Differences in NPI strategies against COVID-19

Redlin, Margarete Article — Published Version Differences in NPI strategies against COVID-19 Journal of Regulatory Economics Provided in Cooperation with: Springer Nature Suggested Citation: Redlin, Margarete (2022) : Differences in NPI strategies against COVID-19, Journal of Regulatory Economics, ISSN 1573-0468, Springer US, New York, NY, Vol. 62, Iss. 1, pp. 1-23, https://doi.org/10.1007/s11149-022-09452-9 This Version is available at: https://hdl.handle.net/10419/307046 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Journal of Regulatory Economics (2022) 62:1–23 https://doi.org/10.1007/s11149-022-09452-9 ORIGINAL ARTICLE Differences in NPI strategies against COVID-19 Margarete Redlin1 Accepted: 17 July 2022 / Published online: 22 August 2022 © The Author(s) 2022 Abstract Non-pharmaceutical interventions are an effective strategy to prevent and control COVID-19 transmission in the community. However, the timing and stringency to which these measures have been implemented varied between countries and regions. The differences in stringency can only to a limited extent be explained by the number of infections and the prevailing vaccination strategies. Our study aims to shed more light on the lockdown strategies and to identify the determinants underlying the differences between countries on regional, economic, institutional, and political level. Based on daily panel data for 173 countries and the period from January 2020 to October 2021 we find significant regional differences in lockdown strategies. Further, more prosperous countries implemented milder restrictions but responded more quickly, while poorer countries introduced more stringent measures but had a longer response time. Finally, democratic regimes and stronger manifested institutions alleviated and slowed down the introduction of lockdown measures. Keywords Pandemics ·COVID-19 ·Non-pharmaceutical interventions ·Lockdown · Economics JEL classification I18 ·C23 1 Introduction With the outbreak of the COVID-19 pandemic, many countries began implementing contact restrictions to reduce contacts and thus counteract the spread of the virus. Non-pharmaceutical interventions (NPIs) have been and continue to be used as an important tool against Corona. However, the lockdown strategies pursued are not homogeneous across countries. While some countries attempted to counteract the virus with very strict lockdown strategies and measures such as travel bans, school BMargarete Redlin marg[email protected] 1Department of Economics, Paderborn University, Warburger Str. 100, 33098 Paderborn, Germany 123 2 M. Redlin closures, and curfew restrictions even at quite low incidences, other countries largely refrained from imposing mandatory restrictions and merely issued recommendations for action. Countries in Southeast Asia as well as Australia tried to pursue a zero covid strategy with early border closures, entry barriers, and isolation by imposing a strict lockdown in entire regions even at low incidence levels and trying to eliminate the virus through extensive testing and tracking. Western European countries also show relatively high restrictions. For example, countries such as Germany, France, Italy, and Greece have made vaccination compulsory in certain professions or age levels, and Austria has made it compulsory for the entire adult population. In France, for example, participation in public life is only possible with a health passport, in Austria introduced a lockdown for the unvaccinated, and Germany has been in lockdown several times and the 2G or 2G plus rule applies to participation in public life (2G only vaccinated or recovered, 2G plus with additional test). Sweden, on the other hand-unlike most of its European neighbors-relied more on voluntarism. And so there were, and still are, primarily only recommendations on how to behave, rather than regulations whose disregard would entail consequences or penalties. And the U.S. version was accompanied by regionally different and in part very strict restrictions. However, these were relaxed early on so as not to harm the economy in the long-term. Thus, it is evident that lockdown strategies across countries were not defined by infection incidence alone. Regional, economic, and institutional factors also appear to be important and thus are the focus of this study. Our empirical investigation examines the determinants that played a role in setting the lockdown course and analyze country characteristics associated with strict and less strict lockdown strategies. Based on daily panel data for 173 countries and the period from January 1, 2020 to October 23, 2021, we identify the factors that were driving the stringency of the lockdowns. Using GMM and IV