Tag: COVID19

Media Influence on Behavior During COVID-19: Insights from a Recent Study

20240304 Media Influence on Behavior During COVID19 Image 01

In their paper, recently accepted by Health Economics, Marcel Garz from Jönköping University and Maiting Zhuang from the Stockholm Institute of Transition Economics (SITE) shed light on the impact of media coverage on individual behavior during the Covid-19 pandemic in Sweden.

Media Coverage and Pandemic Behaviour: Evidence from Sweden

This paper explores the intricate relationship between media depictions of COVID-19 and shifts in individuals’ conduct, focusing on Sweden, a standout nation for not imposing lockdowns or curfews during the pandemic. Instead, Sweden relied on voluntary compliance with public health recommendations, making it a crucial case study.

Researchers Marcel Garz and Maiting Zhuang analyzed Swedish newspaper articles about Covid-19 in 2020, totalling 200,000 articles. The study utilized mobility data from Google and employed a robust methodology, including municipality-day panel models and instrumental variable strategies, to ensure accurate results.

The research contributes to the empirical literature by identifying the causal impact of media coverage on individual behavior during a public health crisis.

Key Findings from the Research

The results unveil a significant correlation between media coverage and alterations in behavior patterns. Specifically, mentions of COVID-19 in the media correlated with reduced visits to workplaces and retail and recreation areas, while simultaneously extending the duration of stays in residential locations. Employing two distinct identification strategies, the researchers established a causal link between media coverage and behavioral changes.

Moreover, the study underscores that the impact of media coverage is most pronounced when news stories are locally relevant, visible, and based on facts. Articles referencing crisis managers and providing explicit public health advice were identified as having significant effects on behavior.

These findings carry broad implications for public communication strategies, emphasizing the pivotal role of local media in shaping individual responses to public health crises.

Full Research Paper Access

For a comprehensive understanding of the research background, methodology, data and variables, as well as the empirical strategy and conclusions, kindly refer to the complete paper on Health Economics.

Gender Equality and Women’s Economic Empowerment in Times of Crisis: Insights Shared at the 2023 FROGEE Conference

20240122 Gender Equality Image 01

On October 19-20, 2023, the International School of Economics at Tbilisi State University Policy Institute (ISET Policy Institute), in partnership with the Forum for Research on Gender Economics (FROGEE), organized the conference “Gender Equality and Women’s Economic Empowerment in Times of Crisis”. The conference addressed critical issues surrounding gender equality and women’s economic empowerment. By bringing together academics and practitioners from various sectors it served as a dynamic platform for knowledge sharing and collaboration on actionable solutions and commitments to address multifaceted challenges faced by women globally. This policy brief outlines the keynote, academic and other presentations and discussions featured at the conference.


Gender equality and women’s economic empowerment are vital issues that have gained increasing global attention in recent years. Their significance is even more pronounced in times of crisis, such as during economic downturns or global health emergencies. Such challenging circumstances often exacerbate existing gender disparities and vulnerabilities, making it crucial to address the specific challenges women face in accessing economic opportunities and resources. Discussions on these matters delve into the complex intersection of gender equality and economic empowerment and how empowering women economically can contribute to more resilient and equitable societies.

The October 19-20 conference was aimed at examining and addressing the various aspects of gender equality and female empowerment. The conference begun with opening introductions by Tamar Sulukhia, Eva Atterlöv and Kaori Ishikawa (see the participant list at the end for all associations). Following the opening remarks were two distinctive keynote presentations, a policy panel discussion, and academic presentations. This policy brief summarizes the key takeaways from the conference.

Keynote Addresses

The conference’s first keynote speaker, Elizabeth Brainerd, deliberated on the impact of World War II on marriage and fertility among Russian women. Brainerd show that the war affected these women’s lives for decades, leading to lower rates of marriage and fertility and higher out-of-wedlock births and divorce rates in urban areas than would have been the case in absence of the war. These effects were likely exacerbated by a war and post-war institutional environment that encouraged nonmarital births (in part by expanding the child benefit program) and increased the cost of binding commitments through marriage, particularly for men (absolving fathers of any financial or legal responsibility for children fathered outside marriage). As shown by Brainerd the shock to sex ratios in the Soviet Union due to World War II was among the largest experienced by any country in the twentieth century. In this sense, the effect on Russian women and men was unique and arguably not directly relevant to other countries or time periods. Yet, highly unbalanced sex ratios characterize many populations – whether due to war, immigration and emigration, or preferences for sons etc., – and the analysis can therefore shed light on the effects of sex ratio imbalance also in other contexts. Brainerd’s work supports the conclusion that sex ratios matter for marital and fertility outcomes, both on the marriage market itself and within marriage. The insights from the Soviet Union also highlights that the institutional context matters for determining both the size and direction of the sex ratio’s impact on marriage markets and family formations.

In the conferences second keynote presentation, Maria Floro discussed the findings from a time-allocation survey in Georgia. Evident from the results, women’s work differs from men’s in the sense that women more often perform unpaid household tasks, and since they are primarily responsible for household and caregiving duties, including childcare and elderly care. Such combined responsibilities, coupled with working in typically low-paid jobs can negatively affect women’s physical and mental wellbeing. As the data shows, 66 percent of Georgia’s population engage in unpaid domestic work, with women (88.3 percent) and men (39.6 percent) participating at starkly different rates. Rural women’s participation is the highest, at 90,3 percent. On average, the Georgian population spends 2.1 hours per day on unpaid domestic services for household and family members – with a large gender disparity. In general, the time spent per day by men is 0.7 hours while, in contrast, the time spent by women on these activities is 5 times higher in rural areas (3.6 hours) and 4.7 times higher in urban areas (3.2 hours). Women working full time spend 2.7 hours per day on unpaid domestic services, five times higher than the 0.5 hours spent by men working full time. For all areas of residence, the time spent on unpaid domestic services by women increases with age up until 64 years of age when the numbers drop. Further, women’s time spent on unpaid caregiving work (0.9 hours per day) is 4.5 times higher than the time spent by men. Even for full time working women, the daily time spent on unpaid caregiving work (0.6 hours) is three times higher than that of their male counterparts (0.2 hours). Women who have completed a higher level of education spend higher time on unpaid caregiving services (0.9-1.1 hours per day) than those with a lower level of education (0.4-0.7 hours per day). The difference in women’s and men’s time spent on unpaid caregiving work is greatest for Georgians aged 25-44. Such unequal sharing of household and caregiving responsibilities limits women’s job prospects and is a major reason behind their low participation rate in the labor force, as well as the gender pay gap.

The South Caucasus Gender Equality Index

Following the keynote presentations, Davit Keshelava, presented the ISET Policy Institute’s most recent work on the South Caucasus Gender Equality Index (SCGEI). The index, developed by ISET Policy Institute in close collaboration with Swiss Cooperation Office in Georgia and updated on an annual basis, draws inspiration from the European Institute for Gender Equality’s Gender Equality Index. It comprises of six domains: work, money, knowledge, time, power, and health, alongside eleven subdomains and nineteen indicators.

The index is calculated for three South Caucasus countries, Georgia, Armenia, and Azerbaijan, and nine benchmark countries: Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, and Slovenia. The 2023 edition, mainly based on data from 2021-2022, reveals that within the South Caucasus Armenia is ahead concerning gender equality in the work domain, while Georgia trails behind its regional counterparts. Gender equality in the work domain is lower in the South Caucasus (64.0) than in the baseline countries (67.3).

Georgia stands out as the South Caucasus leader in gender equality within the money domain but significantly trails the baseline countries (South Caucasus – 51.1 vs. baseline countries – 80.5). This discrepancy is the most prominent across all six domains. Azerbaijan leads in the knowledge domain (with Armenia displaying the greatest inequality), yet the South Caucasus slightly outpaces baseline countries in this domain (South Caucasus – 59 and baseline countries – 58.8). This is however the sole equality domain where the South Caucasus surpasses the benchmark countries.

Georgia and Armenia exhibit higher equality in the power domain than Azerbaijan while, in the time domain, Georgia takes the lead in the South Caucasus. In the health domain, Armenia leads in equality, although the difference in index values is marginal.

In the overall index, Georgia emerges as the regional leader in gender equality (60.4), followed by Armenia (57.5) and Azerbaijan (53.0). However, South Caucasus countries as a whole have a lower index (55.4) than the baseline countries (64.1).

Panel Discussion: Topics and Takeaways

The SCGEI presentation was followed by a policy panel discussion, moderated by Tamar Sulukhia and including the panelists Nino Okribelashvili, Nino Chelidze, Nani Bendeliani and Nino Lortkipanidze. The panelists discussed gender inequalities in different areas such as within academia and the tech industry as well as the role of women during crises and the progress made in Georgia towards ensuring gender equality.

Nino Okribelashvili deliberated on the role of women in academia emphasizing that gender inequalities in higher education attainment become obvious when looking at the representation of women across different fields of science. The share of women in subjects such as social work, education and nursing is more than 80 percent, while it is 20 percent in subjects such as computer science, electrical engineering and mechanical engineering. Science, technology, engineering, and mathematics (STEM) oriented institutions are still generally perceived as male dominated. The second glaring gap concerns the representativeness of women in higher rank and leadership positions in academia, where women remain underrepresented in academic and professorial positions across all subjects.

While Nino Okribelashvili discussed the role of women in academia in general, Nino Lortkipanidze focused specifically on the tech industry. She discussed the industry’s potential to create job opportunities for women through various strategies and initiatives such as STEM education and training, diverse hiring practices, leadership development and flexible work policies – including remote work possibilities. Lortkipanidze emphasized that with the right support and opportunities, the rapidly growing tech industry could allow working mothers to thrive in their careers while also enjoying the advantages of a family-friendly work environment.

Shifting the focus to women in times of crisis, Nino Chelidze emphasized the aggravated impact of war on women using the example of the Nagorno-Karabakh conflict. Chelidze highlighted the need for urgent, coordinated action from the donor community to address the challenges of internally displaced persons, most of whom are women and children.

The panel discussion wrapped up with Nani Bendeliani highlighting Georgia’s advancements in gender equality and female empowerment over the past three decades. Bendeliani mentioned different institutional mechanisms adopted in the country for the advancement of women alongside legislative initiatives implemented in different areas concerning for instance maternity and paternity leave, changes to the labor code and the election code. According to Bendeliani, the progress towards gender equality is visible but slow, with available data and multiple assessments showing there is still much to be done.

Academic Presentations

The remainder of the conference was comprised of several academic sessions all contributing to the overall theme of multifaceted gender-related issues. The topics, as detailed below, were: gender disparities in the labor market, violence against women, gender dynamics during the Covid-19 pandemic, the gender divide in education, women in academia and female empowerment and access to services.

Gender Disparities on the Labor Market

The presenters focused on gender disparities on the labor market, exploring aspects such as the implications of labor protection regulations on both men and women, biases and discrimination in employment and wage negotiation, and the impact of female labor force participation on the advancement of women’s rights.

In his presentation, Michal Myck outlined the consequences of labor protection policies in Poland for employees within four years of retirement (regulation that protects them against layoffs, a lowering of their wages or adjustment of their responsibilities). Preliminary results indicate no economically or statistically significant adverse impacts on the employment of men and women approaching labor protection eligibility. These findings suggest that either the anticipated negative effects are absent, or that any concerns employers may have harbored regarding prospective employment protection were counteracted by robust labor demand during the reform period. The general conclusion is that extending protection to specific groups of workers, both men and women, does not necessarily lead to the adverse outcomes often highlighted in standard economic theory.

While Michal Myck focused on labor protection regulations, Francisco Lagos addressed the topic of weight-related employment discrimination and its impact on hiring outcomes. In an experiment, job applications accompanied either by a facial photo of a normal-weight person or by a photo of the same person manipulated to look overweight were sent out to real job opening across 12 occupations in Spain. The results reveal a significant disparity in callback rates for weight-manipulated male applicants, who received fewer callbacks compared to their normal-weight counterparts, with a more pronounced effect in female-dominated occupations. Conversely, weight-manipulated female applicants experienced a slight increase in callbacks, particularly in female-dominated fields. For men, the weight manipulation effect is attributed to the overweight making them appear less attractive, which translates into an attractiveness wage premium. On the contrary the findings for women suggest evidence of an attractiveness penalty, which is also combined with a weight penalty.

The topics of discrimination and biases were also central to Ramon Cobo Reyes Cano’s presentation, which outlined the results of a field experiment on anticipated discrimination and wage negotiation. The findings show that female applicants ask for a lower salary than male applicants in the baseline treatment group – when the full name of the applicant is visible. In the main treatment group, when the gender of the applicant was no longer visible to the employer, the wage requested by female applicants increased by 86 percent, whereas male applicants’ wage requests were 18 percent lower. Evidently, the gender gap in requested wages completely disappears (and even slightly reverses) when the applicants know that their sex is not visible for the potential employer.

The presentations on gender inequalities in the labor market were concluded by Nisar Ahmad, who empirically investigate the impact of women’s labor force participation on women’s rights.  In general, female labor force participation has a positive effect on women’s rights in countries with at least some legal economic rights for women. In countries where women’s rights are extremely limited or non-existent, female labor force participation has a negative or negligible impact on women’s rights.

Violence Against Women

In the academic session devoted to violence against women, the presenters elaborated on the primary factors influencing such violence in various countries at different time periods, including during the Covid-19 pandemic.

Monika Oczkowska explores how social norms, values, and stereotypes determine beliefs about abuse, including recognition of abuse, what is considered as abuse, whether abuse is ever justified, and societal consent towards gender-based discrimination. In countries where gender inequality is rampant, reported rates of abuse in standard surveys are sensitive to the socio-economic status and beliefs about gender norms of the participants, highlighting a high scale of variation in the perception of gender-based discrimination in Central and Eastern Europe.

These findings are in line with the results presented by Salome Gelashvili, who consider potential determinants of gender-based violence (GBV) in South Caucasus. According to the research, key factors contributing to GBV in Armenia, Azerbaijan and Georgia include alcohol abuse, social stigma, being a member of a marginalized groups, a pervasive patriarchal culture, adherence to traditional gender roles, a high level of bureaucracy when reporting GBV to the police, generally weak legal support, limited awareness about various forms of GBV, and economic factors such as financial dependence on an abusive partner.

