Research Papers

Taye, A., Borga, L.G., Greiff, S., Vögele, C. and D’Ambrosio, C. (2023). "A machine learning approach to predict self-protecting behaviors during the early wave of the COVID-19 pandemic". Scientific Reports, forthcoming.

Using a unique harmonized real‐time data set from the COME-HERE longitudinal survey that covers five European countries (France, Germany, Italy, Spain, and Sweden) and applying a non-parametric machine learning model, this paper identifies the main individual and macro-level predictors of self-protecting behaviors against the coronavirus disease 2019 (COVID-19) during the first wave of the pandemic. Exploiting the interpretability of a Random Forest algorithm via Shapely values, we find that a higher regional incidence of COVID-19 triggers higher levels of self-protective behavior, as does a stricter government policy response. The level of individual knowledge about the pandemic, confidence in institutions, and population density also ranks high among the factors that predict self-protecting behaviors. We also identify a steep socioeconomic gradient with lower levels of self-protecting behaviors being associated with lower income and poor housing conditions. Among socio-demographic factors, gender, marital status, age, and region of residence are the main determinants of self-protective measures.

Lepinteur, A., Borga, L.G., Clark, A.E., Vögele, C. and D’Ambrosio, C. (2023). "Risk Aversion and COVID-19 Vaccine Hesitancy". Health Economics, forthcoming.

We here investigate the role of risk aversion in COVID-19 vaccine hesitancy. The theoretical effect is ambiguous, as both COVID-19 infection and vaccination side-effects involve probabilistic elements. In large-scale data covering five European countries, we find that vaccine hesitancy falls with risk aversion, so that COVID-19 infection is perceived as involving greater risk than vaccination.

Apouey, B., Yin, R., Etilé, F., Piper, A., and Vögele, C. (2022). "Psychological well-being and the tendency to follow official recommendations against COVID-19: A U-shaped relationship?" University of Luxembourg, mimeo.

Using nationally representative panel data on 7,766 individuals (22,878 observations), we investigate the association between several well-being indicators (depression, anxiety, stress, and loneliness) and the general tendency to follow official recommendations regarding self-protection against COVID-19, in France, Germany, Italy, Spain, and Sweden over the course of four data collection waves. Employing a flexible specification that allows the correlation to be non-monotonic, we find a U-shaped relationship, in which transitions to low and high levels of psychological well-being are associated with higher overall compliance, while transitions to medium levels of psychological well-being are associated with less compliance. Moreover, anxiety, stress, and loneliness levels at baseline also have a U-shaped effect on following the recommendations later (i.e., recommendations are followed best by those with lowest and highest levels of anxiety, stress, and loneliness at baseline, while following the recommendations is lowest for those with moderate levels of these variables). These U shapes are in contrast to previous studies which report monotonic relationships between various measures of mental health and compliance, or ambiguous results. Additionally, we observe a U-shaped correlation between the well-being indicators and a number of specific behaviours (including washing hands and mask wearing). Importantly, most of these specific behaviours play a role in the general tendency to follow recommendations. Finally, we uncover the role of gender composition effects in some of our results. While variations in depression and stress are negatively correlated with variations in overall compliance for males, the association is positive for females. The relation in the full sample (composed of males and females) will reflect first the negative slope for males and then the positive slope for females, explaining the U shape.

Jabakhanji, S., Lepinteur, A., Menta, G., Piper, A., and Vögele, C. (2022). "Sleep quality and the evolution of the COVID-19 Pandemic in five European countries." PLoS ONE, 17, e0278971.

The COVID-19 pandemic has led to lifestyle changes across Europe with a likely impact on sleep quality. This investigation considers sleep quality in relation to the evolution of theCOVID-19 pandemic in five European countries. Using panel regressions and keeping policy responses to COVID-19 constant, we show that an increase in the four-week average dailyCOVID-19 deaths/100,000 inhabitants (our proxy for the evolution of the pandemic)significantly reduced sleep quality in France, Germany, Italy, Spain, and Sweden between April 2020 and June 2021. Our results are robust to a battery of sensitivity tests and are larger for women, parents and young adults. Additionally, we show that about half of the reduction in sleep quality caused by the evolution of the pandemic can be attributed to changes in lifestyles, worsened mental health and negative attitudes toward COVID-19 and its management (lower degree of confidence in government, greater fear of being infected). In contrast, changes in one’s own infection-status from the SARS-CoV-2 virus or sleep duration are not significant mediators of the relationship between COVID-19-related deaths and sleep quality.