techniques to account for a potential endogeneity between the stringency level of NPIs and the spread of the virus and taking into account the actual development of infection and the respective vaccination coverage, our results show that less developed countries and countries with less established institutions and autocratic regimes have adopted harsher lockdown measures. We also identify significant regional differences in the adoption of NPIs. All in all, our findings offer a fruitful contribution to the debate on determinants of NPIs. The remainder of this paper is organized as follows. Section 2provides an overview of recent studies and forms the hypotheses for the empirical examination. Section 3 presents the empirical model, the data and our result, and Sect. 4concludes. 2 Literature The COVID-19 pandemic has resulted in extraordinary burdens for all countries worldwide. To slow the spread of infection, many countries have implemented nonpharmacological interventions. These lockdown measures were primarily aimed at containing the spread of the virus by reducing contacts in the population. Containment was intended to keep the virus and mortality in check and to protect the respective care and health systems from being overburdened. 123 Differences in NPI strategies against COVID-19 3 In this regard, empirical studies provide evidence that lockdown with decreasing mobility in the population is an effective tool for pandemic control. The relationship between decreasing mobility in the population and the incidence of infection during the pandemic has been clearly demonstrated empirically. A reduction in mobility has been shown to lower the reproductive numbers (Nouvellet et al., 2021). There is also empirical evidence conforming that the overall set of nonpharmacological interventions had the desired effect on the incidence of infection and thus on the mortality rates (Hsiang et al., 2020). Cross country studies show that lockdown is effective in reducing the number of new cases in the countries that implement it compared with those countries that do not (Bo et al., 2021, Alfano & Ercolano, 2020, Banholzer et al., 2020,Hartl2020). Flaxman et al. (2020) model how many infections and deaths were prevented by the non-pharmacological interventions and lockdowns in 11 European countries by May 2020, with the result that more than 3 million lives were saved. And Askitas et al. (2021) analyses worldwide effects of non-pharmaceutical interventions on COVID-19 incidence and population mobility patterns using a multiple-event study confirming that lockdown had significant effects on reducing COVID-19 infections. However, a lockdown has not only desirable effects, but also negative effects and high psychological, social, and economic costs (Bonaccorsi et al., 2020). Thus, the negative side effects and benefits must be weighed when introducing it (Layard et al., 2020). Thus, some countries implemented strict measures only intermittently and only when viral incidence was high, and strategies were not always consistent when infection histories were similar. In general, we would expect the extent of contact restrictions to be higher the more severely the country is affected by the pandemic event, and the measures to be relaxed as incidences decline. This would be reflected in a positive correlation between lockdown and infection rates. Researchers at Oxford University developed the Government Stringency Index (GSI) during the COVID-19 pandemic, which quantifies the severity of lockdowns in states worldwide. It captures all Corona restrictions in place, such as school closures, closed workplaces, travel restrictions, or contact restrictions in a country, and combines them into one index. The correlation coefficient based on data from 173 countries from January 2020 to October 2021 between the GSI and reported new COVID-19 cases is 0.1465. The correlation is relatively low, indicating that only a small proportion of the variance in the Stringency Index can be explained by the prevailing incidence of infection. Figure 1shows the development of the global averages in daily reported COVID-19 cases, the COVID-19 reproduction rate R0, the Government Stringency Index and the vaccination rate in the population. While the global outbreak at the beginning of 2020 led to a dramatic increase in stringency, no joint movement of COVID-19 cases and the Government Stringency Index is visible in the wider context. The development of the reproduction rate also shows no clear correlation with the policy measures. Thus, lockdown strategies cannot be explained by infection rates alone, but other factors also seem to play an important role in governments’ decisions. The determinants of lockdown policies have been considered only sparsely