Similar outcomes, but with more emphasis put on norms and the patriarchal system, were found by Reina Shehi, who assesses gender-based violence in Albania. The results show that the patriarchal system and gender-based norms are the two main factors contributing to gender-based violence. However, there is a growing awareness of the importance of patriarchal institutions and gender norms when addressing GBV in Albania.

Violence against women increase in times of crisis, as shown by Velan Nirmala, who studies women’s empowerment and intimate partner violence (IPV) in India. The findings reveal that, regardless of socio-economic factors, the main types of IPV during the Covid-19 lockdown were physical and emotional violence. The results also highlight that a large majority of victims, regardless of education, wealth, region, household structure, religion, and caste, do not disclose the abuse due to societal taboos.

Gender Dynamics During the Covid-19 Pandemic

The unequal effect from the Covid-19 pandemic was further examined in an academic session in which the presenters keyed in on repercussions of the pandemic on women in terms of employment outcomes, decisions related to time allocation, and the division of unpaid household labor.

Nabamita Dutta presented work on gender inequality in employment during Covid-19 related lockdowns in India. The results show that during the pandemic, women were, in general, 8 percent less likely to be employed than men. While return migrants generally suffered less in terms of finding alternative jobs, being a female return migrant, increased the probability of joblessness to about 17 percent. For female return migrants belonging to marginalized castes, the probability of joblessness was about 10 percent, an interesting result considering that women belonging to marginalized castes (but not being return migrants) experience a higher likelihood of being unemployed then women that are not part of marginalized castes.

Anne Devlin further elaborated on this topic, assessing the economic impact of the Covid-19 pandemic on people living in disadvantaged areas in Ireland. The results indicate that Pandemic Unemployment Payment (PUP) rates were higher in more deprived areas during lockdown periods and that woman, on average, receive PUP for a slightly longer duration than men. Further, female unemployment has a negative and statistically significant relationship with the length of PUP claims. The findings show that average PUP durations tend to be shorter in areas with a higher share of individuals with lower education levels, and in areas with historically higher levels of female unemployment.

Jacklyn Makaaru Arinaitwe presented work on how gender, culture, norms, and practices contributed to the unequal distribution of unpaid care work during Covid-19 in Uganda. The findings reveal that there are policy gaps in addressing the issue, as current policies don’t acknowledge the value of unpaid care work at a personal and national level. This lack of recognition and failure to come up with new ways to reduce or share women’s disproportionate burden of unpaid care work creates obstacles to girls’ education and hinder women’s economic empowerment in Uganda.

Also, on the topic of the Covid-19 pandemic impacts on women, Alessandro Toppeta presented work on the impacts of the pandemic on the role of parental beliefs in England. The results show that parents believe that the time they spend with their children is more valuable and less risky than the time children spend in formal childcare or with friends and that parents’ beliefs can predict the choices they make in investing time with their children. Further, the findings align with previous indications of the increased burden on women’s time experienced during the pandemic being a consequence of limited availability of alternative childcare options.

The Gender Divide in Education

Within the topic of gender in education, the presenters delved into the connection between education and gender roles and the importance of parental education for children’s education.

Sumit S. Deole presented work on the causal impact of education on gender role attitudes based on evidence from European datasets. The results suggest that an additional year of education prompts egalitarian gender role attitudes. Furthermore, the impact of increases in education is particularly prominent among women and, to some extent, in urban areas.

Fethiye Burcu Türkmen-Ceylan focus specifically on the importance of maternal education for children’s education in Turkey. Preliminary results indicate that maternal education has a distinctive positive impact on households’ budget allocation for children’s education among Turkish households.

Saumya Kumar also presented work on the importance of maternal education, considering the impacts of paternal education as well. The presented research finds that both maternal and paternal education reduce the gender gap in educational enrollment. However, having an educated mother is more important when it comes to increasing girls’ enrollment as compared to boys’ enrollment. The research also indicates that as mothers’ education levels rise, there is a greater increase in spendings on education for both boys and girls.

Further on the gender divide within education, Lubna Naz deliberated on how drought affects school attendance in rural Pakistan. The income decline caused by drought leads to a four-month decrease in schooling for all children, and a six-month decrease for boys. Asset ownership also has a negative impact on school attendance, suggesting a possible reverse causality or Simpson’s paradox. The combined effect of asset ownership and drought, however, has a positive impact on school attendance, Naz concluded.

Women in Academia

Gender inequalities are apparent also in the academic sphere. Liis Roosaar’s research looks into the impact of having children on women’s careers within academia. Roosaar find that becoming a mother doesn’t impact earnings per hour, but that mother’s do work fewer hours. More than four years after having a child, women in academia have lost the equivalent of two years of full-time work. Interestingly, men don’t face the same reduction in work hours after becoming fathers. The study also reveals that the career setback for women in academia after having a child is shorter compared to the general population. However, female academics experience a decline in citations as a consequence of the reduced working hours.

Barbara Będowska-Sójka’s research on women in academia focus on female representation on editorial boards of finance journals.  According to Będowska-Sójka women account for 20 percent of all editors on average, with considerable variance between countries. When it comes to editor’s affiliations they are strongly concentrated in the United States, and to a lesser extent in the United Kingdom. Additionally, a small number of extremely well-connected editors sit on many boards. The gender ratio is consistent in substructures for editors that are better connected (have so-called a high degree of centrality in terms of network analysis) or editors who serve on a large number of boards, yet men outnumber women.

Female Empowerment and Access to Services

Although their research focuses on distinct topics, Fazle Rabbi and Ulrich Wohak both presented research on the overarching theme of women’s empowerment and enhanced access to goods and services.

In his paper, Fazle Rabbi and his co-authors consider a new way to support marginalized individuals, most of whom are women, through the introduction of a new donation model where development agencies provide goats to project beneficiaries. Goat ownership might help beneficiaries generate income and devote more time to education. The research results show that the proposed donation model significantly enhances the economic empowerment of participants, providing them a steady income, better access to education, and more access to the financial system – with the results being more pronounced for women.

Ulrich Wohak evaluated tampon tax reforms (efforts to reduce the taxation of menstrual hygiene products, including tampons, pads, and menstrual cups) as a means to address gender-based tax discrimination. Using transaction-level scanner data, the study finds that when countries lower their standard VAT rates, the extent to which these reductions are passed on to consumers ranges from 57 percent to 119 percent.

Concluding Remarks

The ISET conference “Gender Equality and Women’s Economic Empowerment in Time of Crisis” brought together diverse voices, perspectives, and expertise from various sectors to engage in discussions and knowledge sharing on how to advance gender equality in times of normality and in times of crises. The conference also served as a platform to inspire actionable solutions and commitments to address the multifaceted challenges women face worldwide.

List of Participants

  • Alessandro ToppetaAssistant Professor at SOFI, Stockholm University, Sweden. “Parental Beliefs, Perceived Health Risks, and Time Investment in Children: Evidence from COVID-19” (in collaboration with Gabriella Conti and Michele Giannola).
  • Anne DevlinResearch Fellow, Economic and Social Research Institute, Ireland. “The Impact of COVID-19 on Women’s Employment in Ireland” (in collaboration with Adele Whelan, Seamus McGuinnes, Paul Redmond).
  • Aswathi Rebecca AsokPhD Fellow, University of Portsmouth, United Kingdom. “Unveiling Gendered Dimensions of “Volunteerism”: The COVID-19 Story of Kerala, India”.
  • Barbara Będowska-SójkaHead of Department, Poznań University of Economics and Business, Poland. “Editorial boards of finance journals: the gender gap and social networks” (in collaboration with Claudia Tarantola, C., Mare, C., Ozturkkal, B., Paccagnini, A., Perri, R., Pisoni, G., Shala, A., Skaftad´ottir, H., K.).
  • Davit KeshelavaLead Economist, ISET Policy Institute.
  • Elizabeth BrainerdSusan and Barton Winokur Professor of Economics and Women’s, Gender and Sexuality Studies, Brandeis University.
  • Eva AtterlövDeputy Head of Development Cooperation, Embassy of Sweden.
  • Fazle RabbiDeputy Head of School of Business, Crown Institute of Higher Education, Australia. “From Goats to Education: An Innovative Approach to Community Empowerment” (in collaboration with Laurel Jackson and Zahid Hasan).
  • Fethiye Burcu Türkmen-CeylanResearch Fellow, Ahi Evran University, Turkey. “Educate a Woman, And You Educate a Generation: How Does Maternal Education Affect Intro Household Resource Allocation for Education among the Children?” (in collaboration with Ulucan, H., Çakmak, S.).
  • Francisco LagosProfessor of Economics, Georgetown University, USA. “Weight, Attractiveness, and Gender when Hiring: a Field Experiment in Spain” (in collaboration with Catarina Goulão, Juan Antonio Lacomba, and Dan-Olof Rooth).
  • Jacklyn Makaaru ArinaitweDirector, Ace Policy Research Institute, Uganda. “Gender, culture, norms, and practices that promote gender gaps in the allocation of time to unpaid domestic work in the context of COVID-19 in Uganda” (in collaboration with Twinomugisha David).
  • Kaori IshikawaUN Women Country Representative to Georgia.
  • Liis RoosaarLecturer at the Chair of Economic Modelling, University of Tartu, Estonia. “Child penalty in academia: Event study estimate” (in collaboration with Jaan Masso, Jaanika Meriküll, Kärt Rõigas, and Tiiu Paas).
  • Lubna NazAssociate Professor, Institute of Business Administration. Pakistan. “Left High and Dry: Gendered impacts of Drought on school attainment in Rural Pakistan”.
  • Maria FloroProfessor Emerita Economics, American University in Washington, DC.
  • Michal MyckDirector, Centre for Economic Analysis (CenEA), Poland. “Pre-retirement employment protection: no harm when times are good” (in collaboration with Paweł Chrostek, and Krzysztof Karbownik).
  • Monika OczkowskaSenior Research Economist, CenEA, Poland. “Patterns of harassment and violence against women in Central and Eastern Europe. The role of the socio-economic context and gender norms in international comparisons” (in collaboration with Kajetan Trzcinski and Michal Myck).
  • Nabamita DuttaProfessor of Economics, University of Wisconsin-La Crosse, USA. “Lockdown and Rural Joblessness in India: Gender Inequality in Employment?” (in collaboration with Kar, S.).
  • Nani BendelianiProject Analyst, UN Women Georgia.
  • Nino ChelidzeProgram Director of Women’s Initiative for Security and Equity at Mercy Corps.
  • Nino LortkipanidzeWomen in Tech Ambassador for Georgia and Chief Innovation Officer at The Crossroads.
  • Nino OkribelashviliVice Rector for Research at Ivane Javakhishvili Tbilisi State University.
  • Ramon Cobo Reyes CanoProfessor of Economics, Georgetown University, USA. “Anticipated Discrimination and Wage Negotiation: A Field Experiment” (in collaboration with Gary Charness and Simone Meraglia).
  • Reina ShehiPrimary Appointment Lecturer, Epoka University, Albania. “Patterns of Geographic Gender-Based Violence in Albania” (in collaboration with Endi Tirana and Ajsela Toci).
  • Salome GelashviliLead Economist, ISET Policy Institute, Georgia. “Gender-based violence in the South Caucasus” (in collaboration with Lobjanidze, G., Seturidze, E., Shubitidze I.).
  • Saumya KumarAssistant Professor (Economics), University of Delhi, India. “Gender Differential in Parental Investment in Education: A Study of the Factors Determining Children’s and Adolescents’ Educational Investment in India” (in collaboration with Jawaharlal Nehru).
  • Sumit S. DeoleScientific Assistant, Trier University, Germany. “The Causal Impact of Education on Gender Role Attitudes: Evidence from European Datasets” (in collaboration with Zeydanli, T.).
  • Tamar SulukhiaDirector ISET and ISET Policy Institute.
  • Ulrich WohakTeaching and Research Associate, Vienna University of Economics and Business, Austria. Free the Period? Evaluating Tampon Tax Reforms using Transaction-Level Scanner Data (in collaboration with Kinnl, K.).
  • Velan NirmalaProfessor of Economics, Pondicherry University, India. “Women Empowerment and Intimate Partner Violence in India” (in collaboration with Lusome, R).

Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

Employment and Envelope Wages During the Covid-19 Crisis in Latvia

20231113 Envelope Wages Tax Evasion Image 01

The Covid-19 pandemic created one of the most substantial negative exogenous shocks in decades, forcing firms to rapidly adapt. This brief examines an adjustment mechanism that played a significant role in Latvia, and potentially in other countries in Eastern and Central Europe. Specifically, we focus on the role of envelope wages as a buffer for absorbing the shock. Our analysis demonstrates that this form of tax evasion indeed acted as a cushion during the Covid-19 pandemic. Our results indicate that, in the short run, tax-evading firms experienced smaller employment losses in response to the Covid-19 shock compared to compliant firms.


The Covid-19 pandemic generated one of the largest negative, exogenous shocks in decades. To absorb this shock, firms had to swiftly adapt. Prior literature has demonstrated that firms responded by reducing employment and investment (Lastauskas, 2022; Fernández-Cerezo et al., 2023; Buchheim et al., 2020). In this brief, we discuss another margin of adjustment – potentially important for many countries in the region. We focus on the role of envelope wages as a buffer for negative shock absorption.

Envelope wages is a widespread form tax evasion, in which, for employees that are formally registered, a portion of their salary (often at the minimum wage level) is reported to tax authorities, while the remaining ‘envelope’ portion is paid unofficially. The prevalence of this phenomenon has been extensively documented in Eastern and Central Europe (see Kukk and Staehr (2014) and Paulus (2015) for Estonia, Gorodnichenko et al. (2009) for Russia, Putniņš and Sauka (2015) for the Baltic States, Tonin (2011) and Bíró et al. (2022) for Hungary).

