Globalization and Outbreak of COVID-19: An Empirical Analysis

The purpose of this study is to examine the relationship between globalization, Coronavirus Disease 2019 (COVID-19) cases, and associated deaths in more than 100 countries. Our ordinary least squares multivariate regressions show that countries with higher levels of socio-economic globalization are exposed more to COVID-19 outbreak. Nevertheless, globalization cannot explain cross-country differences in COVID-19 confirmed deaths. The fatalities of coronavirus are mostly explained by cross-country variation in health infrastructures (e.g., share of out of pocket spending on health per capita and the number of hospital beds) and demographic structure (e.g., share of population beyond 65 years old in total population) of countries. Our least squares results are robust to controlling outliers and regional dummies. This finding provides the first empirical insight on the robust determinants of COVID-19 outbreak and its human costs across countries.


Introduction
The spread of the highly contagious coronavirus disease  caused by severe acute respiratory syndrome worldwide has affected 743,201 individuals and has taken the life of 35,000 persons 1 in 192 countries (by 30 March 2020). Yet, the negative impact of the coronavirus outbreak is not limited just to the loss of lives insofar as it has short and long-term socio-economic effects throughout the world.
There are already several reports and studies dealing with the economic consequences of the COVID-19 pandemic in different countries. The coronavirus outbreak has interrupted trade, supply chains and tourism -all of which have had an impact on the global economy (Ahani & Nilashi, 2020). McKibbin and Fernando (2020) demonstrate that, in the short-run, even a controlled outbreak could significantly affect the global economy. Evenett (2020) provides a critical review of the initial trade policy response to  According to the International Monetary Fund Managing Director, COVID-19 outbreak will cause a global recession in 2020 that could be worse than the one triggered by the global financial crisis of 2008-2009. 2 In a recent report, OECD (2020) forecasts that a longer-lasting and more intensive coronavirus outbreak can drop global growth by 1.5%in 2020. So far, it has been estimated that the outbreak will lead to a drop in economic growth in China from 6% to 2% (Khan & Faisal, 2020). Results of Wang et al. (2020)'s study reveal a similar picture where China's expected gross domestic product (GDP) growth rate in 2020 will reduce from 6.50% to 1.72%. Based on different scenarios for the impact of the pandemic on growth, the International Labour Organization (ILO) estimates that the global unemployment could increase by almost 25 million (ILO, 2020). 1 https://ncov2019.live/ (note that data in this website is updated daily. Our analysis is based on data which were available at 30 March) 2 https://www.imf.org/en/News/Articles/2020/03/23/pr2098-imf-managing-director-statementfollowing-a-g20-ministerial-call-on-the-coronavirus-emergency As a result of the COVID-19 pandemic, many countries have banned or imposes restrictions on interpersonal interactions, social, cultural and international trade exchanges 3 .
There is an increasing interest to understand the main explanatory factors of cross-country differences in the pattern of COVID-19 confirmed cases and fatalities.
The pandemic seems to be a major blow to the current form of globalization (Bremmer, 2020), that slows its speed, if does not reverse it (Bloom, 2020), and even may create a new version of globalization which is more regulated (Hutton, 2020). Yet, globalization with the worldwide flow of people, goods, money, information, and ideas, in huge scale and speed, might also be guilty of allowing the speedy spread of the outbreak. Since, for instance, the spread of the COVID-19 disease relies heavily on human-to-human interactions, movement of people internationally could be a dominant driver of its outbreak.
In this paper, we examine whether and to what extent different aspects of globalization are responsible for the outbreak of the COVID-19. In our study, we assess the relationship between different components of globalization, COVID-19 cases, and associated deaths in more than 100 countries. We use multivariate regression analyses, controlling for other plausible factors of COVID-19 outbreak.
There are studies which have examined the negative influence of globalization on health risks (for a review, see Pang et al., 2004 andWoodward et al., 2001). However, our research is the first empirical examination of socio-economic factors (globalization indicators in particular) which may explain, at least partially, the COVID-19 outbreak.
The paper proceeds as follows: Section 2 describes the data and the estimation method; Section 3 presents the findings; and Section 4 concludes.

Data and methodology
We hypothesize that countries with higher levels of globalization are associated with a higher number of COVID-19 cases, ceteris paribus. We, also expect to observe an insignificant relationship between globalization and confirmed deaths of COVID-19, controlling for other explanatory variables such as health infrastructure and demography of countries. To test these hypotheses, we use confirmed cases of COVID-19 and death figures per million by 30 March 2020. The data is regularly updated based on information of local governments' websites/ health departments and can be found at https://ncov2019.live/data.
The base-line econometric model has the following form: Health system capacity: It has been a trending topic around the COVID-19 outbreak (Aleem, 2020). We use log of the number of nurses and midwives (per 1,000 people) and log of the number of hospital beds 5 (per 1000 people), averaged values between 2010 and 2019, as a measure of health system capacity to reduce the negative consequences of COVID-19. We expect to observe a negative correlation between the number of nurses, and hospital beds with death numbers of COVID-19. Modern infrastructures, public health institutions, and efficient medical treatment control the community of infected individuals and keep them far below the critical threshold which is needed for endemic or even epidemic transmission (Murphy, 2006).