in the literature (Aksoy et al., 2020; De Simone & Mourao, 2021; Ferraresi et al., 2020; 123 4 M. Redlin 0100 200 300 400 01jan2020 01oct2021 time GSI Vaccinations R0 New_cases Fig. 1 Worldwide averages in new COVID-19 cases (per million), COVID-19 reproduction rate (*100) R0, GSI and vaccination rate (%) Frey et al., 2020). De Simone and Mourao (2021) analyze the relationship between country characteristics and lockdown timing and find that urban population and political stability are conducive to a prompt activation of a government’s lockdown policy after initial cases while a country’s wealth and the rule of law may produce an opposite effect and be an obstacle to an immediate policy activation. Aksoy et al. (2020) show that countries with high levels of public attention to COVID-19 are more likely to implement non-pharmaceutical interventions. Analyzing political determinants of lockdown differences in 110 Frey et al. (2020) find that autocratic regimes imposed more stringent lockdowns. They show that in authoritarian countries, a doubling of cases is associated with an increase in stringency 17% higher than in democracies. Ferraresi et al. (2020) provide an event-study design which analyses the determinants of differences in timing and intensity of stringency measures undertaken. By analyzing dummies for economics, political and institutional characteristics, they show the different trajectories of lockdown measures depending on the dummies and find that, for the same number of cases identified countries characterized by low political stability low level of development, low level of digitalization and a high degree of decentralization have adopted less stringent measures. Further, openness and being in an electoral year is associated with a more stringent lockdown policy. Our paper extends these analyses by offering a precise and differentiated review of the factors associated with the implementation of NPIs for reducing COVID-19. To examine the determinants differences in lockdown strategies, a holistic view across countries and over time is needed, including not only infection incidence but also regional and country-specific aspect in terms of development status as well as institutional characteristics. In addition, the inclusion of vaccination coverage is relevant, as it is expected that with higher vaccination coverage, lockdown measures can be relaxed. This study, therefore, uses panel regression analysis with daily observations of the Government Stringency Index, reported new COVID-19 cases, proportion of the 123 Differences in NPI strategies against COVID-19 5 population vaccinated and regional, economic, political, and institutional determinants to analyze which factors have an impact on national lockdown strategies. 3 Empirical evidence 3.1 Estimating model In general, implementation of NPIs can be expected to be related to COVID-19 development. If incidence increases, it can be expected that the government will, on average, introduce harsher restrictions to limit contacts in the population. This effect may be mitigated if a large portion of the population is already vaccinated. Vaccinated individuals generally have high protection against the disease and tend to show mild disease courses even when vaccine breakthroughs occur. Following this reasoning our starting point is a model of the form GSIi,tα+β1casesi,t+β2vaccinationsi,t−1+β3xi,t+μi+i,t where GSIi,trepresents the composite Government Stringency Index in country i at day t, casesi,tis the number of new COVID-19 cases per million people in country i at day t, and vaccinationsi,t, is the percent of people fully vaccinated against COVID-19 in country i at day t, and the disturbance term is composed of the individual effect μi and the stochastic disturbance i,t. We account for regional, economic, political, and institutional differences characteristics by including additional specific explanatory variables that capture these characteristics in xi,t. 3.2 Data Our analysis is based on unbalanced panel dataset of daily data covering 173 countries from the period January 1, 2020 to October 23, 2021. 3.3 Dependent variable Our dependent variable GSI is the COVID-19 Government Stringency Index. It is a composite measure calculated by The Oxford Coronavirus Government Response Tracker (OxCGRT) project based on nine government response indicators.1The nine metrics used to calculate the Stringency Index are: school closures; workplace closures; cancellation of public events; restrictions on public gatherings; closures of public transport; stay-at-home requirements; public information campaigns; restrictions on internal movements; and international travel controls. The index on any given day is calculated as the mean score of the nine metrics, each taking a value between 0 and 100. A higher score indicates a stricter response (i.e., 100 strictest response). 