Background

Previous research has conceptualized individuals’ resilience in a multilevel framework in which resilience is an individual’s ability to maintain desired outcomes by accessing resources and adapting resource access through various life-course capitals in order to meet changing resource needs and reduced resource access when facing disturbances (Plach and Aassve, Reference Plach and Aassve2026). These life-course capitals are multidimensional including economic, human, social, and “institutional” capital. Resilience is thus conceptualized and can be measured on the individual level but originates from multiple analytical levels including not only from micro-level economic and human capital but also social capital emerging from social support of meso-level social networks and support from macro-level public institutions. This multidimensional and multilevel view builds on literature theorizing resilience in terms of “capitals” or “assets” or “capacities,” providing resources to achieve well-being (e.g. Norris et al., Reference Norris, Stevens, Pfefferbaum, Wyche and Pfefferbaum2008; Alinovi et al., Reference Alinovi, Mane, Romano, Benedetti, Bee, Espa and Piersimoni2010; European Commission Joint Research Centre., 2017; d’Errico et al., Reference d’Errico, Romano and Pietrelli2018; Mu, Reference Mu2021). But this literature falls short in considering all relevant dimensions, all analytical levels, and potential interaction effects between different factors and levels. Especially the role of public institutions is not explicitly considered.

Economic capital captures the material resources and economic security on the micro level that help households absorb and adapt to disturbances (e.g., savings, arrears, housing security). Income and savings can mitigate economic hardship, enable necessary investments, and outsource demands like childcare and housework. Indeed, individuals with lower socioeconomic status experience greater psychological distress following disasters than those with higher socioeconomic status (Norris et al., Reference Norris, Friedman, Watson, Byrne, Diaz and Kaniasty2002), and children with access to basic necessities exhibit greater resilience (Ungar et al., Reference Ungar, Connelly, Liebenberg and Theron2019). Human capital reflects individual endowments – most prominently education and health – that condition adaptive capacity over the life course. Human capital facilitates adaptation through enabling gathering and interpreting information, reorganizing within a new environment, and reducing uncertainty (Liu et al., Reference Liu, Meng, Wu, Ran, Cao and Liu2023). Social capital concerns the quality and adaptiveness of social ties and thus the access to social support, which can buffer stress, mobilize help, and sustain coping. In the face of adversity, a strong social network can offer crucial support (Ungar et al., Reference Ungar, Connelly, Liebenberg and Theron2019), enhancing coping mechanisms (Sippel et al., Reference Sippel, Pietrzak, Charney, Mayes and Southwick2015) and contributing to lower depression rates and higher life satisfaction (Fuller-Iglesias et al., Reference Fuller-Iglesias, Sellars and Antonucci2008; Richards, Reference Richards2016). At a broader level, social networks play a key role in community and societal recovery (Norris et al., Reference Norris, Stevens, Pfefferbaum, Wyche and Pfefferbaum2008; Sherrieb et al., Reference Sherrieb, Norris and Galea2010; Aldrich, Reference Aldrich2011).

Finally, institutional capital refers to the enabling resources that stem from public institutions, including both the level and composition of welfare state support and the quality of governance with which support is delivered. During the COVID-19 pandemic, for instance, governments in industrialized countries implemented health and economic support policies to mitigate the crisis (Corti et al., Reference Corti, Gros, Ruiz de la Ossa, Liscai, Kiss-Gálfalvi, Gstrein, Herold and Dolls2022; Hemerijck and Huguenot-Noël, Reference Hemerijck and Huguenot-Noël2022). These measures resulted in record-low poverty rates in the U.S. despite high unemployment (Parolin, Reference Parolin2023), widespread short-time work in Europe to prevent mass layoffs (Salerno et al., Reference Salerno, Börsch-Supan, López-Falcón and Rausch2024), and economic support policies that helped stabilize fertility rates in regions with weaker pre-pandemic support (Plach et al., Reference Plach, Aassve, Cavalli, Mencarini and Sanders2023). Public support also strengthens other forms of life-course capital. Welfare states have played a crucial role in enhancing economic resilience during both the Great Recession and the COVID-19 pandemic, particularly where policies have emphasized human capacity building and social inclusion (Hemerijck and Huguenot-Noël, Reference Hemerijck and Huguenot-Noël2022).