In addition to the evident objective of reducing tax obligations, a primary incentive for firms to employ this evasion scheme is the extra flexibility it provides. The unreported portion of wages operates outside of the legal framework, offering firms a means of adaptation in the face of production restrictions, supply chain disruptions, and overall substantial uncertainty caused by the Covid-19 pandemic. In this brief, we argue that firms utilizing envelope wages reduced their employment less than compliant firms during the pandemic in Latvia.

Identifying Firms That Pay Envelope Wages

We identify firms that paid (at least partly) their employees in cash before the pandemic using a rich combination of Latvian administrative and survey data and the methodology proposed by Gavoille and Zasova (2021).

The idea is as follows: We use a subsample of firms for which we can assume that we know whether they pay envelope wages and, using this subsample, train an algorithm that is capable of distinguishing compliant and evading firms based on their observed characteristics and reported financials.

Following Gavoille and Zasova (2021), we use firms owned by Nordic investors as a subsample of tax-compliant firms. To obtain a subsample of non-compliant firms, we combine data on administrative (i.e., reported) wages with several rounds of Labor Force Survey data in order to spot employees who are paid suspiciously little given their personal characteristics (education, experience, etc). Firms employing these employees form the subsample of evading firms. Using these samples of compliant and evading firms, we train a Random Forest algorithm to classify firms according to their type. We then use the algorithm to classify the universe of firms used in this study. Table 1 shows the classification results.

Table 1. Classification results: share of tax-evading firms and employees.

Source: Authors’ calculations.

We find that almost 40 percent of firms (employing about 20 percent of employees) underreport at least some of their workers’ wages. The cross-sectoral heterogeneity is consistent with survey evidence: the construction and transport sectors are the sectors with the highest prevalence of envelope payments. Comparing the share of tax-evading firms with the share of workers working within these firms also indicates that on average, tax-evading firms are smaller than tax-compliant ones. This is yet again in accordance with survey evidence.

Employment Response During Covid-19

Figure 1. Average firm-level change in employment during the Covid-19 pandemic.

Note: This figure shows the average change in employment between January 2020 and any subsequent month, weighted by firm size (average turnover 2017-2019).
Source: Authors’ calculations.

The Covid-19 crisis had a severe impact on Latvia. The government declared a state of emergency as early as March 13, 2020, which entailed significant restrictions on gatherings and on-site work, leading to a six-fold increase in the proportion of remote workers within a matter of months.

During the second wave, in Autumn 2021, Latvia had the highest ranking in the world in terms of new daily positive cases per capita. A substantial number of firms were directly affected by the pandemic (see Figure 1).

We study firm-level employment response at a monthly frequency in compliant and tax-evading firms, from January 2020 to December 2021. Our empirical approach is in the spirit of Machin et al. (2003) and Harasztosi and Lindner (2019), who study the effect of minimum wage shocks. In essence, this approach consists of a series of cross-section regressions, where the dependent variable is the percentage change in employment in a firm between a reference period (set to January 2020) and any subsequent month until December 2021. Our key interest is the difference in cumulative employment response between tax-compliant and evading firms, controlling for a set of (pre-pandemic) firm characteristics, such as the firm’s age, average profitability, average export share, and average labor share over the 2017-2019 period.

The Aggregate Effect

Figure 2 shows the estimated coefficients that measure the difference between employment effects in compliant and tax-evading firms, aggregate for all sectors. Period 0 denotes our reference period, i.e., January 2020, while the estimated coefficients in other periods show the cumulated difference between tax compliant and tax-evading firms in the respective period relative to January 2020 (e.g., the estimated coefficient in period 10 shows the cumulated differential employment response in October 2020 vis-à-vis January 2020).

We document a noticeable difference in the employment response between the two types of firms starting in April 2020. The positive coefficient associated with evading firms indicates that the change in employment growth was not as negative in evading firms as in compliant firms (see Figure 2). Labor tax-evading firms exhibit, on average, a less sensitive employment response than tax-compliant firms. In March 2021, the point estimates are about 0.025, implying that compared to March 2020, tax-evading firms contracted, on average, 2.5 percentage points less than compliant ones. This difference however fades over time and turns insignificant (at the 95 percent level) about halfway through 2021.

Figure 2. Evasion and total employment.

Note: This figure shows the cumulative difference between employment effects in compliant and tax-evading firms, aggregate for all sectors, by month, with respect to January 2020 (reference period).
Source: Authors’ calculations.

Differences by Sector

Figure 3 displays the estimated difference in employment response, disaggregating the sample by sector. We show the results for two sectors: trade and transportation. These two sectors exhibited the most significant differences in employment response between evading and non-evading firms.

For trade, evading firms have been able to maintain employment losses at approximately 5 percentage points less than compliant firms (see Figure 3(a)). This is consistent with the envelope wage margin mechanism. Contrary to the aggregate results, the difference in employment response does not fade over time. This suggests that this margin is not a shock absorber only in the very short run.

The decrease of the evader effect at the aggregate level is caused by negative point estimates of the evasion indicator in the transportation sector, starting in the first quarter of 2021 (see Figure 3(b)). In this sector, evading firms have on average experienced a larger employment decline in 2021 than compliant firms.

Figure 3. Employment effect – by sector.

Note: These figures show the cumulative difference between employment effects in compliant and tax-evading firms, disaggregated by sectors. Source: Authors’ calculations.

The outcome in the transportation sector is likely influenced by the taxi market. There were two major changes in 2021 that particularly affected taxi drivers receiving a portion of their remuneration through envelopes. Firstly, amendments to State Revenues Service’s (SRS) regulations made it more difficult to underreport the number of taxi trips, as each ride was now automatically recorded in the SRS system through taxi apps. Secondly, commencing in July, legal amendments mandated a minimum social security tax, which had to be paid based on at least the minimum wage. Given that many taxi drivers work part-time, and that those associated with evading firms tend to underreport their rides, this new requirement was more binding for evading firms. Additionally, there was a significant shift of taxi drivers to the food delivery sector, where demand for driver services surged during the pandemic.


Our results indicate that employment losses in response to the Covid-19 shock were smaller in tax-evading firms than in compliant firms in the short run. We also demonstrate that by the end of 2021, the discrepancy between the two types of firms had disappeared. This can be explained by significant heterogeneity in employment responses across sectors.

These findings contribute to our understanding of the pandemic’s impact on the size of the informal sector. Despite tax-evading firms generally having more restricted access to finance, the added flexibility provided by unreported wages may have increased their resilience to the negative shock.


This brief is based on a forthcoming working paper COVID-19 Crisis, Employment, and the Envelope Wage Margin. The authors gratefully acknowledge funding from EEA and Norway, grant project “Micro-level responses to socio-economic challenges in face of global uncertainties” (Grant No. S-BMT-21-8 (LT08-2-LMT-K-01-073)).


Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes. 

Media Coverage and Pandemic Behaviour: Evidence from Sweden

20230412 Media Coverage and Pandemic Behaviour Image 01

Individual behaviour can often have wider societal consequences and it is important to understand how to affect positive behavioural change. In this policy brief, we document the ability of the media to increase pro-social behaviour during a public health crisis. In Garz and Zhuang (2022), we collect a unique dataset of 200,000 newspaper articles about the Covid-19 pandemic from Sweden – one of the few countries that did not impose mandatory lockdowns or curfews but largely relied on voluntary social distancing. We show that mentions of Covid-19 significantly lowered the number of visits to workplaces and retail and recreation areas, while increasing the duration of stays in residential locations. The impacts are largest when Covid-19 news stories were more locally relevant, more visible and contained simple and explicit public health advice. These results have wider implications for the design of public communications and the value of the local news media.

The Covid-19 pandemic had devastating health and economic consequences for the entire world (see, for instance, the COVID19 | FREE Network Project for more information on the pandemic experience of the FREE network countries). To stop the spread of the virus, many governments imposed curfews or lockdowns. These mandatory restrictions on people’s movements were and remain controversial. In several countries, these public health policies fuelled conspiracy theories and led to protests and refusals to adopt other protective measures, such as wearing face masks. The policies have also been criticised for placing the burden disproportionately on the poorest members of society, many of whom lost their livelihoods due to the restrictions.

Sweden is one of few countries in the world that did not to impose any lockdowns or curfews. Instead, the Swedish strategy relied on voluntary compliance with public health recommendations aimed at reducing mobility and encouraging social distancing (for FREE policy briefs featuring in-depth discussions of the Swedish pandemic response and experience, see Becker et al., 2020; Hauser, 2020a; Hauser, 2020b; Campa, Roine and Strömberg, 2021; Berlin, 2020.) This strategy was by and large effective, with Sweden seeing similar falls in average mobility in 2020 compared to its Scandinavian neighbours which however imposed strict lockdowns.

What are the drivers of voluntary compliance? A recent SITE working paper “Media Coverage and Pandemic Behaviour: Evidence from Sweden” by Marcel Garz (Jönköping University) and Maiting Zhuang (SITE, Stockholm School of Economics) sheds light on this question. We show that news coverage of the pandemic could have played an important role in shaping public opinion, social norms and ultimately individuals’ health-related actions in Sweden. In this brief, we summarise our approach and findings, and discuss important policy insights.


We collect close to 200,000 newspaper articles about Covid-19 from Swedish newspapers during 2020, which represents almost all newspaper coverage of the pandemic during that year in Sweden. Newspapers remain a major source of information in Sweden, where close to two thirds of the population are regular newspaper readers. We analyse the full texts of articles to identify which aspects of news coverage have the largest impact on behaviour. Figure 1 shows Covid-19 coverage in Swedish newspapers in 2020.

Figure 1. Newspaper coverage of Covid-19

Source: Retriever and authors’ calculations. This graph shows the number of articles including “coronavirus* OR covid*” on a given day, weighted by newspaper circulation. The grey solid lines are local polynomial smooths with 95 percent confidence intervals.

As our main outcome variable, we use Google Community Mobility Report data – anonymous data on the number (or duration) of visits to different types of locations (such as retail and recreation, workplaces, or residential) within a municipality.

Figure 2. Workplace and residential mobility

Source: Google and authors’ calculations. These graphs show the daily percentage change in mobility compared to median values between January 3 to February 6, 2020. The grey solid lines are local polynomial smooths with 95 percent confidence intervals.


We want to study whether reading news concerning Covid-19 affected people’s decisions to reduce their own mobility in Sweden. However, determining the effect of the media on behaviour is difficult for several reasons. First, people choose what media to consume and most often tend towards media outlets that confirm their existing views. Second, there could be factors that determine both media coverage and behaviour.

In our paper, we address the first concern by carefully choosing our data. If we found that people who read more about Covid-19 in newspapers are also more likely to reduce their mobility, this could be driven by many different individual factors, such as how worried a person is about the pandemic. Instead, we measure exposure to Covid-19 news using data on the number of subscriptions every newspaper sells in each municipality. As newspaper subscriptions are annual and were decided before the pandemic, this measure of Covid-19 news exposure should not be correlated with individuals’ beliefs about the pandemic.

The second concern is more relevant, as it is likely that media coverage and individual mobility are both driven by the spread of the pandemic. To alleviate this concern, we adopt several strategies. The first is that we take into account local pandemic severity using a range of different measures, including excess deaths, infections, Covid-19 deaths at different geographic units and with different time lags.

Our second strategy relies on the fact that Swedish newspapers typically circulate in multiple municipalities, but are more likely to respond to events in municipalities where more of their subscribers live. We use an instrumental variable (IV) approach where we use the circulation-weighted excess mortality in a newspaper’s distribution area as the IV. This IV is a strong predictor for the amount of Covid-19 coverage by that newspaper but should not affect mobility in a target municipality – conditional   on the pandemic severity in the municipality itself. We also show that our results hold in a sample of “peripheral” municipalities, that is municipalities that only form a small percentage of any newspaper’s subscriber base, but where these newspapers are nonetheless major sources of information.


We find evidence that media coverage of the pandemic increases compliance with the main public health recommendation to work from home in Sweden. More newspaper articles about Covid-19 on a given day and in a given municipality is associated with increased residential mobility and lower workplace mobility (Table 1), as well as fewer visits to retail areas and recreational facilities. This pattern remains even when controlling for the severity of the pandemic at the local level. The results are robust using different methods and data.

Table 1. Effect of Covid-19 coverage on workplace and residential mobility

Source: Retriever, Google, SMHI, MPRT and authors’ calculations. Observations are at the municipality-day level. The dependent variables are percentage changes in Google mobility for workplaces (columns 1 and 3) and residential areas (columns 2 and 4) relative to the baseline. The main regressor is the circulation-weighted number of Covid-19 articles. Columns 3 and 4 show IV results using the interaction between weighted excess deaths across a newspaper’s circulation area and newspaper exposure as the IV. Local pandemic severity is the latest available excess deaths figure in the municipality at the time of newspaper publication. All regressions control for municipality and day fixed effects. Standard errors in parentheses are clustered by municipality. Significance at the 1%, 5% and 10% is denoted by ***, ** and *, respectively.

The type of coverage matters. Behaviour responds the most to personally relevant news that is easy to understand and visible. In particular, Covid-19 articles which explicitly mention the affected municipality have a larger impact on mobility than, for instance, articles that only relate to developments abroad. There are larger impacts of more factual compared to more subjective reporting. Articles that contain direct and explicit public health advice have a large impact on individual behaviour. In contrast, articles that mention medical experts have a smaller impact on individual behaviour – likely due to the complexity of the language used.

We also find a greater impact on individual behaviour in response to more visible Covid-19 stories, such as articles on the front page or articles whose headlines mention the pandemic. These results are consistent with media coverage not just increasing the salience of the pandemic and reminding individuals to follow official guidelines, but also providing relevant information. Despite fears that the large amount of press coverage could lead to individuals avoiding news about the pandemic, we find little evidence for media fatigue except at very high levels of coverage.