1https://www.bsg.ox.ac.uk/research/research-projects/covid-19-government-response-tracker. 123 6 M. Redlin 3.4 Baseline explanatory variables First, we include the 7-day rolling average of new daily confirmed cases per million people. The data comes from the COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU).2 Additionally, to the level of the infection, we control for the dynamics of COVID-19 by including estimates of the reproduction rate R0 (Arroyo-Marioli et al., 2021). The reproduction rate represents the average number of new infections caused by a single infected individual. Thus, if the rate is greater than 1, the infection is able to spread, while the number of cases will gradually decrease, if the rate is below 1. Further we account for the effect of vaccinations. This variable is defined as the total number of people who received all doses prescribed by the vaccination protocol per 100 people in the total population. The data is provided by the COVID-19 Our World in Data project and is based on public official sources. 3.5 Regional characteristics We control for regional differences by including dummies for continents in the regression on the one hand and running the regressions separately for the individual continents on the other. The dummies represent Africa,Asia,Europe,North America, Oceania, and South America. 3.6 Development First, we investigate the effect of development on the stringency by including GDPpercapita. We use gross domestic product at purchasing power parity (constant 2011 international dollars) from the World Bank World Development Indicators. Second, we include extreme poverty measured as the share of the population living in extreme poverty from the World Bank World Development Indicators.3 Finally, we use the Human Development Index provided by the United Nations Development Programme (UNDP) to account for the effect of development. The HDI is a composite index measuring average achievement in three basic dimensions of human development-a long and healthy life, knowledge, and a decent standard of living. 3.7 Institutional and political characteristics We use the Worldbank’s Worldwide Governance Indicators which measure six broad dimensions of governance to investigate the effect of institutions.4The six dimensions include: 2https://coronavirus.jhu.edu/map.html. 3https://datatopics.worldbank.org/world-development-indicators/. 4https://info.worldbank.org/governance/wgi/. 123 Differences in NPI strategies against COVID-19 7 Table 1 Descriptive statistics Variable Obs Mean Std. dev Min Max GSI 104,655 56.734 20.635 0 100 New cases (per million) 117,314 84.885 165.706 −272.971 3385.473 Reproduction rate R0 100,726 1,002 0.344 −0.030 5.960 Vaccinations (per hundred) 25,445 21.870 22.792 0 118.12 GDP per capita 111,498 19,244.6 20,057.16 661.24 116,935.6 Extreme poverty 74,606 13.499 19.991 0.1 77.6 HDI 111,185 0.726 0.150 0.394 0.957 Voice and accountability 111,738 −0.041 0.987 −2.159 1.725 Political stability 112,281 −0.079 0.977 −2.731 1.913 Government effectiveness 110,885 −0.002 1.004 −2.344 2.335 Regulatory quality 110,885 −0.003 0.993 −2.340 2.206 Rule of law 110,885 −0.025 0.995 −2.346 2.079 Control of corruption 110,885 −0.012 1.013 −1.905 2.270 Voice and Accountability, which captures perceptions of the extent to which a country’s citizens are able to participate in selecting their government, as well as freedom of expression, freedom of association, and a free media. Political Stability and Absence of Violence/Terrorism, a measure of perceptions of the likelihood of political instability and/or politically motivated violence, including terrorism. Government Effectiveness, which captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies. Regulatory Quality, a measure of perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development. Rule of Law, accounting for the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence. And Control of Corruption, which captures perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as "capture" of the state by elites and private interests. The six indicators are reported in their standard normal units, ranging from approximately -2.5 to 2.5. Table 1shows the descriptive statistics of all variable. 