The Truth and Myth of the Advantages of Authoritarian Countries to COVID-19 *

,


Introduction
The number of COVID-19 deaths is reported to exceed one million across the world.
Some argue that people, especially in democratic countries, face a tradeoff between freedom and health (Alsan et al., 2020;Norheim et al., 2020;Thomson and Ip, 2020). Recently published papers also reveal that democratic countries suffer from more COVID-19 deaths than authoritarian states (Cepaluni et al., 2020;Cheibub et al., 2020;Frey et al., 2020). Figure 1 supports these arguments to some extent. It shows the total number of COVID-19 deaths per 1 million (as of December 12, 2020) on the vertical axis, as reported by Worldometer COVID-19 Data, and the level of Polity2 on the horizontal axis from the Polity V Project (Marshall et al., 2020). The correlation coefficient between the two variables is 0.3758, and it is statistically significant at the 1% level. In the case of an alternative measure of political regime, the relationship is more apparent. Figure 2 shows the relationship between the Multiplicative Polyarchy Index (MPI) from the Variety of Democracy (V-Dem) Project . The correlation coefficient between these two variables is 0.4816. These moderate, positive relationships suggest that the arguments should be correct. However, are these relationships accurate? This article attempts to answer this question. Why can these arguments be questioned? For example, it has been reported that the Belarusian President, Alexander Lukashenko, underestimated the risk of COVID-19 spreading in the country (the country's Polity2 is -7). Thus, the president did not take any appropriate measures to prevent the pandemic's spread. As this case implies, authoritarian governments do not necessarily take decisive measures immediately. However, the country has one of the lowest death rates in Europe (Karáth, 2020). If authoritarian governments do not take strong measures to combat COVID-19, the argument that government intervention can reduce COVID-19 deaths by reducing confirmed cases is not persuasive.
It is difficult to imagine that authoritarian states' medical systems can work better than those of democratic countries. Scholars have highlighted that people in democratic countries are likely to have better health than their authoritarian counterparts (Wang et al., 2019;Gerring et al., 2020;Kavanagh and Singh, 2020). Another possibility is that authoritarian countries manipulate death data. Kapoor et al. (2020) analyzed the moving average of the reported number of deaths, revealing that the data are unnaturally produced. Adiguzel et al. (2020) also pointed out a similar result to that of digit-based tests. This paper proposes another possible determinant that affects the advantages of authoritarian countries in combating COVID-19. It demonstrates that authoritarian countries tend to perform more tests to detect COVID-19 carriers, leading to lower death rates than their democratic counterparts.

Determinants of COVID-19 Deaths
First, this paper analyzes the relationship between political regimes and COVID-19 deaths, as the graphs above suggest. The total number of deaths was obtained from Worldometer COVID-19 data. Daily data are available from another source. However, almost all other covariates necessary to be included in the analysis are yearly data, such as GDP per capita. This study constructs cross-sectional data on over 100 countries for all statistical analyses below. Political regime variables are obtained from the Polity Project and Variety of Democracy (V-Dem) Project. Control variables such as GDP per capita, trade ratio to GDP, total population, population density, and population ratio age 65 and above are taken from the World Bank. The latitude and days since the first confirmed case were also included in the analysis. The latest available yearly data were used. Negative binomial regression was applied to consider the dependent variable's skewed distribution.
The independent variables (except for latitude and days since the first confirmed case) are logged to consider skewed distributions. Taking the logs of the dependent variables leads to missing values for countries where COVID-19 cases and deaths are not reported.
But this is not a problem because it is almost meaningless for this study to analyze such countries if those countries are not affected by COVID-19. The descriptive statistics are presented in Appendix 1. Table 1 shows the results of the regression results for the determinants of the death cases that report the incidence rate ratio (IRR) instead of coefficients.
Model 1 analyzes the relationship between Polity2 from the Polity Project and death cases. Model 2 explores the relationship between MPI from the V-Dem Project and death cases. These results confirm the association in Figures 1 and 2 Cepaluni et al. (2020) also reported a statistical analysis in which the political regime is positively correlated with COVID-19 deaths. Their research included confirmed cases as a control variable, which were also positively associated with COVID-19 deaths, as expected. As it is puzzling that the political regime variable is statistically significant, even after controlling for confirmed cases, this leads to questioning of a common explanation for the advantages of authoritarian countries. If authoritarian governments can reduce death cases by stringent intervention, the total confirmed cases must be reduced first; (1) however, the statistical analysis shows a significant effect of political regime on deaths, even after controlling for confirmed cases. It is true that the coefficients are smaller when including confirmed cases as control; however, MPI in model 4 is more significant than in model 3. This indicates that political regime affects COVID-19 death through another factor other than confirmed cases.