Prior work shows that all four sets of capitals matter for an individual’s resilience (Plach and Aassve, Reference Plach and Aassve2026). Health and social inclusion were found to be relatively most important. Institutional capital provides the crucial macro-to-micro transmission channel through which policies can capacitate citizens and citizens themselves can in turn strengthen public institutions. This resonates with findings that generalized and institutional trust and good governance are strongly associated with subjective well-being and can mitigate the well-being costs of adverse shocks (Helliwell et al., Reference Helliwell, Huang, Grover and Wang2018, Reference Helliwell, Huang, Wang, Norton, Helliwell, Layard, Sachs and De Neve2020).

We hypothesize that objective measures of the welfare state are linked to individual resilience through two broad channels. First, the quantity and composition of public support matter [Hypothesis 1a]: how much is spent, and on which policy areas, including family, health, active labor market programs, unemployment support, housing, and incapacity-related spending. Family support can buffer the additional resource needs of households with children and for instance reduce poverty risks, especially when benefits are generous and well-designed (Misra et al., Reference Misra, Moller and Budig2012; Van Lancker and Van Mechelen, Reference Van Lancker and Van Mechelen2015). Health spending can prevent and mitigate health shocks, with higher expenditures associated with better population health and lower mortality, particularly when complemented by social spending (Nixon and Ulmann, Reference Nixon and Ulmann2006; Bradley et al., Reference Bradley, Elkins, Herrin and Elbel2011). Active labor market programs, especially human-capital-intensive interventions (training, employment services), show improving employment impacts over time (Card et al., Reference Card, Kluve and Weber2018, Reference Card, Kluve and Weber2010) and can support coping during job transitions. In the same vein, unemployment insurance helps households bridge periods without work by smoothing consumption and easing liquidity constraints, thereby sustaining adaptive capacity (Gruber, Reference Gruber1997; Chetty, Reference Chetty2008). Housing-related expenditures that reduce insecurity and unaffordability are linked to better mental health and well-being, lowering stressors that undermine resilience (Clapham et al., Reference Clapham, Foye and Christian2018; Carrere et al., Reference Carrere, Vásquez-Vera, Pérez-Luna, Novoa and Borrell2022; Keen et al., Reference Keen, Chen, Slopen, Sandel, Copeland and Tiemeier2023). Finally, incapacity-related benefits provide income security and access to supports for those whose health limits labor-market participation, stabilizing resources during adverse life-course events (Bound and Burkhauser, Reference Bound, Burkhauser, Ashenfelter and Card1999).

Second, we hypothesize that the quality of governance matters [Hypothesis 1b] as citizens are more likely to cope with adversity when public action is effective, rules-based, predictable, and fair. A substantial body of work links governance quality to citizens’ subjective well-being: studies consistently show that political stability, rule of law, and control of corruption are associated with higher life satisfaction and lower psychological distress (see e.g. Helliwell and Huang, Reference Helliwell and Huang2008; Welsch, Reference Welsch2008; Ott, Reference Ott2010). It is therefore illuminating to analyze governance beyond “government effectiveness” to include voice and accountability, political stability and absence of violence, regulatory quality, rule of law, and control of corruption. Voice and accountability provide channels to express preferences and sanction poor performance, which improves responsiveness and strengthens perceived agency, efficacy, and trust – factors associated with better well-being and coping (Frey and Stutzer, Reference Frey and Stutzer2000; Besley and Burgess, Reference Besley and Burgess2002; Helliwell and Putnam, Reference Helliwell and Putnam2004; Tyler, Reference Tyler2006). Political stability and absence of violence reduce uncertainty and disruption, enabling households to plan and invest in coping strategies and avoiding the psychological toll associated with insecurity and conflict (Alesina et al., Reference Alesina, Ozler, Roubini and Swagel1996; Steel et al., Reference Steel, Chey, Silove, Marnane, Bryant and van Ommeren2009). And this stability also supports continuity in essential services during shocks (Kruk et al., Reference Kruk, Freedman, Anglin and Waldman2010). Regulatory quality – clear, proportionate, and implementable rules – lowers frictions and administrative burdens, clarifies entitlements, and facilitates adaptation during shocks and life-course transitions (Djankov et al., Reference Djankov, La Porta, Lopez-de Silanes and Shleifer2002; Herd and Moynihan, Reference Herd and Moynihan2018). The rule of law – impartial and predictable enforcement – curbs arbitrariness, supports cooperation, and allows individuals to mobilize resources with confidence, thereby enhancing capacity to adjust when adversity strikes (North, Reference North1990; Acemoglu and Johnson, Reference Acemoglu and Johnson2005; Rothstein, Reference Rothstein2011). Finally, control of corruption improves the fairness and reliability of service delivery and sustains generalized and institutional trust, both linked to better health and higher well-being and resilience (Mauro, Reference Mauro1995; Holmberg and Rothstein, Reference Holmberg and Rothstein2011; Helliwell et al., Reference Helliwell, Huang, Grover and Wang2018).