3.8 Regression results Table 2presents the results of our baseline specification. The results from ordinary least squares (OLS) estimation and fixed effect (FE) estimation are provided in column 123 8 M. Redlin Table 2 Baseline regression (1) (2) (3) (4) (5) OLS FE GMM Lewbel IV Panel event New cases (per million) 0.0150*** 0.0139*** 0.0132*** 0.0143*** 0.0117*** (0.003) (0.003) (0.003) (0.003) (0.003) Vaccinations (per hundred) −0.2306*** −0.3301*** −0.3624*** −0.3326*** − 0.2682*** (0.034) (0.035) (0.051) (0.036) (0.054) R2 0.1396 0.3485 – 0.3485 0.3751 Hansen j – – 144.11 0.889 3.66 0.160 – AR2 ––−1.22 0.223 –– Instruments ––1692– Countries 173 173 173 173 173 Obs 20,258 20,258 20,258 20,189 20,258 Estimates based on (1) OLS, (2) fixed effects, (3) two-step system GMM, (4) Lewbel instrumental variables regressions and Clarke and Tapia-Schythe (2021) panel event study estimation. R2 denotes the coefficient of determination. Hansen j denotes the Hansen test statistic for overidentifying restrictions. AR2 denotes the Arellano and Bond second order serial correlation test. Dependent variable is the Government Stringency Index. Clustered standard errors in parentheses; for GMM Windmeijer (2005) standard errors *p< 0.10, **p< 0.05, ***p<0.01 (1) and (2). Both coefficients show the expected results. The coefficient for new cases is positive and highly significant. It is thus evident that, on average, the increase in COVID-19 cases is associated with the introduction of harsher NPIs to reduce the contacts and counteract the spread of the virus. The coefficient for the vaccination rate, on the other hand, is negatively significant. This indicates that the lockdown measures could be relaxed as the vaccination rate increased–holding all other factors constant. This result is consistent with Patel et al. (2021) modeling that NPIs and vaccination coverage are both levers that can be used to control spread and showing that with increasing vaccination rate, the restrictions on NPIs can be relaxed. In a second step we provide robustness checks regarding potential endogeneity and the event character of the data. Technically, endogeneity occurs when explanatory variables in a regression model are correlated with the error term. This can occur (1) when important variables are omitted from the model and (2) in case of reverse causality. In our model, the Stringency Index is not only an outcome of the corona cases but can also help to explain the further course of the cases as a predictor. The policy measures serve to reduce the contacts, which in turn have an impact on the spread of the virus. The issue of endogeneity can be addressed by using instrumental variables. This potential endogeneity could bias our OLS and FE results. Therefore, we use the system generalized method of moment (GMM) estimator developed by Blundell and Bond (1998), which relies on a set of “internal” instruments contained within the panel itself. Further we use the instrumental variable (IV) approach developed by 123 Differences in NPI strategies against COVID-19 15 Table 4 Institutional and Political Effects in the Government Stringency Index (1) (2) (3) (4) (5) (6) (7) (8) (9) New cases (per million) 0.0139*** – 0.0129*** 0.0003*** 0.0061*** 0.0004*** 0.0133*** 0.0131*** 0.0128*** (0.005) – (0.002) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Vaccinations (per hundred) −0.3386*** −0.3649*** −0.3635*** −0.5082*** −0.4467*** −0.5001*** −0.3575*** −0.3582*** −0.3581*** (0.043) (0.052) (0.000) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) Reproduction rate R0 9.7230 10.6117 – – – – – – – (7.000) (7.023) – – – – – – – High income country – – −2.8428*** ––– –−1.8100*** 0.5167 – – (0.134) – – – – (0.164) (0.346) ln GDP pc – – – −16.3301*** –– ––– – – – (0.170) – – – – – Extreme poverty – – – – 0.5506*** –––– – – – – (0.035) – – – – HDI –––– –−149.5101*** ––– – – – – – (2.305) – – – Good governance – – – – – – −3.5108*** −2.5101*** −0.6822** – – – – – – (0.182) (0.132) (0.285) 123 16 M. Redlin Table 4 (continued) (1) (2) (3) (4) (5) (6) (7) (8) (9) High income * Good governance – – – – – – – – −3.4041*** –––– –– ––(0.237) Hansen j 159.31 163.21 146.25 140.35 99.87 138.21 −1.22 143.61 142.20 0.523 0.459 0.791 0.878 0.787 0.921 0.222 0.819 0.826 AR2 −0.09 −0.07 −1.23 −1.17 −0.72 −1.18 144.83 −1.23 −1.24 0.931 0.944 0.218 0.241 0.470 0.236 0.815 0.217 0.213 Instruments 165 165 165 165 116 167 165 165 165 Countries 169 169 173 166 116 168 173 173 173 Obs 19,944 19,944 20,258 19,690 14,311 19,988 20,258 20,258 20,258 Estimates based on two-step system GMM regressions. Hansen j denotes the Hansen test statistic for overidentifying restrictions. AR2 denotes the Arellano and Bond second order serial correlation test. Dependent variable is the Government Stringency Index. Robust Windmeijer (2005) standard errors in parentheses *p<0.10,**p< 0.05, 0***p<0.01 123 Differences in NPI strategies against COVID-19 17 show that high quality of institutions can have a detrimental effect on the implementation of strict measures and on response time. Sophisticated institutional processes can slow down implementation, for example through a bloated legal