
Hungary	10	Partially Free
India	9	Free
Indonesia	9	Partially Free
Iran	−7	Not Free
Iraq	6	Not Free
Ireland	10	Free
Israel	6	Free
Italy	10	Free
Jamaica	9	Free
Japan	10	Free
Jordan	−3	Partially Free
Kazakhstan	−6	Not Free
Kenya	9	Partially Free
Korea, Rep.	8	Free
Kuwait	−7	Partially Free
Kyrgyz Republic	8	Partially Free
Latvia	8	Free
Lebanon	6	Partially Free
Liberia	7	Partially Free
Libya	−7	Not Free
Country	Polity Score	Freedom Classification
Lithuania	10	Free
Luxembourg	10	Free
Madagascar	6	Partially Free
Malawi	6	Partially Free
Malaysia	7	Partially Free
Mali	5	Partially Free
Mauritius	10	Free
Mexico	8	Partially Free
Moldova	9	Partially Free
Mongolia	10	Free
Morocco	−4	Partially Free
Mozambique	5	Partially Free
Myanmar	8	Not Free
Nepal	7	Partially Free
Netherlands	10	Free
New Zealand	10	Free
Nicaragua	6	Not Free
Niger	5	Partially Free
Nigeria	7	Partially Free
Norway	10	Free
Oman	−8	Not Free
Pakistan	7	Partially Free
Panama	9	Free
Paraguay	9	Partially Free
Peru	9	Free
Philippines	8	Partially Free
Poland	10	Free
Portugal	10	Free
Qatar	−10	Not Free
Romania	9	Free
Russian Federation	4	Not Free
Saudi Arabia	−10	Not Free
Singapore	−2	Partially Free
Slovakia	10	Free
Slovenia	10	Free
Spain	10	Free
Sri Lanka	6	Partially Free
Sudan	−4	Not Free
Suriname	5	Free
Country	Polity Score	Freedom Classification
Sweden	10	Free
Switzerland	10	Free
Syrian Arab Republic	−9	Not Free
Tajikistan	−3	Not Free
Tanzania	3	Partially Free
Thailand	−3	Partially Free
Timor-Leste	8	Free
Togo	−2	Partially Free
Trinidad and Tobago	10	Free
Tunisia	7	Free
Turkey	−4	Not Free
Uganda	−1	Not Free
Ukraine	4	Partially Free
United Arab Emirates	−8	Not Free
United Kingdom	8	Free
United States	5	Free
Uruguay	10	Free
Uzbekistan	−9	Not Free
Venezuela	−3	Not Free
Vietnam	−7	Not Free
Yemen, Rep.	3	Not Free
Zambia	6	Partially Free
Zimbabwe	4	Partially Free

Table 4

Time to first cases and political regime

	0.4 c. per 100,000	0.6 c. per 100,000	100 c.
Democracy	−0.27	−0.20	0.95
	(0.54)	(0.59)	(0.79)
R-sq	0.58	0.58	0.42
Observations	137	137	137

Heteroskedasticity-robust standard errors in parentheses * p < .10 **p < .05 ***p < .01. Controls included are: GDP per capita, population, density, urbanization rate, share of 65 and above, number of hospital beds per capita and the square of all preceding variables, latitude, longitude, World Bank region fixed effect.