A social investment perspective stresses capacitation: policies that expand people’s real opportunities and capabilities ex ante through person-facing services (e.g., early childhood education and care, training and lifelong learning, active labor market programs, work-life balance supports, long-term care) (Pöyliö and Hemerijck, Reference Pöyliö and Hemerijck2026). Recent work anchors social investment in a capabilitarian view of justice: policy should secure “fertile functionings” through “secure capabilities,” providing stepping-stones that help individuals traverse life-course transitions (Bagadirov et al., Reference Bagadirov, Hemerijck and Roig2025). Predictable and impartial institutions – those socially investing consistently and generously in citizens and those with high governance performance such as political stability, rule of law, and control of corruption – decrease citizens’ uncertainty and arbitrariness, thereby enabling them to plan, adapt, and act – thus increasing their resilience.

Such capacitating can generate subjective well-being returns – not only better labor market outcomes, but also enhanced agency (a sense of control), recognition (being treated fairly and respectfully), and resilience itself (Pöyliö and Hemerijck, Reference Pöyliö and Hemerijck2026; Wagner et al., Reference Wagner, Heta and Hemerijck2025). Person-facing services and fair procedures reduce material and psychological uncertainty, build skills and self-efficacy, and signal procedural justice and respect – each of which supports perceived control and optimistic evaluations under stress (Bandura, Reference Bandura1997; Sen, Reference Sen1999; Tyler, Reference Tyler2006). These well-being returns can, in turn, foster positive sociopolitical feedbacks (e.g., trust, participation, solidaristic attitudes) by strengthening generalized and institutional trust and civic efficacy, thereby reinforcing institutional legitimacy and the effectiveness of public action (Rothstein and Stolle, Reference Rothstein and Stolle2008; Porumbescu, Reference Porumbescu2017; Helliwell et al., Reference Helliwell, Huang, Grover and Wang2018).

In line with this, there are multiple ways in which objective measures of public institutions can affect citizens’ resilience. First, directly received transfers or services from public institutions “today” can provide resources and security. Second, past received transfers or services across the life-course can position individuals more favorably to adapt to stressors. Third, transfers or services received by other citizens can indirectly affect an individual.

The social investment logic also motivates our attention not only to objective spending levels and governance performance but also to how citizens subjectively experience them. Citizens’ subjective experiences of public service quality, fairness, and trust are underrepresented in previous research. But the above-mentioned sociopolitical feedback means that subjective views of governance might shape citizens’ resilience through the social investment mechanism just highlighted – perceived agency, recognition, and optimism. Citizens form resilience-relevant expectations at the state–citizen interface: whether childcare and schools are reliable, whether public employment services are helpful, whether healthcare is responsive, and whether public authorities are trustworthy and democratically accountable. Positive service experiences and fair and trustworthy procedures might matter above and beyond spending levels because they signal recognition, build confidence, and reduce stress – each a plausible pathway to resilience.

In line with his, we hypothesize that subjective views of welfare state and governance are linked to resilience. First, we hypothesize that subjective perceptions of the quality of childcare services, education systems, healthcare, and public employment services matter for individuals resilience [Hypothesis 2a]. The perception of reliable, high-quality childcare can reduce work–family conflict and parental stress, helping households maintain routines and employment during shocks. In fact, evidence shows that family well-being is sensitive to the design and quality of early-childhood provision of childcare (Baker et al., Reference Baker, Gruber and Milligan2008; Burger, Reference Burger2010; Havnes and Mogstad, Reference Havnes and Mogstad2011). Perceived education quality is arguably tied to both cognitive and non-cognitive skill formation – skills that underpin problem solving, self-regulation, and forward planning, all central to coping capacity (Cunha and Heckman, Reference Cunha and Heckman2007; Oreopoulos and Salvanes, Reference Oreopoulos and Salvanes2011). In healthcare, better patient experience is associated with greater adherence and improved outcomes, which can stabilize health – a foundational resource for resilience (Doyle et al., Reference Doyle, Lennox and Bell2013; Manary et al., Reference Manary, Boulding, Staelin and Glickman2013). As for the subjective views on public employment services, supportive and competent counselling and job-search assistance can enhance job-search self-efficacy and expedite transitions back to work, buffering the psychological and economic strains of unemployment (Vinokur et al., Reference Vinokur, Schul and Price1995; Crépon et al., Reference Crépon, Duflo, Gurgand, Rathelot and Zamora2013).