Determinants of Non-Pharmaceutical Interventions
Next, this study considers the determinants of non-pharmaceutical interventions (NPIs) -such as lockdowns-by utilizing data from Johns Hopkins University and the University of Oxford (Hale et al., 2020), to test the different levels of intervention by political regimes. If authoritarian governments can reduce COVID-19 deaths by reducing the number of confirmed cases, it must be through stringent interventions in peoples' lives. However, this inference is doubtful based on the above analyses. Whether authoritarian governments more forcefully intervene in peoples' lives than their democratic counterparts, as is often said, was tested.
The dependent variable was NPIs, operationalized by the Stringent Index (Hale et al. 2020). This variable records the daily change of a government's response to the pandemic. The Stringent Index is averaged by each country since the first case was confirmed until December 11, 2020. Ordinary least squares (OLS) was applied for this analysis.

Figure 3: Average Marginal Effects of Confirmed Cases per 1M on SI with 95% CIs
These results suggest that the number of confirmed cases in each country affects government response to COVID-19 as expected, regardless of the political regime.
These models imply that authoritarian governments do not necessarily intervene in civil society more than democratic states. Only when considering the interaction between Polity2 and confirmed cases does the political regime affect government response. However, model 4 and Figure 3 suggest that the tendency is counter to that of previous literature, although this analysis does not consider how swiftly the government responds, which may make a difference (Cepaluni et al., 2020;Cheibub et al., 2020;Frey et al., 2020). In short, strong government measures should not be regarded as an essential factor for authoritarian countries' advantage. What then is the reason behind the advantage of these countries? Next, this article explores the reasons for this. Liang et al. (2020) reported that COVID-19 mortality is negatively correlated with the number of tests in 101 countries. On the other hand, Cepaluni et al. (2020) also included testing as a control variable and reported that it is positively associated with COVID-19 deaths, not negatively. However, their analysis only included 49 countries. The samples in their study are at least 50 countries fewer than those in Liang et al.'s (2020) and this article. The number of tests obtained from Worldometer COVID-19 data is included in the next regression models, considering these contradictory reports.  Liang et al. (2020). GDP per capita is also no longer significant. These results are not affected by the difference in the sample size. Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 mortality rate. Rich countries are likely to have fewer death cases in models 1 and 2 in Table1. However, this advantage is also lost by the inclusion of the number of tests.

Number of tests
Next, the determinants of testing are analyzed.  tests to detect COVID-19 carriers, although Petersen (2020) reported that most authoritarian countries conduct fewer tests. This difference between political regimes may result in an advantage for authoritarian countries. However, this factor implies a different image of the advantages of authoritarian countries. It is often supposed that authoritarian governments intervene in citizens' lives and reduce confirmed cases by limiting their liberty, thus contributing to lower death rates. This study demonstrates another scenario where authoritarian countries cope with COVID-19. Rich countries also tend to conduct more tests than poor countries, contributing to fewer deaths.

Data Manipulation
This study also considers the effects of data manipulation on the advantages of authoritarian countries. It may be possible that some authoritarian governments manipulate death data to overstate their successes, which may affect the results. Some studies suggest this possibility using statistical methods (Adiguzel et al., 2020;Kapoor et al., 2020). This article utilizes the HRV Transparency Index (Hollyer et al. 2014) as an additional control to capture data credibility. The HRV Transparency Project creates this index based on the WDI data. The project regards the missing values in the WDI data as the government's unwillingness to disclose its country's internal affairs. This index can be a proxy for data transparency. Table 5 shows the results considering the index as an additional control. Models 11 and 14 are the results including the index in models 1 and 2 in Table 1. The index in these models is statistically significant. Polity2 is no longer significant in model 11.
This result seems to suggest that the HRV index is more important to COVID-19 deaths than Poliy2. This may imply data manipulation in authoritarian states. However, when including test numbers, the index is no longer significant (models 12 and 15). It is also probable that the sample size of the variables may affect the results. Models 13 and 16, in which the data missing countries on the HRV index and test numbers are dropped from models 11 and 14, confirm this possibility. These results imply that data manipulation may affect authoritarian countries' advantages, but testing should be more crucial.