system and high bureaucratic burdens and thus hinder effective execution. Furthermore, government authorities may need to limit certain human rights in these national emergencies when combating the spread of the virus. This is more difficult the more these rights are enshrined in national jurisprudence and laws. Therefore, democratic regimes may find more obstacles to imposing mandatory and harsh measures while autocratic countries face fewer administrative hurdles and can enforce measures more quickly and with less resistance (De Simone & Mourao, 2021; Frey et al., 2020). Since it is often the richer countries that are also institutionally better off, we go on to test the joint effect of the two to see whether it is income, the institutional framework, or the combination of the two that makes more of a difference. Therefore, we introduce models in which we estimate the two dummy variables for good governance and high income together and additionally together with their interaction. While the estimates with both variables still show significant negative coefficients, the results change with the inclusion of the interaction. Looking at the interaction, we see that countries that have both high income and are institutionally well-positioned had lower constraints on average. Good governance by itself still shows a negative effect, but the effect size and significance are now smaller. It is evident that when considering income and institutions jointly, income alone does not have a significant effect on policy measures. Thus, it is more the institutional framework, especially in combination with higher income, that matters. Overall, our results show that differences in lockdown strategies can be explained only to a limited extent by differences in infection numbers and vaccination rates. Rather, we show that regional differences co-determine lockdown strategies. This is in line with Petherick et al. (2020) who show that the development of the number of cases and the response of the state is not parallel. Rather the stringency of policy response has varied substantially, with many coutries experiencing a rise in cases in the summer and fall even as levels of stringency remained approximately constant or fell. Similar results were also found for the US, where Hallas et al (2021) find significant variation in both the measures that states adopt and when they adopt them. Their results shows that after initial peaks in stringency, policy variation by state, region, and political affiliation continued into the fall, with Northeastern and Democrat-led states experiencing more stringent responses overall. In order to better compare the results and their quantitative significance, we present the effects in relation to the standard deviations in Table 5. In addition to the coefficients and standard deviations of the explanatory and dependent variable, the standardized coefficients are presented. The coefficients for COVID-19 cases shows that the change by one standard deviation increases the Government Stringecy Index by 22 units, i.e., about 17% of its standard deviation. In contrast, a one standard deviation change in the vaccination rate has an effect that is three times as large, at 50%. The effects on income and government quality are relatively small. However, the combination of both leads to a 10% change in the standard deviation of stringency. In summary, we show that economic, institutional, and political factors of the country play a significant role in the implementation and harshness of interventions. In 123 18 M. Redlin Table 5 Standardized estimates (1) (2) (3) (4) (6) (7) Coef Std dev X Std dev Y x-stand. coefficient y-stand. coefficient Fully stand. coefficient New cases (per million) 0.0128*** 205.009 15.7197 22.0180 1,228.1016 0.1669 (0.000) – – – – – Vaccinations (per hundred) - 0.3581*** 22.1125 15.7197 −7.9185 −43.8975 −0.5038 (0.002) – – – – – High income country 0.5167 0.3630 15.7197 0.1876 30.4233 0.0119 (0.346) – – – – – Good governance −0.6822** 0.1979 15.7197 −0.1350 −23.0427 −0.0086 (0.285) – – – – – High income * Good governance - 3.4041*** 0.4959 15.7197 −1.6881 −4.6179 −0.1074 (0.237) – – – – – Obs 20,258 – – – – – *p< 0.10, **p< 0.05, ***p<0.01 the future, governments at all levels would benefit from adopting an evidence-based approach to the actions they take. 4 Concluding remarks NPIs are an effective strategy in combating the COVID-19 pandemic. Recent studies show that a suitable combination of NPIs is necessary to curb the spread of the virus (Haug et al., 2020) and that vaccination alone is insufficient to contain the outbreak (Moore et al., 2021). However, the timing and stringency to which these measures have been implemented varied between the countries and regions. The differences