Second, we hypothesize that citizens’ subjective satisfaction with democracy and trust in core public institutions – the government, the European Union, the police, and the health system – are linked to resilience [Hypothesis 2b]. Institutional trust, and likewise satisfaction with democracy, reduces uncertainty and fosters a sense of control and collective efficacy, which are associated with better well-being and adaptive behaviors in crises (Helliwell et al., Reference Helliwell, Huang, Grover and Wang2018; Sibley et al., Reference Sibley, Greaves, Satherley, Wilson, Overall, Lee, Milojev, Bulbulia, Osborne, Milfont, Houkamau, Duck, Vickers-Jones and Barlow2020; Devine et al., Reference Devine, Gaskell, Jennings and Stoker2021). Trust in the health system is related to service uptake, adherence, and continuity of care, reinforcing the health foundations of coping (Gilson, Reference Gilson2003; Ozawa and Sripad, Reference Ozawa and Sripad2013). Confidence in law-enforcement and justice institutions is associated with lower fear and higher perceived safety, conditions that support planning and daily functioning under stress (Sunshine and Tyler, Reference Sunshine and Tyler2003). Perceived performance and fairness of government and European Union institutions also shape citizens’ expectations about credible support and policy continuity during shocks, with implications for both compliance and subjective well-being (Van Ryzin, Reference Van Ryzin2011; Armingeon and Ceka, Reference Armingeon and Ceka2014).

In addition, we hypothesize that subjective perceptions of the quality of social services and trust and satisfaction with public institutions matter differently across country-contexts [Hypothesis 2c]. We expect some heterogeneity across broad country-groups which differ in their welfare regimes, economic setting, and socio-cultural landscape. Positive subjective perceptions of public institutions might matter particularly in less extensive welfare regimes by compensating objective limitations. Instead, subjective perceptions might also matter particularly where citizens rely most on public institutions, that is, in the most extensive welfare regimes.

Data and methods

Data

To empirically analyze the determinants of individuals’ resilience in our theoretical framework, we use data from the Eurofound Living, Working and COVID-19 survey (LWC) (Ahrendt et al., Reference Ahrendt, Mascherini, Sándor, Ganko, Jansova, Kärkkäinen, Mereuta, Monteleone, Prina and di Vozzo2022; Eurofound, 2022). The survey was carried out in 28 countries (the European Union and the UK) in five survey waves during the COVID-19 pandemic (Apr/May 2020; Jun/Jul 2020; Feb/Mar 2021; Oct/Nov 2021; and Apr/May 2022). We use survey wave 2-5 because social inclusion was not measured in wave 1. We use 23 countries for which additional data on public social expenditure is available (i.e. dropping Bulgaria, Cyprus, Croatia, Malta, and Romania). The COVID-19 pandemic is a suitable setting, making resilience particularly relevant and inducing variation in life-course capitals.

Resilience is measured by an index, constructed as the average of the two items “When things go wrong in my life, it generally takes me a long time to get back to normal” and “I find it difficult to deal with important problems that come up in my life,” measured on a 5-point scale from strongly disagree to strongly agree. These items are included in the validated and widely recognized CD-RISC scales (Connor and Davidson, Reference Connor and Davidson2003; Campbell-Sills and Stein, Reference Campbell-Sills and Stein2007; Vaishnavi et al., Reference Vaishnavi, Connor and Davidson2007; Kuiper et al., Reference Kuiper, van Leeuwen, Stolwijk-Swüste and Post2019).