In Garz and Zhuang (2022), we study the effects of media coverage on individual behaviour during a public health crisis. We focus on Sweden, a country that did not impose any lockdowns or curfews during the Covid-19 pandemic and where newspapers remain an important source of information. Using close to the universe of all Swedish newspaper articles about Covid-19 in 2020, we find that media coverage of the pandemic encouraged people to stay at home. The effect is largest when news stories are of local relevance and contain explicit public health advice. These results have important implications for the design of future public communication strategies that aim to foster behavioural change.

We also find little evidence of media fatigue or a preference of opinion pieces relative to factual reporting when it comes to Covid-19 in Sweden. While there has been much discussion about misinformation and media bias during the pandemic, our paper shows a positive effect of the local news media in terms of encouraging voluntary adherence to public health measures.

More broadly, our study adds an important dimension to the policy discussion about the decline of local news, beyond local political accountability and community participation. We find that local news remains an important source of local information, and that personally relevant information is more important for behavioural change. A lack of trusted local media could adversely affect compliance with government recommendations during a crisis, as well as a range of other campaigns, such as those encouraging the take-up of vaccines or adoption of more environmentally friendly behaviours.


Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

What Can We Learn from Regional Patterns of Mortality During the Covid-19 Pandemic?

Doctor outside COVID-19 isolation center representing covid-19 pandemic mortality

Given the nature of the spread of the virus, strong regional patterns in fatal consequences of the Covid-19 pandemic are to be expected. This brief summarizes a detailed examination of the spatial correlation of deaths in the first year of the pandemic in two neighboring countries – Germany and Poland. Among high income European countries, these two seem particularly different in terms of the death toll associated with the pandemic, with many more excess deaths recorded in Poland as compared to Germany. Detailed spatial analysis of deaths at the regional level shows a consistent spatial pattern in deaths officially registered as related to Covid-19 in both countries. For excess deaths, however, we find a strong spatial correlation in Germany but little such evidence in Poland. These findings point towards important failures or neglect in the areas of healthcare and public health in Poland, which resulted in a massive loss of life.


While almost all European countries currently refrain from imposing any Covid-19 related restrictions, the pandemic still takes a huge economic, health and social toll across societies worldwide. The regional variation of incidence and different consequences of the pandemic, observed over time, should be examined to draw lessons for ongoing challenges and future pandemics. This brief outlines a recently published paper by Myck et al. (2023) in which we take a closer look at two neighboring countries, Germany and Poland.  Within the pool of high-income countries, these are particularly different in terms of the death toll associated with the Covid-19 pandemic. In 2020 in Poland, the excess deaths rate (with reference to the 2016-2019 average) was as high as 194 per 100,000 inhabitants, over 3 times higher than the 62 deaths per 100,000 inhabitants in Germany (EUROSTAT, 2022a, 2022b). While, in relative terms, the death toll officially registered as resulting from Covid-19 infections in 2020 was also higher in Poland than in Germany, the difference was considerably lower (about 75 vs 61 deaths per 100,000 inhabitants, respectively) (Ministry of Health, 2022; RKI, 2021). Population-wise Germany is 2.2 times larger than Poland and, before the pandemic struck, the countries differed also in other relevant dimensions related to the socio-demographic structure of the population, healthcare and public health. The nature of Covid-19 and the high degree of regional variation between and within the two countries along some crucial dimensions thus make Germany and Poland an interesting international case for comparison of the pandemic’s consequences. We show that the differences in the spatial pattern of deaths between Germany and Poland may provide valuable insight to the reasons behind the dramatic differences in the aggregate numbers of fatalities (Myck et al., 2023).

Regional Variation in Pandemic-Related Mortality and Pre-Pandemic Characteristics

We examine three measures of mortality in the first year of the Covid-19 pandemic in 401 German and 380 Polish counties (Kreise and powiats, respectively): the officially recorded Covid-19 deaths, the total numbers of excessive deaths (measured as the difference in the number of total deaths in year 2020 and the 2015-2019 average) and the difference between the two measures. Figure 1 shows the regional distribution of these three measures calculated per 1000 county inhabitants. All examined indicators were generally much higher in Poland as compared to Germany. In Poland, deaths officially registered as caused by Covid-19 were concentrated in the central and south-eastern regions (łódzkie and lubelskie voivodeships), while in Germany they were concentrated in the east and the south (Sachsen and Bayern). Excess mortality was predominantly high in German regions with high numbers of Covid-19 deaths, but also in nearby regions. As a result, these same regions also show greater differences between excessive deaths and Covid-19 deaths. On the contrary, high excessive deaths can be noted throughout Poland, including the regions where the number of Covid-19 deaths were lower. In the case of Poland, spatial clusters are much less obvious for both excess deaths and the difference between excess and Covid-19 deaths. To further explore the degree of regional variation between and within countries with respect to the mortality outcomes, we link them to regional characteristics such as population, healthcare and economic conditions, which might be relevant for both the spread of the virus and the risk of death from Covid-19. In Figure 2 we illustrate the scope of regional disparities with examples of (a) age structure of the population, (b) the pattern of economic activity and (c) distribution of healthcare facilities in years prior to the pandemic.

Figure 1. Regional variation of death incidence in 2020: Germany and Poland.

Note: The panels share a common legend based on the quintile distribution of Covid-19 deaths, with two additional categories added at the top and bottom of the scale. County borders in white, regional borders in yellow and country border in grey. Source: Myck et al. (2023).

Figure 2. Pre-pandemic regional variation of socio-economic indicators: Germany and Poland.

Note: Two top and bottom categories in the legend cover 10% of observations each, the rest of categories cover 20% of observations each. County borders in white, regional borders in yellow and country border in grey. Source: Myck et al. (2023).

Shares of older population groups (aged 85+ years) are clearly substantially higher in Germany compared to Poland, and within both countries these shares are higher in the eastern regions. On the other hand, the proportion of labor force employed in agriculture is significantly higher in Poland and heavily concentrated in the eastern parts of the country. In Germany, this share is much lower and more evenly spread. This indicator illustrates that socio-economic conditions in 2020 were still substantially different between the two countries. The share of employed in agriculture is also important from the point of view of pandemic risks – it reflects lower levels of education, and specific working conditions that make it challenging to work remotely yet entail less personal contact and more outdoor labor. The distribution of hospital beds reflects the urban/rural divide in both countries. It is also a good proxy for detailing the differences in the overall quality of healthcare between the two countries, i.e. displaying significantly better healthcare infrastructure in German counties.

Uncovering the Spatial Nature of Excess Deaths in Germany and Poland

While spatial similarities among regions are present along many dimensions, they are particularly important when discussing such phenomena as pandemics, when infection spread affects nearby regions more than distant ones. With regard to the spatial nature of excess deaths in the first year of the pandemic, a natural hypothesis is thus that the pattern of these deaths should reflect the nature of contagion. This applies primarily to excess deaths directly caused by the pandemic (deaths resulting from infection with the virus). At the same time, some indirect consequences of Covid-19 such as limitations on the availability of hospital places and medical procedures, or lack of medical personnel to treat patients not affected by Covid-19, are also expected to be greater in regions with a higher incidence of Covid-19. On the other hand, spatial patterns are much less obvious in cases where excess deaths would result, for example, from externally or self-imposed restrictions such as access to primary health care, reduced contact with other people, diminished family support, or mental health problems due to isolation. While these should also be regarded as indirect consequences of the pandemic, as they would arguably not have realized in its absence, these consequences do not necessarily relate to the actual spread of the virus. Our in-depth analysis of the spatial distribution of the three examined mortality-related measures, therefore, allows us to make a crucial distinction in possible explanations for the dramatic differences in the observed death toll in the first year of the pandemic in Germany and Poland. We explore the degree of spatial correlation in the three mortality outcomes using multivariate spatial autoregressive models, controlling for a number of local characteristics (for more details see Myck et al., 2023).

We find that in Germany, all mortality measures show very strong spatial correlation. In Poland, we also confirm statistically significant spatial correlation of Covid-19 deaths. However, we find no evidence for such spatial pattern either in the total excess deaths or in the difference between excess deaths and Covid-19 deaths. In other words, in Poland, the deaths over and above the official Covid-19 deaths do not reflect the features to be expected during a pandemic. As the results of the spatial analysis show, these findings cannot be explained by the regional pre-pandemic characteristics but require alternative explanations. This suggests that a high proportion of deaths results from a combination of policy deficits and individual reactions to the pandemic in Poland. Firstly, during the pandemic, individuals in Poland may have principally withdrawn from various healthcare interventions as a result of fear of infection. Secondly, those with serious health conditions unrelated to the pandemic may have received insufficient care during the Covid-19 crisis in Poland, and, as a consequence, died prematurely. This may have been a result of lower effectiveness of online medical consultations, excessive limitations to hospital admissions – unjustified from the point of view of the spread of the virus, and/or worsened access to healthcare services as a result of country-wide lockdowns and mobility limitations. The deaths could also have resulted from reduced direct contact with other people (including family and friends as well as care personnel) and mental health deterioration as a consequence of (self)isolation. Our analysis does not allow us to differentiate between these hypotheses, but the aggregate excess deaths data suggests that a combination of the above reasons came at a massive cost in terms of loss of lives. The consequences reflect a very particular type of healthcare policy failure or policy neglect in the first year of the pandemic in Poland.

Our study also shows that a detailed analysis of country differences concerning the consequences of the ongoing pandemic can serve as a platform to set and test hypotheses about the effectiveness of policy responses to better tackle future global health crises.


The authors wish to acknowledge the support of the German Research Foundation (DFG, project no: BR 38.6816-1) and the Polish National Science Centre (NCN, project no: 2018/31/G/HS4/01511) in the joint international Beethoven Classic 3 funding scheme – project AGE-WELL. For the full list of acknowledgements see Myck et al. (2023).


Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

Personality Traits, Remote Work and Productivity

20221024 Personality Traits, Remote Work Image 01

The Covid-19 pandemic generated a massive and sudden shift towards teleworking. Survey evidence suggests that remote work will stick in the post-pandemic period. The effects of remote work on workers’ productivity are however not well understood, some workers gaining in productivity whereas others experience the opposite. How can this large heterogeneity in workers productivity following the switch to teleworking be explained? In this brief, we discuss the importance of personality traits. We document strong links between personality, productivity, and willingness to work from home in the post-pandemic period. Our results suggest that a one-size-fits-all policy regarding remote work is unlikely to maximize firms’ productivity.


The Covid-19 pandemic triggered a large and sudden exogenous shift towards working from home (WFH). Within a few months in Spring 2020, the share of remote workers increased from 8.2 percent to 35.2 percent in the US (Bick et al., 2020), and from 5 percent to more than 30 percent in the EU (Sostero et al., 2020). Surveys of business leaders suggest that WFH will stick in the post-pandemic period (e.g., Bartik et al., 2020).

The prevalence of teleworking will ultimately depend on its impact on workers’ productivity and well-being. This impact however remains ambiguous, some studies reporting an overall positive impact, some studies a negative one. Overall, the balance of these pros and cons can vary greatly across individuals. The existing literature emphasizes the importance of gender and occupation for workers’ productivity under WFH arrangements, but a large share of this heterogeneity remains unexplained.

In a recent paper (Gavoille and Hazans, 2022) we investigate the link between personality traits and workers’ productivity when working from home. Importance of non-cognitive skills, in particular personality traits, for individual labor market outcomes is well documented in the literature (e.g., Heckman et al., 2006; Heckman and Kautz, 2012). In the context of WFH, soft skills such as conscientiousness or emotional stability, are good candidates for explaining heterogeneity in relative productivity at the individual employee level.

The Latvian context provides an ideal setup for studying the effect of teleworking on productivity. First, Latvia has a large but unexploited potential for teleworking. Dingel and Neiman (2021) estimate that 35 percent of Latvian jobs could be done remotely, which is about the EU average. However, prior to the pandemic only 3 percent of the workforce was working remotely – one of the smallest figures in the EU. Second, the Latvian government declared a state of emergency in March 2020, which introduced compulsory WFH for all private and public sector employees, except for cases where on-site work is indispensable due to the nature of the work. This led to a six-fold increase in the share of remote workers within a couple of months. This stringent policy constitutes a massive exogenous shock in the worker-level adoption of WFH, well suited for studying.

Survey Design

To study the link between personality traits, teleworking, and productivity, we designed an original survey, implemented in May and June 2021 in Latvia. The target population was the set of employees who experienced work from home (only or mostly) during the pandemic. To reach this population, we used various channels: national news portals, social media (Facebook and Twitter) and radio advertisement. More than 2000 respondents participated in the survey, from which we obtained more than 1700 fully completed questionnaires.

Productivity and Remote Work

In addition to the standard individual characteristics such as age and the likes, we first collect information about respondents’ perception of their own relative productivity at the office and at home. More specifically, we ask “Where are you more productive?”. The five possible answers are “In office”, “In office (slightly)”, “No difference”, “At home (slightly)” and “At home” (plus a sixth answer: “Difficult to tell”). Table 1 provides a description of the answers. Roughly one third of the respondents reports a higher productivity at home, another third a higher productivity at the office, and one third do not report much of a difference. This measure of productivity is self-assessed, as it is the case with virtually any “Covid-19-era” paper on productivity. Note however that our question is not about absolute productivity as such, but relative productivity of teleworking in comparison with productivity at the office, which is arguably easier to self-assess.

Second, we ask “Talking about the job you worked at mostly remotely, and taking into account all difficulties and advantages, what would you choose post-pandemic: working from home or in office for the same remuneration (if you had the choice)?” The five possible answers are “Only from home”, “Mostly from home”, “Indifferent”, “Mostly in office”, “Only in office” (and a sixth option: “Difficult to tell”). The main aim of this question is to study who would like to keep working remotely in the post-pandemic period, irrespective of productivity concerns. Notably, the answers are much different than from the productivity question (see Table 1), which suggests the latter does not reflect preferences.

Finally, we ask respondents about the post-pandemic monthly wage premium required by the respondent to accept i) working at the office for individuals preferring to work from home; ii) working from home for individuals preferring to work at the office. Median values of these premia for workers with different preferences are reported in Table 1 (panel C). These values appear to be economically meaningful both in absolute terms and relative to the median net monthly wage in Latvia (which was 740 euro in 2021), reinforcing the reliability of the survey.