in stringency can only be explained to a limited extent by the number of infections and the prevailing vaccination strategies. Our study aims to shed more light on the lockdown strategies and to identify the determinants underlying the differences between countries on regional, economic, institutional, and political level. Based on daily panel data for 173 countries and the period from January 2020 to October 2021, we analyze the factors that were driving the stringency of the lockdowns. We identify significant regional differences. It is evident that some regions and countries were more responsive to global developments of the pandemic, while others adjusted their NPI measures more on country specific virological development. Asian countries introduced relatively strict measures, which were not directly related to domestic infection rates. In North America and Europe, on the other hand, the stringency of the lockdown was comparatively small. An investigation of the relationship between economic development and NPIs shows that the associated high economic lockdown costs in high 123 Differences in NPI strategies against COVID-19 19 developed countries led to a weakened lockdown reaction, while poorer countriesin terms of GDP per capita, poverty level and HDI-have introduced more stringent measures. On the other hand, wealthier countries showed a quicker response, while in poorer countries the response time was longer. Further, democratic regimes and stronger manifested institutions alleviated and slowed down the introduction of lockdown measures. In most political systems and administrative organizations, there was maximum uncertainty about pandemic response and the introduction of NPIs. Our results indicate, that for the future a more structured pandemic policy is needed that provides quick and clear guidelines and recommendations for action. Better conceptual, personnel and material recourses are important prerequisites for fast and effective pandemic response. Rapid and consistent implementation and targeted adaptation of NPIs can both save lives and reduce lockdown duration. Preventive measures, such as the installation of air filters in workplaces and schools and comprehensive testing strategies, may also reduce the need for NPIs and the associated economic and social costs. Overall, our analysis makes a valuable contribution to the discussion of lockdown determinants. However, our estimation is limited by the type of data utilized. Although we control for temporal effects, it should be noted that we do not explicitly control for the prevailing virus variants, since this kind of data is not available in the panel format. It is also important to note, that the study does not take into account the Omicron wave, as only the time period until October 2021 is considered. Funding Open Access funding enabled and organized by Projekt DEAL. Declarations Conflict of interest There is no conflict of interests for this paper. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/ by/4.0/. Appendix: 1 Table A1 Mean values for specific groups 123 20 M. Redlin Variable Overall Africa Asia Europe North America Oceania South America High income (yes) High income (no) Good governance (yes) Good governance (no) GSI 56.734 52.694 60.028 54.071 60.645 46.900 68.583 58.055 55.325 55.976 57.353 New cases (per million) 84.885 26.580 74.692 162.56 86.939 15.839 120.054 124.203 39.594 122.914 50.193 Vaccinations (per hundred) 21.870 6.533 21.983 26.22 25.363 18.705 19.701 25.448 10.420 26.028 23.770 GDP per capita 19244.6 5488.3 23862.9 33478,5 19231.3 14170.3 13885.9 33206.6 5179.0 31393.9 9046.9 Extreme poverty 13.499 33.836 5.667 0.901 5.599 8.359 2.845 1.268 22.363 5.397 19.143 HDI 0.726 0.562 0.742 0.880 0.757 0.731 0.764 0.840 0.610 0.832 0.642 Voice and accountability −0.041 −0.615 −0.667 0.828 0.413 0.799 0.230 0.298 −0.415 0.763 −0.677 Political stability −0.079 −0.691 −0.391 0.549 0.455 0.768 −0.190 0.301 −0.500 0.730 −0.726 Government effectiveness −0.002 −0.772 0.073 0.868 0.101 0.152 −0.272 0.551 −0.603 0.874 −0.684 Regulatory quality −0.003 −0.770 0.006 0.940 0.147 0.078 −0.317 0.517 −0.566 0.853 −0.668 Rule of law −0.025 −0.699 −0.102 0.855 0.076 0.491 −0.390 0.485 −0.579 0.874 −0.725 Control of corruption −0.012 −0.635 −0.164 0.808 0.127 0.590 −0.238 0.484 −0.550 0.907 −0.727 123 Differences in NPI strategies against COVID-19 21 Variable Overall Africa Asia Europe North America Oceania South America High income (yes) High income (no) Good governance (yes) Good governance (no) Polity2 4.291 2.889 0.188 8.750 7.177 3.237 6.969 4.975 3.615 7.825 2.068 Military exp (in GDP) 1.982 1.791 3.070 1.678 1.067 1.464 1.718 2.253 1.710 1.860 2.068 123 22 M. Redlin References Aksoy, C. G., Ganslmeier, M., & Poutvaara, P. (2020). Public