Population density:
A higher density of the population may mean more interactions among people and thus a higher risk of contagion. Tarwater and Martin (2001) found a significant effect of population density on the epidemic outbreak of measles or measles-like infectious diseases. We use the average values of population density between 2010 and 2019.
Demographic structure: the Coronavirus infects people, regardless of their age.
However, evidence suggests that the infection rate is likely age-dependent (Suwanprasert, 2020) and older people are at a higher risk of getting severe COVID-19 disease 6 . A higher share of elderly in the population may also mean a higher vulnerability versus COVID-19. Analysis of Zhou et al. (2020) show in-hospital death due to COVID-19 is more likely for patients with older age. Early data from China suggest that a majority of coronavirus disease 2019 deaths have occurred among adults aged more than 60 years and among persons with serious underlying health conditions. 7 Evans and Werker (2020) also argue that uncontrolled virus could have a far lesser death toll in a much younger population. We use an average share of population ages 65 and above in the total population, from 2010 to 2019, and expect it to have a positive correlation with fatalities of COVID-19.
Costs of health care: to control for financial costs of health care for people, we use out-of-pocket expenditure on heath per capita, PPP (current international $) averaged from 2010 to 2019. Out of pocket payments are spending on health directly out of pocket by households in each country. Its higher levels may indicate a higher burden of health care and thus higher vulnerability of individuals against COVID-19. Earlier studies show that ineffective health financing systems and lack of social protection 6 http://www.euro.who.int/en/health-topics/health-emergencies/coronavirus-covid-19/statements/statement-older-people-are-at-highest-risk-from-covid-19,-but-all-must-act-to-preventcommunity-spread 7 https://www.cdc.gov/mmwr/volumes/69/wr/mm6912e2.htm?s_cid=mm6912e2_w#suggestedcitation networks are main drivers of out-of-pocket health expenditure which consequently leads to consumption a large portion of household's budget (e.g., van Doorslaer et al., 2006). Table 1 presents summary statistics of key variables.  Table 2 presents the correlation between COVID-19 confirmed cases and associated deaths by 30 March 2020 (in log) and economic, social, and political KOF globalization indices. As can be seen from Table 2, there is a stronger correlation between social globalization index followed by economic and political dimensions of globalization with COVID-19 outbreak. We will examine the robustness of these correlations through multivariate regression analysis.

Main analyses
We apply the ordinary least squares (OLS) estimation method with robust standard errors. The main variables of interest are different KOF globalization (sub-)indices. All models include GDP per capita (PPP, US$), as a proxy for the relative wealth of nations and economic activities, population density, to account for the higher chance of humanto-human interaction which itself makes infection more likely, ratio of people over 65 years old in the total population, to take into account countries with more high-risk population as well as health system infrastructure proxies such as number of nurses and number hospital beds, both per 1000 population, and out-of-pocket spending on health per capita (PPP, US$), to consider degree of government involvement in the health system and financial burden of health care on people. Moreover, we incorporated regional dummies which control for regional specific characteristics which may also impact the outbreak of COVID-19 such as geography, cultural and behavioral norms and attitudes. (columns 1-8 of Table 3). In addition, in majority of models, the share of older population and the number of hospital beds are positively and negatively correlated with confirmed cases, respectively. Among regional dummies, the Europe and Central Asia dummies have the most effect which is in-line with the fact that Europe is the most globalized region in the world.
In Table 4, we use the log of total confirmed deaths associated with COVID-19 per million. 9 There are some differences in results reported in Table 4 in comparison with estimations for confirmed cases in Table 3. In contrast to Table 3 Table 4).
The size effect of number of nurses per 1000 population is also comparable with the effect of hospital beds (with around 0.5% decreasing impact).
Countries in which people have a higher level of out-of-pocket spending on health suffer more from higher numbers of deaths of COVID-19 pandemic. People in countries with a weaker insurance system coverage and a higher private burden of health costs may visit less frequently doctors and are thus more vulnerable against high-risk diseases. They may not survive due to earlier health deficiencies amplified by COVID-19. The effect is also sizable, as countries in which the out-of-pocket spending on health (per capita) is on average 1% higher, experience approximately 0.70% higher numbers of COVID-19 deaths per million, ceteris paribus (column 1-8 of Table 4) Among regional dummies, the EU and Central Asia dummy still has the most effect on the number of deaths due to the Coronavirus per million of population, while population density (although with expected positive sign) has no statistically significant impact on the number of these deaths.

Sensitivity analysis
To control for the possible effects of outliers in our cross-country estimations, we re-

Conclusion
In our study, we examined cross-country variation in exposure to COVID-19 and associated fatalities in a multivariate regression analysis, covering more than 100 countries. Based on ordinary least squares regressions and several robust estimators for linear regression models which address the possibility of outliers, we find a robust and significant positive association between records of almost all KOF globalization sub-indices with the current level of accumulated COVID-19 confirmed cases, but not with the level of accumulated COVID-19 confirmed deaths. These findings are robust in different models and after control for other possible drivers of the disease including health system infrastructures, demographic structure, and regional dummies. as well as training and employing skilled medical staff (e.g., physicians, and nurses).
In addition, we show that a higher level of out-of-pocket spending on health is explaining part of larger numbers of human costs of COVID-19 pandemic. Therefore, policymakers should improve the efficiency and affordability of access to health care for all individuals and reduce the financial cost of health care on households. Tracing the demographic developments of societies and planning for health needs of the elderlies are also important parts of the resistance package for future pandemics.   Note: MM estimation aims to obtain estimates that have a high breakdown value and more efficient. Breakdown value is a common measure of the proportion of outliers that can be addressed before these observations affect the model. Robust estimators should be resistant to a certain degree of data contamination. MM-estimator has a breakdown point of 50%, e.g., it is resistant to a contamination of up-to 50% of outliers

Data Availability Statement
The datasets generated during and/or analysed during the current study are available from the corresponding author on request.