Economic capital is measured by three variables broadly reflecting income, wealth, and economic security. First, lower arrears is the sum of six reverted survey items asking “Has your household been in arrears at any time during the past 3 months, that is, unable to pay as scheduled …” for rent/accommodation, utility bills, consumer loans, phone/internet bills, informal loans, or healthcare/health insurance. We use this variable as it reflects the grade of sufficiency of income in covering living expenses. A direct measure of income is not available in the LWC survey but income sufficiency to pay bills is arguably easier to report for respondents and still gives us an indication of their income relative to living expenses. Second, savings is measured as “If your household would not receive any income, how long would your household be able to maintain the same standard of living?” on the scale “no savings,”, “less than 3 months, “3 to 6 months,” “6 to 12 months; “12 or more months.” Third, housing security is measured as “How likely or unlikely do you think it is that you will need to leave your accommodation within the next 3 months?” on a 5-point scale. Human capital is reflected by education level, split in tertiary versus non-tertiary education, and health, measured as “In general, how is your health?” on a 5-point scale. Social capital is measured as social inclusion by the reverted survey item “I feel left out of society” on a 5-point scale.

To measure objective institutional capital, we complement our data set with country-level data. First, public social expenditure per capita at constant purchasing power parity per year comes from the OECD (OECD, 2022). The total social expenditure included in our baseline model is constructed from cash and in-kind benefits in the policy areas of old age, survivors, incapacity-related benefits, health, family, active labor market programs, unemployment, housing, and other social policy areas (excluding expenditure in education). We also conduct more specific analysis using data on spending by area. These more granular data are available only until 2021. That means that any analysis using these data will employ only survey waves 2–4, but not survey wave 5.

Second, we use data from the World Governance Indicators (WGI) of the World Bank. These data are available for all years in which we have survey data. We employ the government effectiveness index (GEI) in our main model. The GEI captures the quality of public services, civil services and their independence from political pressures, the quality of policy formulation and implementation, together with the credibility of a government’s commitment to such policies (Kaufmann et al., Reference Kaufmann, Kraay and Mastruzzi2011; Kaufmann and Kraay, Reference Kaufmann and Kraay2023). We then conduct more specific analysis using the other WGI governance indicators of Voice and Accountability (VAI), Political Stability and Absence of Violence/Terrorism (PVI), Regulatory Quality (RQI), Rule of Law (RLI), and Control of Corruption (CCI). We standardize all variables to allow comparisons of coefficients to assess relative importance.

In the final part of our analysis, we use subjective measures on people’s perceptions of governance. First, we use LWC survey respondents’ subjective perceptions on the quality of various public services: public employment services, childcare services, long-term care services, the education system, doctors, and hospitals. The survey item is “In general, how would you rate the quality of each of the following public services in your country?”. These items are only available in the last two survey waves. Therefore, any analysis using these items will be restricted to survey waves 4 and 5.

Second, we employ LWC measures on the satisfaction with democracy in the respondent’s country (“On the whole, how satisfied are you with the way democracy works in your country?”) and the respondent’s trust (“How much do you personally trust …?”) in government, the European Union, the police, and the health system. We also use the survey item of generalized trust (“Generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people?”) as a comparison for the institutional trust measures. Generalized trust is available for survey waves 3–5, while satisfaction with democracy and institutional trust measures are available for all survey waves used (2–5).

Table 1 gives an overview of the key variables and data used on them. In addition, Figure A1 reports correlation coefficients between the eight main life-course capital variables. These are only weakly correlated with exception of the three economic capital variables (weakly-moderately correlated among each other) and the two public institutions variables (moderately-strongly correlated among each other).

Table 1.

Overview of variables.

Table 1. Long description

The table consists of three columns: Variable, Description, and Data used.

* Lower arrears: Sum of six reverted items regarding household arrears in the past 3 months for rent, utilities, loans, phone, or healthcare. Data from L W C survey waves 2 to 5.

* Savings: Household ability to maintain standard of living without income on a scale from no savings to 12 or more months. Data from L W C survey waves 1 to 5.

* Housing security: Likelihood of needing to leave accommodation within 3 months on a 5-point scale. Data from L W C survey waves 1 to 5.

* Tertiary education: Tertiary versus non-tertiary education. Data from L W C survey waves 2 to 5.

* Health: General health rating on a 5-point scale. Data from L W C survey waves 2 to 5.

* Social inclusion: Feeling left out of society on a 5-point scale. Data from L W C survey waves 2 to 5.

* Public social expenditure: Total social expenditure including cash and in-kind benefits across various policy areas. O E C D data by country-year.