Table 1. Outcome variables

Source: reproduced from Gavoille and Hazans (2022).

Measuring Personality Traits

The survey contains a section aiming at evaluating the personality of the respondent through the lens of the so-called Five Factor Model of Personality. The psychometrics literature offers several standardized questionnaires allowing to build a measure for each of these five factors – Openness to Experience, Agreeableness, Extraversion, Emotional Stability and Conscientiousness. We rely on the Ten-Item-Personality-Inventory (TIPI) measure (Gosling et al., 2003). This test is composed by only ten questions, making it convenient for surveys, and it has been widely used, including in economics. As simple as this approach seems, the performance of this test has been shown to be only slightly below those with more sophisticated questionnaires, and to provide measures highly correlated with the existing alternative measures of personality traits.


Overall, the results indicate that personality traits do matter for productivity at home vs. at the office. The personality trait most strongly related to all three outcome variables is Conscientiousness. Controlling for a battery of other factors, individuals with a higher level of conscientiousness are reporting a higher productivity when working from home as well as a higher willingness to keep working from home after the pandemic. This link is not only statistically significant but also economically meaningful: an individual with a level of conscientiousness in the 75th percentile is 8.4 percentage points more likely to report a higher productivity from home than a similar individual in the 25th percentile. Considering that the sample average is 31 percent, this difference is substantial.

Previous studies documented a positive correlation between Conscientiousness and key labor market outcomes such as wage, employment status and supervisor evaluation. A usual concern of employers is a possible negative selection of workers in teleworking. Observing that highly conscientious workers are more willing to work from home, where they are more productive, suggests that firms do not need to exert a very strict control on employees choosing to telework.

Openness to Experience shows a similar positive relationship with productivity. Extraversion on the other hand is only weakly negatively related to productivity. The relationship between this trait and willingness to work from home is however much stronger. These findings are intuitive: workers with a high Openness to Experience are more likely to cope easily with the important changes associated with switching to WFH. On the other hand, extravert individuals may find it more difficult to remain physically isolated from colleagues.

The literature studying the relationship between WFH and productivity suggests a conditional effect based on gender. In parallel, the literature investigating the role of personality traits on labor market outcomes also documents gender-specific patterns. As our work builds on these two strands of literature, we provide a heterogeneity analysis of the personality traits/productivity relationship conditional on gender.

When disaggregating the analysis by gender, it appears that the relationship between personality traits and productivity is stronger for women than for men. Conscientiousness and (to a smaller extent) Openness to Experience have a strong positive relationship with relative productivity of teleworking for women, while Extraversion and Agreeableness feature economically meaningful negative relationships. Noteworthy, the effects of Agreeableness and Openness to Experience do not concern the probability to be more productive at the office but only the willingness to work from home after the pandemic. For men, only Conscientiousness is significant, with a much smaller magnitude than for women.


We document that personality traits matter for changes in productivity when switching to a WFH regime. In particular, individuals with high levels of Conscientiousness are much more likely to report a better productivity from home than from the office. Additionally, Openness to Experience and Extraversion also do play a role.

Taken together, these results suggest that a one-size-fits-all policy is unlikely to maximize neither firms’ productivity nor workers’ satisfaction. It also highlights that when estimating firm-level ability in switching to remote work, characteristics of individual workers should be considered. In particular, employers practicing remote work should invest in socialization measures to compensate the negative effect of teleworking on the wellbeing of more extravert workers. Finally, several surveys (e.g., Barrero et al., 2021) document that more than a third of workers in the US would start looking for a new job allowing (some) work from home if their current employer would impose a strict in-office policy. Our results support this finding but also indicate that the opposite also holds: some workers would strongly oppose to remaining in a WFH setup after the pandemic. Personality traits are important determinants of the value attached to working from home.


This research is funded by Iceland, Liechtenstein and Norway through the EEA Grants. Project Title: The Economic Integration of the Nordic-Baltic Region through Labour, Innovation, Investments and Trade (LIFT). Project contract with the Research Council of Lithuania (LMTLT) No is S-BMT-21-7 (LT08-2-LMT-K-01-070).


Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

Social Norms, Conspiracy Theories and Vaccine Scepticism: A Snapshot from the First Year of the Covid-19 Pandemic in Poland

20220419 Social Norms and Vaccine Scepticism Image 05

In January 2022, Poland experienced the highest rate of SARS-CoV-2 transmission since the beginning of the COVID-19 pandemic. Considering the widespread consensus among experts about the efficacy of vaccines in preventing hospitalisation and death resulting from the virus, low vaccination rates and widespread anti-vaccine sentiments in Poland are of great concern. We use data from the DIAGNOZA+ Survey to demonstrate the relationship between various demographic characteristics, opinions around certain gender norms, the propensity for conspiratorial thinking, concern about the pandemic, and vaccine scepticism. While controlling for exogenous demographic characteristics, we measure the strength of the relationship between various beliefs that people hold and how they feel about the COVID-19 vaccine. Our analysis indicates that while respondents who hold more traditional views on gender roles are 6 percentage points less likely to get vaccinated, those who agree with a variety of conspiratorial statements are 43 percentage points less likely to vaccinate against COVID-19.


As of January 2022, Europe finds itself well into the 4th wave of the COVID-19 pandemic, with some countries, including Poland, experiencing the highest rates of transmission since the virus was first detected. There are a few tools available to policymakers and healthcare professionals for combating the spread of the virus, ranging from preventative measures to strict social lockdowns aimed at reducing interpersonal interaction. A comprehensive literature review of 72 academic studies conducted by the BMJ found that the implementation of preventative measures such as hand washing, mask wearing, and social distancing decreased the risk of transmission by 53% (Talic et al., 2021). But even though such measures reduce transmission, the shortcomings in adherence and enforcement make high vaccination rates much more effective in diminishing the risk of hospitalization and death (Moline et al., 2021). With a consensus among experts reaffirming the effectiveness of vaccines in minimising the more severe cases of COVID-19 illness,  the widespread availability of the vaccine has become the most effective and cost-efficient tool in limiting morbidity while avoiding future instances of economically unsustainable lockdowns. The drawbacks of the alternative scenario have already been made evident in 2020, before the development and distribution of COVID-19 vaccines. Over the course of the year, hospital capacities were overwhelmed in many countries around the world, leading to significant spikes in excess deaths. Poland saw an increase of over 18% in all-cause mortality in 2020 (OECD, 2021), the fourth-highest in the OECD and second-highest in the European Union (Eurostat, 2021).

Considering the central role that prevalent vaccination plays in combating the impact of COVID-19, it is important to understand the underlying factors and demographic characteristics of individuals who are driving the low vaccination rates in countries such as Poland. With this in mind, we use an online survey: DIAGNOZA+ (DIAGNOZA Plus, 2020-2021), conducted on a representative sample of adults in Poland throughout the pandemic, allowing for the identification of characteristics that are most strongly correlated with vaccine scepticism. This kind of analysis can provide useful indicators for the targeting of certain policies and information campaigns that encourage vaccinations, and thereby suppress future outbreaks of SARS-CoV-2, as well as any other future pandemics. Below, we first outline the key features of the DIAGNOZA+ data, describe the methodology adopted in this study, and present results on the relationship between key demographic characteristics, social norms, views of respondents, and attitudes towards COVID-19 vaccination. We show a strong correlation between traditional family values, conspiratorial views, and reservations relating to the vaccination programme. Having traditional family values (expressed by about 40% of the sample) is associated with an over 10 percentage point (p.p.) lower probability to declare a willingness to get vaccinated. This drops to about 6 p.p. when we extend the model to account for conspiratorial thinking, which strongly dominates the relationship. Individuals who express strong conspiratorial and anti-establishment views (about a quarter of the sample), conditional on other demographic characteristics, were more than 40 p.p. less likely to declare a willingness to get vaccinated.


The following analysis is based on data from DIAGNOZA+, an online survey collected in seven waves over the course of the COVID-19 pandemic (DIAGNOZA Plus, 2020-2021). The panel survey was conducted with the purpose of assessing changes in the labour market situation of adults in Poland between April 2020 and July 2021. The survey consistently included standard questions on individual and household characteristics such as age, gender and education, as well as questions on as well as questions about the respondent’s labor market status, hours worked, and financial situation. Waves 3 and 4 included additional modules where respondents were asked to express their opinions on a variety of statements surrounding gender norms such as “In general, fathers are as well suited to look after their children as mothers”, “A pre-school child is likely to suffer if his or her mother works” and “When jobs are scarce, men should have more right to a job than women”. The questions were answered on a scale of 1 (strongly agree) to 4 (strongly disagree). For the analysis, these categorical variables are dichotomised, with a value of 1 assigned to responses 1 and 2 (strongly agree or agree) and a value of 0 assigned to responses 3 and 4 (disagree or strongly disagree). Thus, for each question, we develop a binary variable that categorises respondents as either having a progressive or traditional reaction to each particular gender norms statement.

In consecutive waves, the same respondents were asked questions surrounding their willingness to vaccinate against the virus (in wave 5) and their trust in experts and the government response to the COVID-19 pandemic (in wave 6). For this analysis, we select questions that may influence an individual’s likelihood to vaccinate, starting with their level of concern about the pandemic or their fear of the virus itself. Furthermore, we identify individuals with a high predisposition for conspiratorial beliefs based on information from wave 6. Each variable included in this module is converted into a binary measure of agreement or disagreement, as outlined above for the social norms questions. We consider seven statements from the survey related to conspiratorial views, including “Secret organisations influence political decisions” or “I trust my intuition more than the so-called experts” (see the full list of statements in Figure 2). For each of them, the variable is converted into a binary measure of agreement or disagreement, similarly to the social norms questions above. Those who agreed or strongly agreed with all seven statements are classified as having conspiratorial views.

Due to sample attrition and after dropping respondents who did not answer one (or more) of the questions needed for our analysis, the sample reduces to 726 individuals (see table A1 in the Annex). Although each wave of the DIAGNOZA+ survey is carefully weighted to ensure population representativeness of the survey, these cross-sectional weights are only relevant to each independent wave of the survey. Therefore, for our sample, we develop frequency weights by sex and age using population data from Statistics Poland (Statistics Poland, 2021), which are utilised throughout the analysis. Given the low number of participants in the oldest age groups (those above 60 years old), we limit the sample to individuals aged between 21 and 60. Unfortunately, calibrating the weights according to additional characteristics such as education and municipal population is not feasible with a sample of this size. Clearly, the requirement of consistent consecutive participation in at least three waves of the survey has implications for its representativeness. For example, after the sample of respondents that participated in wave 6 is cut to include only those who also participated in waves 3, 4 and 5, we observe a bias in favour of conspiratorial views among the remaining observations, indicating that individuals who hold these views were more likely to continue their participation in the survey. For example, while 18.1% of the total cross-sectional sample of individuals in wave 6 hold conspiratorial views, the proportion is 23.4% in the sample we analyse (falling slightly to 23.2% when weights are applied). From this perspective, while indicative of existing correlations, the results ought to be treated with some caution.

Limiting the sample to respondents who answered all sets of questions across several rounds of the survey allows us to study vaccine scepticism and respondents’ susceptibility to conspiracy theories in relation to a number of personal characteristics. Furthermore, we consider the relationship between a respondent’s attitudes towards certain social norms (asked in waves 3 and 4), their individual response to COVID-19 (asked in wave 5), and their trust in the government’s response to the pandemic (asked in wave 6). We begin the analysis by assessing the relationship between respondents’ demographic characteristics and their opinions on gender roles, their propensity to hold conspiratorial beliefs, and their concern about the pandemic. This is followed by two models measuring respondents’ willingness to vaccinate. In the first of these models, demographic characteristics and traditional family values are used as explanatory variables, while in the second model conspiratorial views are included as well. Finally, we conclude with a summary of results and policy considerations.

Survey Results

Traditional Family Values in Poland

The respondents of the DIAGNOZA+ survey vary, on average, in the ‘traditionality’ of their attitudes towards gender and family depending on the selected indicator. The shares of answers to the three questions about gender norms are presented in Figure 1. The results demonstrate that progressive views on gender norms in Poland were more common in relation to the workplace than the home and family. For example, the statement to which most respondents were opposed was “When jobs are scarce, men have more right to a job than women”, with 37.2% of respondents disagreeing and 50.3% of respondents strongly disagreeing. On the other hand, slightly fewer respondents disagreed (50.5%) or strongly disagreed (34.8%) with “In general, fathers are not as well suited to look after their children as mothers”. Finally, respondents were most ‘traditional’ in their views in reaction to the statement “A pre-school child is likely to suffer if his or her mother works”, with 28% agreeing and 10% strongly agreeing. There is a natural correlation between these different views, and in our analysis, we examine the significance of different combinations of the three indicators. Given the relatively small sample, only the last indicator proved to be significantly related to our main outcome of interest and we use this one to represent the view on the ‘progressive-traditional’ spectrum

 Figure 1. Gender norms in the survey sample

Source: DIAGNOZA+ survey, waves 3 and 4. Note: Data weighted using weights generated from Statistics Poland’s data on population by sex and age. Sample limited to individuals aged 21-60. The statement “In general, fathers are as well suited to look after their children as mothers” from the questionnaire was adjusted in the graph for better readability.

Conspiratorial Views

In wave 6 of the DIAGNOZA+ survey respondents were asked seven different questions relating to trust in government, politicians, media, and the recommendations of experts. As shown in Figure 2, for five out of the seven statements, a majority of respondents agreed or strongly agreed that the government or media are dishonest, intentionally share misinformation, or have ulterior motives. Nearly three quarters of respondents agreed that “politicians and the media deliberately hide certain information”. This result supports data published by the OECD in 2020 showing that, out of the 38 member countries, Poland had the second-lowest trust in government, with only 27.3% of the population expressing confidence (OECD, 2022). However, the DIAGNOZA+ survey goes further to find that nearly half of respondents in our sample reported that they trust their own intuitions more than the experts during the pandemic, while the least widespread belief out of the seven was that “secret organisations influence political decisions”. Still, even this statement, which suggests deep-seeded nefarious behaviour behind the scenes of government, found 39.8% of respondents to be in agreement. Note that we aim to identify individuals who have a general propensity for conspiratorial thinking, rather than those who simply find some of the statements particularly compelling. To this end, we only categorise those respondents who agreed with all seven statements as having a high propensity for conspiratorial thinking, which was the case for 23.2% of our sample after reweighting.