Chen, N., Chen, X., Pang, J., Borga, L.G., D’Ambrosio, C., and Vögele, C. (2022). “Measuring COVID-19 Vaccine Hesitancy: Consistency of Social Media with Surveys." University of Luxembourg, mimeo.

We validate whether social media data can be used to complement social surveys to the public’s COVID-19 vaccine hesitancy. Taking advantage of recent artificial intelligence advances, we propose a framework to estimate individuals’ vaccine hesitancy from their social media posts. With 745,661 vaccine-related tweets originating from three Western European countries, we compare vaccine hesitancy levels measured with our framework against that collected from multiple consecutive waves of surveys. We successfully validate that Twitter, one popular social media platform, can be used as a data source to calculate consistent public acceptance of COVID-19 vaccines with surveys at both country and region levels. In addition, this consistency persists over time although it varies among socio-demographic sub-populations. Our findings establish the power of social media in complementing social surveys to capture the continuously changing vaccine hesitancy in a global health crisis similar to the COVID-19 pandemic.

Borga, L.G., Clark, A.E., D’Ambrosio, C., and Lepinteur, A. (2022). "Characteristics associated with COVID-19 vaccine hesitancy." Scientific Reports 12, 12435

Understanding what lies behind actual COVID-19 vaccine hesitancy is fundamental to help policy makers increase vaccination rates and reach herd immunity. We use June 2021 data the COME-HERE survey to explore the predictors of actual vaccine hesitancy in France, Germany, Italy, Luxembourg, Spain and Sweden. We estimate a linear-probability model with a rich set of covariates and address issues of common-method variance. 13% of our sample say they do not plan to be vaccinated. Post-Secondary education, home-ownership, having an underlying health condition, and one standard-deviation higher age or income are all associated with lower vaccine hesitancy of 2–4.5% points. Conservative-leaning political attitudes and a one standard-deviation lower degree of confidence in the government increase this probability by 3 and 6% points respectively. Vaccine hesitancy in Spain and Sweden is significantly lower than in the other countries.

Clark, A.E., D’Ambrosio, C., Lepinteur, A., and Menta, G. (2022). “Pandemic Policy and Individual Income Changes across Europe.” ECINEQ Working Paper Series No. 600

We use data from the COME-HERE panel survey collected by the University of Luxembourg to assess the effects of COVID-19 policy responses on disposable incomes in France, Germany, Italy, Spain and Sweden between January 2020 and October 2021. Policy responses are measured by the Stringency and Economic Support Indices from the Oxford COVID-19 Government-Response Tracker. Controlling for the evolution of the pandemic itself, we find that the income cost of greater stringency measures is borne only by the most economically-vulnerable, while government economic-support measures have a positive effect across the income distribution.

Clark, A.E., and Lepinteur, A. (2022). "Pandemic Policy and Life Satisfaction in Europe." Review of Income and Wealth, 68, 393-408

We use data from the COME-HERE longitudinal survey collected by the University of Luxembourg to assess the effects of the policy responses to the COVID-19 pandemic on life satisfaction in France, Germany, Italy, Spain and Sweden over the course of 2020. Policy responses are measured by the Stringency Index and the Economic Support Index from the Blavatnik School of Government. Stringency is systematically associated with lower life satisfaction, controlling for the intensity of the pandemic itself. This stringency effect is larger for women, those with weak ties to the labour market, and in richer households. The effect of the Economic Support is never statistically different from zero.

Riera Mallol, G., Lepinteur, A, Clark, A.E., Vögele C., and D’Ambrosio, C. (2022). “The Consequences of Economic and Health Shocks on Resilience during COVID-19.” University of Luxembourg, mimeo.