attention and policy responses to COVID-19 pandemic. CESifo Working Paper No. 8409, Available at SSRN 3638340. Alfano, V., & Ercolano, S. (2020). The efficacy of lockdown against COVID-19: A cross-country panel analysis. Applied Health Economics and Health Policy, 18, 509–517. Arroyo-Marioli, F., Bullano, F., Kucinskas, S., & Rondón-Moreno, C. (2021). Tracking R of COVID-19: A new real-time estimation using the Kalman filter. PloS one, 16(1), e244474. Askitas, N., Tatsiramos, K., & Verheyden, B. (2021). Estimating worldwide effects of non-pharmaceutical interventions on COVID-19 incidence and population mobility patterns using a multiple-event study. Scientific Reports, 11(1), 1–13. Banholzer, N., van Weenen, E., Kratzwald, B., Seeliger, A., Tschernutter, D., Bottrighi, P., & Feuerriegel, S. (2020). Impact of non-pharmaceutical interventions on documented cases of COVID-19. MedRxiv. Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115–143. Bo, Y., Guo, C., Lin, C., Zeng, Y., Li, H. B., Zhang, Y., & Lao, X. Q. (2021). Effectiveness of nonpharmaceutical interventions on COVID-19 transmission in 190 countries from 23 January to 13 April 2020. International Journal of Infectious Diseases, 102, 247–253. Bonaccorsi, G., Pierri, F., Cinelli, M., Flori, A., Galeazzi, A., Porcelli, F., & Pammolli, F. (2020). Economic and social consequences of human mobility restrictions under COVID-19. Proceedings of the National Academy of Sciences, 117(27), 15530–15535. Clarke, D., & Tapia-Schythe, K. (2021). Implementing the panel event study. The Stata Journal, 21(4), 853–884. De Simone, E., & Mourao, P. R. (2021). What determines governments’ response time to COVID-19? A cross-country inquiry on the measure restricting internal movements. Open Economics, 4(1), 106–117. Ferraresi, M., Kotsogiannis, C., Rizzo, L., & Secomandi, R. (2020). The ‘Great Lockdown’ and its determinants. Economics Letters, 197, 109628. Flaxman, S., Mishra, S., Gandy, A., Unwin, H. J. T., Mellan, T. A., Coupland, H., & Bhatt, S. (2020). Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe. Nature, 584(7820), 257–261. Frey, C. B., Chen, C., & Presidente, G. (2020). Democracy, culture, and contagion: Political regimes and countries responsiveness to Covid-19. Covid Economics, 18, 1–20. Hallas, L., Hatibie, A., Majumdar, S., Pyarali, M., & Hale, T. (2021). Variation in US States’ Responses to COVID-19. Blavatnik Centre for Government Working Paper 2020/034, University of Oxford. Hartl, T., Wälde, K., & Weber, E. (2020). Measuring the impact of the German public shutdown on the spread of Covid-19. Covid Economics, 1, 25–32. Haug, N., Geyrhofer, L., Londei, A., Dervic, E., Desvars-Larrive, A., Loreto, V., & Klimek, P. (2020). Ranking the effectiveness of worldwide COVID-19 government interventions. Nature Human Behaviour, 4(12), 1303–1312. Hsiang, S., Allen, D., Annan-Phan, S., Bell, K., Bolliger, I., Chong, T., & Wu, T. (2020). The effect of large-scale anti-contagion policies on the COVID-19 pandemic. Nature, 584(7820), 262–267. Karabulut, G., Zimmermann, K. F., Bilgin, M. H., & Doker, A. C. (2021). Democracy and COVID-19 outcomes. Economics Letters, 203, 109–840. Kashnitsky, I., & Aburto, J. M. (2020). COVID-19 in unequally ageing European regions. World Development, 136, 105–170. Layard, R., Clark, A., De Neve, J. E., Krekel, C., Fancourt, D., Hey, N., & O’Donnell, G. (2020). When to release the lockdown? A wellbeing framework for analysing costs and benefits. SSRN Electronic Journal. Lewbel, A. (2012). Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models. Journal of Business & Economic Statistics, 30(1), 67–80. Moore, S., Hill, E. M., Tildesley, M. J., Dyson, L., & Keeling, M. J. (2021). Vaccination and nonpharmaceutical interventions for COVID-19: A mathematical modelling study. The Lancet Infectious Diseases, 21(6), 793–802. Nouvellet, P., Bhatia, S., Cori, A., Ainslie, K. E., Baguelin, M., Bhatt, S., & Donnelly, C. A. (2021). Reduction in mobility and COVID-19 transmission. Nature Communications, 12(1), 1–9. 123 Differences in NPI strategies against COVID-19 23 Patel, M. D., Rosenstrom, E., Ivy, J. S., Mayorga, M. E., Keskinocak, P., Boyce, R. M. & Swann, J. L. (2021). The Joint Impact of COVID-19 Vaccination and Non-Pharmaceutical Interventions on Infections, Hospitalizations, and Mortality: An Agent-Based Simulation. MedRXiv. Petherick, A., Kira, B., Hale, T., Phillips, T., Webster, S., Cameron-Blake, E. & Angrist, N. (2020). Variation in government responses to COVID-19. Blavatnik Centre for Government Working Paper 2020/032, University of Oxford. Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step GMM estimators. Journal of Econometrics, 126(1), 25–51. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 123