Figure 2. Conspiratorial beliefs and trust in authority

Source: DIAGNOZA+ survey, wave 6. Note: Data weighted using weights generated from Statistics Poland’s data on population by sex and age. Sample limited to individuals aged 21-60.


Table 1 presents regression results on the relationship between specific beliefs reported in the different waves of the survey and a number of individual characteristics. We show these results for three dependent variables: traditional family values, as defined by the opinion that a pre-school child is likely to suffer if his or her mother works; propensity for conspiratorial views, which identifies the respondents that agreed with all seven statements presented in Figure 2; and concern about the pandemic, a binary variable that identifies individuals who expressed great worry or fear about the pandemic. The results indicate that parents who live with their children are 10.1 p.p. more likely to hold traditional family values. After controlling for age, gender and education, living in a small town or village is associated with a 10.9 p.p higher probability of ascribing to more traditional gender norms, while individuals holding a tertiary degree are 18 p.p. less likely to agree that “a pre-school child is likely to suffer if his or her mother works” compared to those with primary education. Interestingly, neither age nor gender significantly correlates with family values, suggesting that the DIAGNOZA+ survey did not capture an intergenerational or gender-driven divide on these issues. This might relate to the online nature of the survey and the implied sample selection, in particular among older individuals.

 Table 1. Regression results on views and attitudes

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. Note: Data weighted using weights generated from Statistics Poland’s data on population by sex and age. Sample limited to individuals aged 21-60. Estimates using the linear probability model.

The results presented in Table 1 also demonstrate a relationship between some demographic characteristics and the likelihood to hold conspiratorial views (as defined by expressing agreement to the seven related statements in wave 6). A number of characteristics strongly correlate with conspiratorial thinking: being a parent living with their children aged 0-17, and living in small cities, towns and villages. Each of these characteristics is associated with a higher probability of believing in secret organisations and mistrusting experts. A number of characteristics strongly correlate with conspiratorial thinking: holding such views are 9.3 p.p. more likely among parents living with their underaged children and 10 p.p. more likely among individuals living in smaller towns or villages compared to those living in cities of over 500 thousand inhabitants. Higher education is strongly negatively correlated with the likelihood of holding conspiratorial views – those with tertiary education are 14.5 p.p. less likely to have these views compared to individuals with primary education.

One simple explanation for the increased vaccination rates among certain demographic groups in Poland could be that some segments of the population are more worried about the virus, and thus choose to take greater precautions. The analysis presented in Table 1 demonstrates that people were increasingly likely to be concerned about the pandemic in higher age groups. When asked “To what extent are you concerned about the COVID-19 pandemic?”, the probability of expressing serious concern increases progressively with age. This is an intuitive result considering the strong relationship between age and the severity of COVID-19 symptoms and the associated risk of mortality (CDC, 2021).  Respondents aged between 31 and 40 were 10 p.p. more likely to report being very concerned or frightened than respondents between the age of 21 and 30, while in the age groups 41-50 (12.6 p.p.) and 51-60 (21.4 p.p.) the probability was even higher. There is also a weak but positive correlation (7.7 and 8.6 p.p.) between living in a city with a population of 10,000 to 500,000 inhabitants and expressing fear about the pandemic, as compared to respondents who lived in cities with a population of more than 500,000 people. The relationships between the remaining demographic characteristics and the probability of being seriously concerned about the pandemic are not statistically significant. Below, we use this data to examine the link between people’s beliefs and the likelihood of getting vaccinated.

Vaccine Scepticism, Demographic Characteristics and Conspiratorial Views

In light of the widespread scientific consensus on the safety and effectiveness of COVID-19 vaccines, low vaccination rates in Poland are difficult to explain. In this section, we analyse to which extent they may be driven by the underlying beliefs, on top of the socio-demographic characteristics. Overall, 54% of respondents in the selected sample from the DIAGNOZA+ survey planned to be or had already been vaccinated. Thus, the survey sample closely reflects the actual proportion of the population that was fully vaccinated in Poland as of January 2022. (ECDC, 2022). In Model A of Table 2, we present the relationship between the response to the question “Do you plan to get vaccinated against COVID-19 or are you already vaccinated?” and traditional family values, alongside the usual demographic characteristics. We find that those in the 51-60 age group were 14.5 p.p. more likely to plan to vaccinate than those aged between 21 and 30. This also reflects the higher level of concern about the virus expressed by those over the age of 50, as presented in Table 1, and the risk of serious illness associated with increasing age. However, the relationship between age and the probability of vaccination was much weaker than the relationship between age and the probability of expressing general concern about the pandemic, implying that concern does not translate directly into a willingness to vaccinate. We also find that tertiary education has a particularly strong effect, and respondents who have a university degree were much more likely (17.7 p.p.) to get vaccinated than those with less than secondary education.

Through this analysis we also discover several less intuitive relationships between individual characteristics and the propensity to vaccinate. We find that women are 11.5 p.p. less likely to plan to vaccinate against COVID-19 than men. Moreover, individuals living in a city with less than 500,000 inhabitants were much less likely to vaccinate, with the strongest correlation (-23.5 p.p.) observed for respondents living in medium-sized cities of 100,000 to 500,000 people. However, a strong relationship can also be seen for smaller cities of 10,000 to 100,000 inhabitants (-19.3 p.p.) and small towns and villages (-17.2 p.p.). Respondents’ expressions of traditional family values are also a strong predictor of their propensity to vaccinate. After controlling for gender, age, education and municipality size, those categorised as holding traditional views are 10.6 p.p. less likely to plan to vaccinate against COVID-19. Our findings demonstrate that while population density, education, age and gender, are all strong indicators of vaccine scepticism in Poland, so is the degree of traditionalism in people’s beliefs.

Table 2. Regression results on vaccination: probability of being vaccinated or planning to get vaccinated

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. Note: Data weighted using weights generated from Statistics Poland’s data on population by sex and age. Sample limited to individuals aged 21-60. Estimates using the linear probability model.

A commonly cited explanatory factor for vaccine scepticism is the susceptibility to conspiratorial beliefs, as well as scepticism towards information disseminated by figures of authority (Hornsey et al., 2018). Thus, in Model B, we seek to identify a relationship between conspiratorial beliefs and scepticism towards the COVID-19 vaccine in Poland. When adding to our model a binary indicator for agreement with all seven of the conspiratorial statements included in the survey, we find that those who agreed across the board were 43.3 p.p. less likely to get vaccinated. Therefore, it seems that the propensity for conspiratorial thinking is a very strong correlate of willingness to vaccinate, and the characteristic most strongly associated with vaccine scepticism. The impact of the demographic factors goes in the same direction for both models, although the scale diminishes in Model B after controlling for conspiratorial views, reflecting the higher propensity of older individuals to hold such views. Furthermore, the effect of traditional family values is much weaker in Model B, suggesting a positive correlation between traditional family values and conspiratorial beliefs (Figure A1 in the Annex shows how values and views in the analysis views overlap with each other). This is in line with past research that ties traditional moral values and conservatism with conspiratorial beliefs, both before and during the COVID-19 pandemic (Pennycook et al., 2020; Romer and Jamieson, 2021).

One explanation for the strong relationship between conspiratorial beliefs and vaccine scepticism could be that respondents who do not trust the media and figures of authority believe that the dangers of the pandemic have been exaggerated and would thus not be concerned about its consequences. We account for this possibility in Model C by including the indicator for fear of the pandemic. We find that those who are very concerned or frightened are 21.1 p.p. more likely to vaccinate than those who are not. However, including this variable in the model has little effect on the estimates of the relationship between traditional gender views or conspiratorial thinking and the likelihood to vaccinate. Further research is needed to understand what is driving these relationships in this particular context. These findings demonstrate that while individuals that believe in conspiracies are the most susceptible to vaccine scepticism, other elements such as fear of the pandemic, education attainment, and where people live play an important role as well.


By January 2022 most European countries have reached a plateau in their vaccination rates, with free vaccines readily available since the summer months of 2021 to all those who are willing to take them. Not only have the high rates of hospital admissions among the non-vaccinated population proven the epidemiological models about the efficacy of vaccines in reducing hospitalisation and death to be true (a study in the United States showed a more than tenfold reduction in the risk of each measure; Scobie et al., 2021), but disparities between countries in the proportion of the population that is vaccinated have created a natural experiment that further substantiates this hypothesis. Poland, a country with a vaccination rate that is 15 p.p. lower than neighbouring Germany, had virtually the same number of cases per 100,000 people in the first two weeks of December, but almost threefold the number of deaths from COVID-19 (ECDC, 2021). Due to the burden COVID-19 related hospitalisations place on healthcare systems, the issues arising from the significant scale of vaccine scepticism are not only related to physical well-being, but also directly impact economic and fiscal stability.

Despite a fairly small sample size available for our analysis from the DIAGNOZA+ survey, a number of important correlations are identified in this study. We find that people living in cities and towns smaller than 500,000 people are less likely to vaccinate than those living in big cities. We show that women, those with less than secondary education, and young people are less likely to be vaccinated. Moreover, those believing that pre-school-aged children suffer when their mothers work are less likely to vaccinate compared to those with more progressive gender views. The most significant predictor of vaccine scepticism, however, is whether a respondent expressed low trust in authority and belief in the conspiracy theories presented in the survey, which was the case for 23.2% of the sample. These individuals are more than 40 p.p. less likely to express willingness to get vaccinated than the rest of the population. This suggests that the low rate of vaccination in Poland can, in part, be attributed to widespread distrust of government, the media, and scientific experts. Poland has already suffered the consequences of the high magnitude of anti-vaccine sentiments in the population, with the severity of the fourth wave of COVID-19 being one of the harshest in Europe (ECDC, 2021). If the government intends to prevent future outbreaks and protect the healthcare system and the economy, it must present a consistent, clear, and transparent message about the safety and efficiency of vaccines to minimise the misinformation that is driving vaccine scepticism among certain demographic groups.


Annex is available in the PDF version.


This Policy Paper was prepared under the FROGEE project, with financial support from the Swedish International Development Cooperation Agency (Sida). FROGEE papers contribute to the discussion of inequalities in Central and Eastern Europe.  For more information, please visit www.freepolicybriefs.com. The views presented in the Policy Paper reflect the opinions of the authors and do not necessarily overlap with the position of the FREE Network or Sida.

Gender Gap Widens During COVID-19: The Case of Georgia

20220411 Gender Gap Widens Image 01

Gender inequality has been a persistent (albeit steadily improving) problem for years. The COVID-induced crisis put women in a disproportionately disadvantaged position, jeopardizing decades of progress achieved towards equality between men and women. However, these effects of the pandemic were not universal across countries. This policy brief aims to evaluate the gender-specific effects of the COVID-19 crisis in Georgia, looking at labor market outcomes and entrepreneurial activities. As expected, the impact of the pandemic was not gender-neutral in this regard, being especially harmful for women. As the Georgian economy rebounds after the crisis, we show that the widened gender gaps are partially offset only in certain aspects. In order to countervail the disproportionate effects of the pandemic, targeted policy measures are needed to stimulate women’s economic activity.


Past economic recessions, including the COVID-induced crisis, have never been gender-neutral (e.g., Liu et al., 2021; Ahmed et al., 2020). While economic crises are usually associated with disproportionate negative impacts on labor market outcomes of men compared to women, the impact of the crisis is, debatably, more severe for women-led businesses as compared to their male-led counterparts (e.g., Torres, 2021; Nordman and Vaillant, 2014; Grimm et al.,2012).

The disproportionate labor market outcomes of economic crises are claimed to be due to the fact that men are predominantly employed in cyclical sectors such as construction or manufacturing; therefore, women have to increase their employment during economic downturns as a means of within-family insurance (Alon et al., 2021). The recent COVID-induced crisis, due to its unique nature, turns out to be an exception in this regard. The pandemic and the subsequently-adopted measures primarily adversely affected contact-intensive sectors (where the worker is required to perform tasks in close physical proximity to other people) that predominantly employ women (Mongey, Pilossoph, and Weinberg 2020; Albanesi and Kim 2021). Moreover, large-scale lockdowns increased the burden of unpaid care, which is generally shouldered by women disproportionately (Babych, 2021), leaving less available time for them to work. It should be noted that gender gaps in the labor market were a persistent (albeit steadily improving) problem even before the pandemic (Eurofound, 2016). Therefore, COVID-19 poses a threat jeopardizing the progress achieved in this direction and worsening gender inequality.

COVID-19 brought unprecedented adverse consequences for not only employed workers but entrepreneurs as well. Increased unpaid care and housework pose additional burdens on female top managers, making women-led businesses more vulnerable to the crisis.

The unequal gender implications of the COVID-19 crisis have been widely debated. Growing evidence (Albanesi and Kim 2021; Torres et al., 2021; Alon et al., 2020; Caselli et al., 2020, Fabrizio et al., 2021) attests that, on average, the effects of the pandemic put women in a disproportionately disadvantaged economic position. However, the extent of this effect varies across countries and is absent in some cases (Campa et al., 2021; Torres et al., 2021).

This policy brief aims to examine the gender-specific nature of the COVID-19 crisis in Georgia. With this aim, we study the differential effects of the pandemic on the economic activity of women in terms of labor market outcomes and entrepreneurship. First, we contrast labor market outcomes for Georgian men and women during the COVID-19 crisis. Secondly, we try to assess the magnitude of the disproportionate impact on women-led businesses compared to men-led ones. We calculate gender gaps across different measures of firm-level performance, such as sales revenue, liquidity and owners’ expectations of falling into arrears. Finally, we examine whether there are any signs of recovery yet in 2021 and draw policymakers’ attention to emerging issues.