We use the novel COME-HERE panel dataset collected at the University of Luxembourg to estimate the impact of economic and health shocks occurring during the COVID-19 pandemic on individual resilience. We explore both the incidence and the intensity of shocks. We show that resilience is reduced in the wake of unfortunate events. In particular, experiencing a negative shock at least once during the pandemic, such as job loss or diagnosis of a mental disorder, decreases resilience. Experiencing an additional isolation episode or an additional diagnosis of a mental disorder also reduces resilience. There is heterogeneity in our results. Those suffering the worst consequences of a job loss are people with no children, while those with a lower income at the beginning of the pandemic and those with children are the most sensitive to additional isolation periods.

Schifano, S., Clark, A.E., Greiff, S., Vögele, C., and D’Ambrosio, C. (2022). "Well-being and working from home during COVID-19." Information Technology & People. Forthcoming.

We track the well-being of individuals across five European countries during the course of the coronavirus disease 2019 (COVID-19) pandemic and relate their well-being to working from home. We also consider the role of pandemic-policy stringency in affecting well-being in Europe. Using both cross-section and panel regressions, we find that well-being among workers is lower for those who work from home, and those who are not working have the lowest well-being of all. The panel results are more mitigated, with switching into working at home yielding a small drop in anxiety. The panel and cross-section difference could reflect adaptation or the selection of certain types of individuals into working at home. Policy stringency is always negatively correlated with well-being. The well-being penalty from working at home is larger for the older, the better-educated, those with young children and those with more crowded housing.

Clark, A.E., D’Ambrosio, C., and Lepinteur, A. (2021). “The fall in income inequality during COVID-19 in four European countries.” Journal of Economic Inequality 19, 489–507

We here use panel data from the COME-HERE survey to track income inequality during COVID-19 in France, Germany, Italy, and Spain. Relative inequality in equivalent household disposable income among individuals changed in a hump-shaped way between January 2020 and January 2021, with an initial rise from January to May 2020 being more than reversed by September 2020. Absolute inequality also fell over this period. Due to the pandemic some households lost more than others, and government compensation schemes were targeted towards the poorest, implying that on average income differences decreased. Generalized Lorenz domination reveals that these distributive changes reduced welfare in Italy.

Menta, G. (2021). “Poverty in the COVID-19 Era: Real-time Data Analysis on Five European Countries.” Research on Economic Inequality: Poverty, Inequality and Shocks, 29, 209-247

Using real-time data from the University of Luxembourg’s COME-HERE nationally representative panel survey, covering more than 8,000 individuals across France, Germany, Italy, Spain, and Sweden, the author investigates how income distributions and poverty rates have changed from January to September 2020. The author finds that poverty rates increased on average in all countries from January to May and partially recovered in September. The increase in poverty is heterogeneous across countries, with Italy being the most affected and France the least; within countries, COVID-19 contributed to exacerbating poverty differences across regions in Italy and Spain. With a set of poverty measures from the Foster–Greer–Thorbecke family, the author then explores the role of individual characteristics in shaping different poverty profiles across countries. Results suggest that poverty increased disproportionately more for young individuals, women, and respondents who had a job in January 2020 – with different intensities across countries.

Vögele C., Lutz, A., Yin, R., and D’Ambrosio, C. (2020). "How do different confinement measures affect people in Luxembourg, France, Germany, Italy, Spain and Sweden?" First COME‐HERE Report.

COME-HERE is an international longitudinal study designed to understand the consequences of COVID-19 on individuals and societies, and to develop guidelines for policy makers to boost economic performance while at the same time protecting the health and well-being of the population. The survey covers five European countries [France, Germany, Italy, Spain and Sweden]. In each country, the survey follows nationally representative sample of adults (aged 18 or over) in terms of gender, region, and age. Ethics approval was granted by the Ethics Review Panel of the University of Luxembourg. All research was carried out in accordance with relevant guidelines, and informed consent was obtained from all participants. Respondents complete an online questionnaire that takes approximately 25 min. The survey closely assessed people's regular health status, their experiences during COVID-19 (including social environment, financial and health worries, isolation and family environment or degrees of confidence towards authorities) and mental health changes. Results highlight parallels in some of the countries surveyed, but also surprising differences and can be accessed here, while the complete second report is available here.