* Government effectiveness G E I: Quality of public and civil services, policy formulation, and government commitment. World Governance Indicators W G I by country-year.

* Public social expenditure by area: Specific policy areas including family, health, labor, unemployment, housing, and incapacity. O E C D data by country-year until 2021.

* Other W G I indicators: Includes V A I, P V I, R Q I, R L I, and C C I. World Governance Indicators W G I by country-year.

* Perceptions of quality of public services: Rating of services like employment, childcare, education, and healthcare. Data from L W C survey waves 4 to 5.

* Satisfaction and trust: Measures of interpersonal trust, satisfaction with democracy, and trust in the government, E U, police, and health system. Data from L W C survey waves 2 to 5.

To construct our sample, we remove individuals with inconsistencies in demographic variables across survey waves or missing values in any of the used variables (resilience, the life-course capital variables, and age and gender). Social inclusion is unavailable in the first survey wave, while social expenditure is unavailable for the non-OECD countries Bulgaria, Croatia, Cyprus, Malta, and Romania. The UK is not included in survey waves 4 and 5 and has rather small sample sizes in waves 2 and 3. We are restricting our sample to an unbalanced panel of individuals observed at least twice across survey waves (on average 3.05 observations) to obtain a consistent three-level data structure: within-individual, between-individual, between-country or -country-groups, broadly reflecting welfare regimes and broad economic and socio-cultural heterogeneity, for instance in family ties.Footnote ¹ This results in 45,294 observations from 16,071 individuals across 23 countries, and, in total, four survey waves.

Methods

In the first baseline part of our analysis, we report results from a model including the eight core determinants of resilience – measuring economic, human, social, and institutional capital – identified in previous research (Plach and Aassve, Reference Plach and Aassve2026). To leverage variation between countries, between individuals, and within individuals over time we estimate various multilevel models and an individual FEs model based on the following OLS model:

$$ {\displaystyle \begin{array}{c}{Resilience}_{i,t}=\alpha +{\beta}_1{lowerArrears}_{i,t}+{\beta}_2{Savings}_{i,t}+{\beta}_3{HousingSec}_{i,t}+{\beta}_4{TertEdu}_{i,t}\\ {}+\hskip.4em {\beta}_5{Health}_{i,t}+{\beta}_6{SocIncl}_{i,t}+{\beta}_7{PublSocExp}_{c(i),y(t)}+{\beta}_8{GovEffect}_{c(i),y(t)}\\ {}+\hskip.4em \sum \limits_k\sum \limits_{m\ne k}{\gamma}^{k,m}\hskip1em {D}_{i,t}^k\cdot {D}_{i,t}^m+{\delta}_1{Age}_{i,t}+{\delta}_2{Female}_i+{\unicode{x025B}}_{i,t}\end{array}} $$

$ Resilience $ represents the resilience index for individual $ i $ at time $ t $ (survey waves 2–5). Explanatory variables include the eight life-course capital variables: lower arrears, savings, housing security, tertiary education, health, social inclusion, public social expenditure, and government effectiveness. We also include two-way interaction terms between life-course capital variables to estimate potential substitution effects in $ \gamma $ . We control for age and gender. Standard errors are clustered at the individual level. We are accounting for individual REs and REs or FEs on country or country-group level (multilevel models), or, most restrictively, directly individual FEs. Country-groups are broadly reflecting welfare regimes and economic and socio-cultural heterogeneity.

In a second step, we take the baseline model and vary measures of objective governance. We exchange total social expenditure for social expenditure in specific spending areas. We then exchange the GEI for the other five governance indices. We run one model per governance indicator instead of including all variables in one model, as their correlation would lead to collinearity issues and thus an imprecise estimation of coefficients.

In a third step, we add to the baseline model measures of citizens’ subjective perceptions of public institutions – their rating of the quality of public services and their trust in and satisfaction with public institutions – one variable at a time to estimate the association with citizens’ resilience beyond the objective, country-level institutional capital measures.

For these models on citizens’ subjective perceptions, we also estimate country-group-specific coefficients to assess how citizens’ views matter differently for their resilience across country-group contexts (Southern Europe; Eastern Europe; Baltics; Anglo-Saxons; Northern Europe; Central Europe; Western Europe). We perform this heterogeneity analysis only for models on subjective measures, as objective measures on social expenditure and governance indicators vary only across countries and there is thus not enough between-country-group variation to precisely estimate country-group-specific coefficients.