Labor market highlights

The adverse effects of the pandemic on female employment were conditioned by both supply and demand-side factors. The latter include decreased economic activity, mainly in service-related sectors (hospitality, personal care, etc.) that are dominated by women (Eurofound, 2021). In Georgia, as of 2019, women constituted the majority of workers in sectors such as hospitality (56%), education (83%) and activities of households as employers of domestic personnel (99%) that experienced some of the sharpest declines in employment during 2020. Moreover, women are more likely to be employed in part-time and temporary jobs (14% of women, as opposed to 11% of men, were employed part-time as of 2019, Geostat Labor Force Survey 2019), leaving them more vulnerable during times of crisis.  Supply-side factors were triggered by the unequal burden of unpaid work generally undertaken by women in Georgia, mainly due to cultural reasons as well as the higher opportunity cost of time for men (women in Georgia on average earned 64% of men’s salaries in 2019, Geostat). School and daycare closures and decreased childcare involvement of grandparents increased household responsibilities for women. A UN Women survey-based study showed that in the midst of the pandemic in Georgia, around 42% of women reported spending more time on at least one extra domestic task as opposed to 35% of men (UN Women, 2020). This would naturally lead to more women than men leaving the labor force. Indeed, looking at the data, we see that in one year after the COVID-19 outbreak, women contributed to 98% (48,000 individuals) of the decrease in the Georgian labor force in 2020 (Geostat). Moreover, a close look at the percentage point difference between the labor force participation rates of Georgian men and women reveals a notable growth in the gender gap starting from 2020. The same can be said about employment rates (Figure 2).

Figure 1. Difference between male and female labor force participation and employment rates

Source: Geostat

To further elaborate on the tendencies in employment, Bluedorn et al. (2021) look at the differences between employment rate changes among male and female workers in 38 advanced and emerging economies. Replicating the exercise with the Georgian data, we can observe results similar to those obtained in Bluedorn et al. (2021). In Figure 2, we see differences between female and male employment rate changes. For each gender group, the latter is computed as an absolute difference between the quarterly employment rate and its annual average level from the previous year. Once the difference takes a negative value, implying that the drop in employment was sharper for women, one could say that we observe a “She-cession” phenomenon as termed by Bluedorn et al. (2021). As we can see, in 2020, the employment rate of women fell more than that of men. This widened gender gap was partially offset in 2021.

Figure 2. Employment rate changes by gender (deviation from the previous year average)

Source: Geostat

Remote work: a burden or a blessing for women?

One important aspect of the COVID-19 crisis was a wide-scale switch to remote work. This development had some gender-specific implications as well. The evidence shows that the prevalence of the switch to remote work was higher among women compared to men (41% vs. 37%) in the EU (Sostero et al., 2020). This tendency also holds in Georgia, where 11% of women as opposed to only 3% of men reported usually working from home in the last three quarters of 2020 (Julakidze and Kardava, 2021). It is not clear whether this tendency can be explained by gender-related occupational differences of male and female jobs (Dingel and Neiman, 2020; Boeri and Paccagnella, 2020; Sostero et al., 2020) or, rather, different personal choices of men and women working in the same occupations. Interestingly, across different countries, we observe a positive correlation between gender inequality (as measured by the Gender Inequality Index) and gender differences in the switch to remote work (measured by the ratio of the share of remote workers among female and male workers). To account for this observation, we can stipulate that gender differences in switching to remote work might be explained by differing gender roles in households, and in society at large, across countries (as proxied by the gender inequality index).

Figure 3. Relative prevalence of remote work among female and male workers

Source: Eurostat, Statistics Sweden, Statista, Geostat, UNDP Human Development Reports

Regardless of the reason, remote work is likely to have some important implications on gender roles. However, the directionality of these implications is not straightforward. On the one hand, remote work offers flexibility for women to juggle household and work responsibilities. On the other hand, since women compared to men have been shown to be more likely to use the time saved from commuting to engage in housework, the switch to remote work might increase their “total responsibility burden” (Ransome, 2007) and lead to time poverty (Peters et al., 2004; Hilbrecht, Shaw, Johnson and Andrey, 2008). Indeed, according to CARE International South Caucasus (2020), around 48% of female survey participants in Georgia placed additional effort into housework and childcare in the midst of the pandemic. Moreover, as women are more likely and expected to use remote working as a means of balancing work-life responsibilities (Moran and Koslowski, 2019) their bargaining power at work decreases relative to their male counterparts. This could have some adverse career implications for female workers. Recent enforced lockdowns might pose an opportunity in this regard, as once-remote work becomes something close to a “new normal” employers will likely decrease the penalty for remote workers.

Spotlight on women-led business performance during the COVID-19 crisis

Calamities brought by the pandemic worsened financial outcomes for enterprises, affecting their ability to operate and have stable financial income. Similar to other crises, the pandemic has not been gender-neutral (Liu et al., 2021; Ahmed et al., 2020) in terms of the effect on business performance.

Gaps in the performance of women- and men-led businesses have been prevalent beyond any economic crisis as well, and have been documented in a number of studies (e.g., Amin, 2011; Bardasi et al., 2011), registering gender differences in sales and productivity in favor of men-owned enterprises. As suggested by Campos et al. (2019), these performance gaps may be due to lower levels of capital owned by women as opposed to men, a smaller number of employees hired by women-owned firms, as well as different practices in using advanced business tools and innovation. In addition, the existence of these gender gaps has also been explained as stemming from the prevailing social norms that assign certain obligations to women. Nordman and Vaillant (2014) and Grimm et al. (2012) suggest that unpaid housework and family-care led to a constrained number of hours women could afford to spend on the work and management of firms, negatively affecting their productivity.

According to the Women Entrepreneurship Report (Global Entrepreneurship Monitor (GEM), 2021), the pandemic imposed an additional burden in terms of increasing family-care duties on women. The GEM survey (2021) conducted in 43 countries worldwide shows that the likelihood of enterprise closure is 20% higher for women-led compared to men-led businesses. The higher likelihood of closure reflects the adverse factors that may have hindered the operating capacity of firms. For example, a survey conducted by UNIDO (2020) suggests that, as a result of the Coronavirus crisis, African and Middle Eastern women-led firms experienced diminished revenues. In addition, 41% of women-led firms were short of cash flow and unable to fulfill financial obligations, while only 32% of male entrepreneurs were exposed to the same problem.

More rigorous analysis on this matter has been conducted by Torres et al. (2021) and Liu et al. (2021). They try to examine the asymmetric effects of the COVID-19 crisis on women-led firms in several dimensions utilizing new datasets from the World Bank: COVID-19 Follow-up Enterprise Survey and the World Bank Business Pulse Survey. The findings of Liu et al. (2021) for 24 countries from Central Europe & Central Asia and Sub-Saharan Africa confirm that during the pandemic women-led businesses are subject to a higher likelihood of closure than men-led businesses and that female top managers are more pessimistic about the future than their male counterparts. Finance and labor factors were mentioned to be the major contributors to these disadvantages; for example, women-led businesses were found to be less likely to receive bank loans compared to men-led businesses. Lastly, the disadvantages experienced by women-led firms were claimed to widen in highly gender-unequal economies and developing countries. Torres et al. (2021) study the impact of the early phase of the COVID-crisis on gender gaps in firm performance for 49 mostly low- and middle-income countries. The results demonstrate that women-led businesses experienced a greater reduction in sales and lower liquidity compared to their male counterparts, which has been reflected in a higher likelihood for women-led companies in several sectors to fall into arrears. On the other hand, as a response to changing circumstances, women-led firms were found to be more likely to increase the utilization of online platforms and make product innovations. Nevertheless, they struggled to obtain any form of public support.

The impact of the pandemic on firms was not gender-neutral in Georgia

The pandemic-induced fragile environment had an adverse impact on entrepreneurs in Georgia– the effects of the shock were significantly more severe for female entrepreneurs than for their male counterparts. In order to assess the gender differences in the impact of the pandemic on firms, we utilize firm-level data on Georgian enterprises from the second round of the World Bank COVID-19 Follow-up Enterprise Survey, conducted in October – November 2020.

Following the methodology as presented in Torres et al. (2021), we assess whether there are differences in the magnitude of reduction in sales revenue (self-reported percentage change in sales revenue one month before the interview as compared to the same period of 2019) and available liquidity for women- and men-led businesses, and whether falling into arrears in any outstanding liabilities is more expected by female top managers (in the next six months from the interview).

Depending on the type of dependent variable, continuous or binary, either Ordinary Least Squares (OLS) or Probit models are estimated, respectively. Along with the gender of the top manager of firms, we also control for sector and firm size. The Georgian database contains a total of 701 enterprises (581 SMEs and 120 micro-businesses).

Table 1. Magnitude of the disproportionate impact of COVID-19 on women-led businesses in Georgia, October-November 2020

Source: The World Bank COVID-19 Follow-up Enterprise Survey, Second Round. Author’s calculations. ***Significant at the 1% significance level; ** significant at the 5% significance level.

Table 1 presents the results of the regression analysis of gender differences among Georgian enterprises in terms of the impact of the pandemic. As observed, women-led businesses reported larger declines in sales, revenues, and liquidity. The predicted drop in sales was 18 percentage points (pp) higher for enterprises with a female top manager than for men-led firms. The larger drop in sales should have been reflected in the reduced cash flow availability and in hardship to cover operating costs. Indeed, as the results demonstrate, women-led enterprises are on average 12.9 pp more likely to have reduced availability of liquidity. This may explain women’s negative future expectations. Moreover, the average predicted probability of expecting to fall into arrears is 11.3 pp higher for women-led firms in Georgia as compared to men-led businesses.

The unequal effect of the COVID-19 crisis on women-led businesses might have been fueled by the disproportionate burden of unpaid care and housework shouldered by women in Georgia, leaving less time available for work and managing enterprises. On the other hand, as Torres et al. (2021) claim, female business owners tend to employ more female workers (the social group more exposed to the unequal burden of the pandemic) than male owners. This, in turn, could further hamper the productivity of women-led businesses and increase their vulnerability to economic shocks.

On the road to recovery

2021 has been characterized by a rather rapid recovery for the Georgian economy, as evidenced by the 10.6% (preliminary estimate) annual growth rate of real GDP. Signs of recovery can also be observed in the labor market – the labor force increased by 4% (YoY) in the 3rd quarter of 2021, while employment was also characterized by a growing trend (1%, YoY).

Along the lines of economic recovery, the gender gap in the labor market also seems to be narrowing. For instance, the steadily growing gap between male and female labor force participation rates seems to stagnate over 2021 (Figure 1). Moreover, as is illustrated in Figure 2 above, the difference between women’s and men’s employment rate changes is positive in 2021, meaning that the employment rate was increasing more (or decreasing less) for women. If this tendency persists, we might stipulate that the disproportionate effects of the COVID-19 crisis on female employment are on the way to recovery.

To examine whether Georgian firms have experienced concurrent movement in their performance along with the economic recovery, we utilize third-round data (from September 2021) of the World Bank COVID-19 Follow-up Enterprise Survey and scrutinize whether the gender differences have narrowed since the previous round of the survey (Table 2).

Table 2. Magnitude of the disproportionate impact of COVID-19 on women-led businesses in Georgia, September 2021.

Source: The World Bank COVID-19 Follow-up Enterprise Survey, Third Round. Author’s calculations. ***Significant at the 1% significance level.

Although the third-round survey data suggests that the predicted percentage drop in sales sharply declined for both men- and women-led businesses, the findings are not statistically significant and therefore cannot claim any signs of recovery in the gender gap in this respect. No signs of recovery are observed in terms of average predicted probability of reduced liquidity of firms and expectations of falling into arrears, either. Gender gaps in these two indicators still persist and are as strong in magnitude as in the second-round survey estimates (from October-November 2020). It seems that despite the economic rebound, not all traces of the pandemic crisis for firms have been eradicated from a gender perspective.


The pandemic came with high economic costs. It hit women disproportionately harder, adversely affecting their employment and entrepreneurial prospects. The unequal burden of the COVID-crisis shouldered by women in Georgia could be one of the reasons for the massive labor force dropouts among female workers and poor performance of women-led businesses. Georgian enterprises with female owners experienced a significantly larger decline in sales compared to their male-owned counterparts, consequently suffering from a shortage of cash flow and fears of falling into arrears.

Despite the great rebound in growth after the initial COVID-19 shock, the pandemic-associated increase in the gender gap seems to have been only partially offset in Georgia. In particular, there is a larger positive upsurge in women’s employment rate, as well as a diminishing difference between male and female labor force participation and employment rates. Following the ongoing recovery in sales revenue of Georgian enterprises (though the predicted gender difference was statistically insignificant), the gender gap in sales is shrinking too. But, in spite of the economic rebound, differences in available liquidity and expectations of falling into arrears have not yet been eradicated, indicating that the adverse influence of the pandemic on women still persists. It leaves female entrepreneurs a still more vulnerable group, which could be of special interest to policymakers to ease their liquidity problems.

Policies should also be directed towards encouraging women to become more economically active. In this regard, remote work seems to pose an opportunity if coupled with affordable childcare support policies.


Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

Social Distancing and Ethnic Diversity

20211214 Social Distancing and Ethnic Diversity Image 01

Voluntary social distancing plays a vital role in containing the spread of the disease during a pandemic. As a public good, it should be more commonplace in more homogeneous and altruistic societies. For healthy people, social distancing offers private benefits, too. If sick people are more likely to stay home, healthy ones have fewer incentives to do so, especially if asymptomatic transmission is perceived to be unlikely. This interplay may lead to a stricter observance of social distancing guidelines in more diverse, less altruistic societies. Consistent with this prediction, we find that mobility reduction following the first local case of COVID-19 was stronger in Russian cities with higher ethnic fractionalization and cities with higher levels of xenophobia and we confirm that mobility reduction in the United States was also higher in counties with higher ethnic fractionalization. Our findings highlight the importance of creating strategic incentives for different population groups in crafting effective public policy.

During the COVID-19 pandemic, governments in almost all affected countries have imposed restrictions aimed at promoting social distancing. However, enforcing these restrictions is logistically and politically costly. The effectiveness of these measures depends heavily on people voluntarily observing social distancing guidelines. The conventional wisdom is that informal social norms are more difficult to sustain in ethnically diverse societies (Alesina and La Ferrara, 2000; Algan et al., 2016). In Egorov et al. (2021), we challenge this notion by showing that during the COVID-19 pandemic ethnic diversity has increased prosocial behavior in Russia and the United States.

At least at the beginning of the pandemic, most people considered themselves healthy. For them, the decision to stay home has been driven more by the fear of getting infected than by the desire to avoid infecting others. The likelihood of getting infected is higher if sick people cannot be expected to self-isolate, which, in turn, depends on their prosocial considerations. If people are subject to out-group biases and care less about people from other groups, then the sick are less likely to engage in social distancing in more diverse places. This makes people who consider themselves healthy more likely to self-isolate. Since healthy people constitute a majority, at least in the early stages of a pandemic, we expect to see more social distancing in more diverse societies. Generally, in these circumstances, the private benefits of those who consider themselves healthy align with social objectives.

In Egorov et al. (2021) we formalize this argument and provide causal evidence of the differential decline in social distancing based on ethnic diversity in Russia and the United States.


Our theory predicts that people engage in social distancing more in places with higher ethnic fractionalization when the probability of getting infected becomes nontrivial. To test this prediction empirically, we use two approaches. First, we report difference-in-differences estimates, where we compare cities with higher and lower levels of ethnic fractionalization before and after the first reported case of COVID-19 infection in their region. Second, we combine the difference-in-differences approach with a two-stage least-squares approach, in which the timing of the first reported case is instrumented using measures of preexisting migration.

One potential concern with the first approach is that the timing of the first case is not fully random. For example, regions could report late COVID-19 cases because their medical capacity precluded them from correctly identifying the virus in time, or because their testing policies could be ineffective, or because their administration was prone to conceal the first cases for a longer time. To deal with these potential confounds in the first approach we use predicted timing of the first case. Specifically, we use the fact that travel connections between various cities and Moscow (where the first major outbreak occurred) could affect the timing of the first case in those cities’ respective regions. We rely on internal migration as a proxy for these types of connections (Mikhailova and Valsecchi, 2020; Valsecchi and Durante, forthcoming) and use a shift-share instrument for internal cross-regional migration to deal with the endogeneity of migration.

Data and Results

To measure social distancing, we use data on people’s movements provided by Russia’s largest technology company, Yandex, which tracks individuals’ cell phones with its mobile apps. In particular, we use daily averages of the Yandex Isolation Index, which aggregates data on people’s movements at the city level and is analogous to the Google Mobility Index. The index is calibrated for each city to be 0 for the busiest hour of the working day, and 5 for the quietest hour of the night before the coronavirus outbreak. We use daily data for 302 cities with a population over 50,000 from February 23, 2020, through April 21, 2020.

Information on the first reported case of COVID-19 in each region is taken from the government-agency website that contains official information about the pandemic. Data on ethnic fractionalization is based on the 2010 Census. Information on interregional migration and control variables comes from the Russian Federal State Statistics Service.

Figure 1. Isolation Over Time for Places with High and Low Ethnic Fractionalization

Source: Egorov et al. (2021)

Figure 1 shows no visible difference in the behavior of people in cities with low and high levels of ethnic fractionalization before the first coronavirus case. In both groups of cities, people have engaged in more social distancing since the discovery of the first case. However, after one week, people in more fractionalized cities have been more likely to stay home than people in less fractionalized cities. The effect does not manifest itself immediately after the discovery of the first case, which likely reflects the fact that a certain time is needed to disseminate information about the discovery of the coronavirus in the region. Moreover, the growth in self-isolation in more fractionalized cities is somewhat lower in the first days after the discovery of the first case, which may be driven by people catching up on unfinished tasks that require mobility, such as last-minute purchases, in anticipation of more stringent self-isolation in the future.

The results of the difference-in-differences and IV estimation confirm the results of the visual analysis. The magnitudes of the IV estimation imply that a one-standard-deviation increase in ethnic fractionalization leads to 3.7% higher social distancing following the report of the first local COVID-19 case. In other words, a one-standard-deviation increase in ethnic fractionalization can explain 5.7% of the average mobility reduction after the report of the first case or, alternatively, 4.7% of the weekday-weekend gap for an average locality.

To make sure that the results are not Russia- specific, we also show that ethnic fractionalization led to a bigger reduction in mobility following the first local COVID-19 case using the United States county-level data.


Overall, the results in Egorov et al. (2021) highlight the role of ethnic diversity in voluntary adherence to socially beneficial norms, such as self-isolation and social distancing during a pandemic. We show that people in more diverse places were more likely to restrict their mobility following the reports of the first local COVID-19 cases.

Our study has important implications for government policy. It highlights not only that the propensity of different groups of people to engage in prosocial behavior may differ but also that there may be important strategic effects. In the context of the pandemic, decisions by healthy and sick individuals to self-isolate are strategic substitutes. This means, for example, that in a homogeneous society with high levels of tolerance, extensive testing would allow people to learn that they are sick and self-isolate, enabling the rest to go out with little fear. In a heterogeneous society with low levels of tolerance, the same policy may spur people who learn that they are contagious to go out more because they have little to lose, with the exact opposite implications for the healthy population.


  • Alesina, A., La Ferrara, E., 2000. Participation in heterogeneous communities. Quarterly Journal of Economics. 115, 847–904.
  • Algan, Y., Hémet, C., Laitin, D.D., 2016. The social effects of ethnic diversity at the local level: a natural experiment with exogenous residential allocation. Journal of Political Economics. 124, 696–733.
  • Egorov, G., Enikolopov, R., A., Makarin, and M. Petrova. 2021. Divided We Stay Home: Social Distancing and Ethnic Diversity” Journal of Public Economics. 194: 104328.
  • Mikhailova, T., Valsecchi, M., 2020. Internal migration and Covid-19 (in Russian). In: Economic Policy in Times of Covid-19, New Economic School, pp. 26–33.
  • Valsecchi, M., Durante, R., forthcoming. Internal Migration Networks And Mortality In Home Communities: Evidence From Italy During The Covid-19 Pandemic. European Economic Review.

Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

Russian Exporters in the Face of the COVID-19 Pandemic Crisis

20211108 Russian Exporters in the Face of the COVID-19 Image 01

This brief summarizes the results of recent work on the effects of the COVID-19 pandemic on Russian exporting companies (Volchkova, 2021).  We use data from the CEFIR NES survey of exporters conducted in 2020. 72% of respondents reported that they were affected by the crisis. We scrutinize this impact. Contrary to popular wisdom, we observe little difference in delays of inputs by domestic and foreign suppliers. On the other hand, exporters experienced more disruptions in their sales in foreign destinations than in the domestic market. Possible reasons for this may be due to restrictions on international travel.


According to experts at the Gaidar Institute (Knobel, Firanchuk, 2021), in 2020, Russia’s non-resource non-energy exports, decreased by 4.3%, while export prices fell by 4.1 % on average. The export of high-tech goods decreased by 14% due to a reduction in the physical volume of export. These changes in export intensity are mainly associated with the COVID-19 pandemic crisis. But are exporting firms more affected by the crisis than firms only active in the domestic market? What are the main channels through which the crisis influenced exporters?  And how do exporters adjust to the COVID-19 related shocks?

The analysis in this brief is based on forthcoming publication in the Journal of New Economic Association (Volchkova, 2021). We use data from a survey of Russian non-resource exporters conducted in 2020. We show that involvement in international trade did not affect the company’s vulnerability to the crisis on the production side: supply delays were equally likely to occur from domestic and foreign suppliers. These findings are consistent with Bonadio et al. (2021) who consider a numerical multi-sectoral model for 64 countries around the world linked by supply chains. They show that, in the face of the employment shocks associated with quarantine measures and switching to a remote work format, the contribution of global chains to the decline of real GDP is about one quarter. Importantly, the authors show that the “re-nationalization” of supply chains does not make countries more resilient to shocks associated with quarantine measures on the labor market because these shocks are also bad for domestic industries.

At the same time, our results indicate that exporting companies are exposed to additional risks associated with the need to adjust to shocks in the sales markets. According to the data, exporters find it more difficult to adjust their sales in foreign markets than in the domestic one. This is consistent with the fact that, during the pandemic, all countries introduced a strict ban on international travel, reducing the possibility of establishing new business ties through personal contacts. Similarly, Benzi et al. (2020) show a significant negative effect of international travel restrictions on the export of services.

Survey of Non-resource Exporters

The survey of exporters was carried out in June – November 2020 by CEFIR NES. The primary purpose of the survey was to identify and estimate barriers to the export of non-primary non-energy products. In the context of the developing economic crisis caused by the COVID-19 pandemic, we have added several questions to identify how the crisis influenced companies’ operations and how the respondent firms adjusted to the new conditions.

The survey was conducted using a representative sample of Russian exporting firms. As a control group, we interviewed non-exporting firms with (observable) characteristics (region, industry, labor productivity) similar to those of the surveyed exporters. Altogether, 928 exporting companies and 344 non-exporting companies were interviewed during the field stage of the study.

Most exporting companies that took part in the survey produce food products, chemicals, machinery and equipment, electrical equipment, metal products, and timber. On average, a surveyed exporter had 827 full-time employees; 25% of the firms had fewer than 26 employees. More than half of the surveyed exporting firms (53%) are also importers: 81% import raw materials and other inputs, 66% import equipment, and 22% import technology. Most interviewed exporters sell their products both abroad and on the domestic market. On average, an enterprise supplies 67% of its output to the domestic market and 32% abroad.

Impact of the COVID-19 Crisis on Firms’ Performance

Among exporters that participated in the survey, 25% reported that their business was not affected by the COVID-19 crisis, while 72% of respondents stated that the crisis did have an impact. Like any crisis, the COVID-19 pandemic created problems for some enterprises and provided new beneficial opportunities for others. According to the data, exporting businesses were significantly more likely to be negatively affected by the crisis than their non-exporting counterparts, and the impact of the crisis was not correlated with the size of the enterprise. Figure 1 presents the exporters’ answers to the question of how their sales in the domestic and foreign markets have changed with the COVID-19 pandemic.

The distribution of changes in sales volume in domestic and foreign markets significantly differ from each other. Estimates of the mean values of changes in sales volumes also differ significantly: the average drop in sales in the domestic market was 5%, while for the external market, it reached 17%. Hence, in times of the COVID-19 crisis, opportunities for growth were less prominent in foreign markets than in the domestic one, while significant market losses were more frequent.

Figure 1. Change in sales of export companies associated with the COVID-19 pandemic

Source: Survey of non-resource exporters, CEFIR NES, 2020.

Adjustment to the Crisis

The most frequently used crisis adjustment measure was employees transition to remote work – it was reported by 70% of the surveyed companies. 25% of exporters were forced to suspend their work during the crisis, while 72% were not. 14% of respondents stated they had to cut their payroll expenditures and other non-monetary benefits for employees (food, insurance, etc.), 12% of companies sent workers on unpaid leave. Only 6.5% of export firms had to lay off workers, while 91% handled the crisis without layoffs.

Comparing exporters’ answers with those of non-exporters while controlling for enterprise size, we conclude that exporting firms were more rigid in their adjustment  to the crisis. They were significantly more likely to suspend enterprise activities, dismiss of employees, send workers on unpaid leave, and reduce of wages. Also, these events were more likely to occur for smaller companies than for larger ones.

At the same time, flexible adjustment measures such as remote work were equally likely to be used by exporters and non-exporters, as well as by firms of different sizes. In general, Russian exporters of non-primary goods maintained their efficiency mainly by adjusting the labor relations to the new epidemiological conditions rather than by reducing employee-related expenses.

Dealing with Counterparties

Delays in the supply of components and raw materials were reported by 36% of the surveyed companies, and such delays were equally likely for shipments from abroad and domestic shipments. There is a perception that international supply chains in the context of the pandemic crisis are an additional risk factor. Our results indicate that domestic and international supply chains were equally challenged in 2020. Nevertheless, non-exporting companies faced the problem of delayed deliveries significantly less often than exporters did, and about 60% of companies experienced no problems at all on the input supply side.

27% of surveyed exporters stated that they delayed payments to counterparties. Non-exporting companies reported these reactions much less frequently regardless of firm size.

On the sales side, half of the surveyed exporters experienced delays in payments from their customers during the pandemic crisis. Non-exporting enterprises encountered the problems with the same frequency, and companies of all sizes were affected by this obstacle equally.

The cases of planned purchases cancellation on behalf of buyers were reported by 34% of exporting companies. Exporters experienced these problems significantly more often than non-exporters, and smaller companies experienced them much more often than larger ones.

Crossing international borders presented a certain problem for Russian exporters when it concerns product delivery. Just over half of the respondents indicated that they had to delay deliveries due to difficulties with border crossing. However, about the same share of companies (48%) reported that they delayed products delivery due to the introduction of lockdowns. Thus, during the COVID-19 pandemic, exporters’ operations were complicated to the same extent by problems related to border crossings as by those associated with lockdown regimes.


It is widely believed that international exposure of companies in the context of the COVID-19 pandemic crisis creates additional risks. Our study shows that, regarding existing inputs supply, international relations pose problems for Russian companies just as often as relations with domestic partners. As far as sales are concerned, adjustment to the crisis was better on the domestic market than on foreign markets. A possible explanation of this phenomenon is that, in addition to the shocks associated with quarantine measures in the labor market, access to foreign markets was hampered by restrictions on international travel, which is essential for readjusting contractual relations to explore new opportunities brought by crises (Cristea, 2011). Without personal interaction, new contracts were more difficult to launch. Thus firms’ opportunities to adjust foreign sales were more restricted than the ones in the domestic market.


Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.

Media mentions: Key takeaways from this policy brief have been published by one of the most influential media outlets in Russia Kommersant – Коммерсант: «Ковид сильнее ударил по экспортерам». Исследование ЦЭФИР РЭШ.