The role of financial stress in mental health changes during COVID-19

On March 11, 2020, WHO declared COVID-19 a pandemic, raising concerns about population mental health. Prior pandemics and early COVID-19 studies indicated mixed mental health outcomes, with common symptoms including PTSD, depression, anxiety, insomnia, and loneliness. Three pathways were proposed for pandemic effects on mental health: (1) the disease and infection threat, (2) containment measures (quarantine/social distancing) disrupting routines and activities, and (3) economic consequences. Socioeconomically disadvantaged groups were considered particularly vulnerable due to higher infection risk, greater disruption from lockdowns, and exposure to economic shocks. Financial stress—a subjective sense of insufficient resources and lack of control, with responses including worry and short-term focus—may mediate links between financial vulnerability and health. This study leveraged pre- and during-pandemic longitudinal data to examine how changes in financial stress relate to changes in mental health and how pre-pandemic income, savings, and debts, as well as income changes during COVID-19, relate to financial stress. The authors tested three hypotheses: (1) Increases in financial stress during COVID-19 relate to decreased mental health, whereas decreases relate to increased mental health. (2) Falling incomes during COVID-19 and low incomes, low savings, and high debts before COVID-19 relate to increases in financial stress during COVID-19. (3) Changes in financial stress during COVID-19 mediate the association between financial vulnerability (income drops, low incomes, low savings, high debts) and mental health changes.

Evidence on mental health during COVID-19 is mixed: some studies find worsened mental health, others no change or even improvements. Frequently reported issues include PTSD, depression, anxiety, insomnia, and loneliness. Conceptual frameworks identify three pathways—disease threat, containment measures, and economic consequences—through which COVID-19 can affect mental health. Socioeconomic disadvantage is linked to higher infection rates, less flexibility to work from home, greater job and income insecurity, and thus elevated mental health risks. Early pandemic surveys in the EU and US associated job insecurity, unemployment, financial stressors, and low assets with worse mental health. However, some studies reported no SES–mental health differences. Entrepreneurs/self-employed often faced larger working hour losses and may be at elevated mental health risk, though the broader evidence is mixed. Longstanding literature links lower socioeconomic status to poorer physical and mental health, with financial stress posited as a mediator. Components related to financial stress include low income, income volatility, lack of savings, and consumer debt; savings can buffer shocks, whereas debts and volatility increase uncertainty and stress. Prior work suggests financial stress mediates associations between poverty and health outcomes. This study builds on these literatures by integrating pre-pandemic measures of savings, debts, income, and financial stress to examine mental health changes during COVID-19.

Design and data: Longitudinal analysis using the Dutch LISS panel (initial N = 1114), a probability-based, nationally representative household panel. Three waves spanning pre- and during-pandemic periods were used: t=0 (Apr–Nov 2018), t1 (Dec 2019–Mar 2020), and t=2 (Dec 2020–Mar 2021). Mental health measures were from Nov/Dec 2018, 2019, 2020; financial stress from Apr 2018, Feb 2020, Aug 2020. Methods followed ethical guidelines; informed consent obtained by CentERData. Measures: - Mental health: MHI-5 (subset of SF-36; Cronbach's alpha = 0.87), items recoded so higher scores indicate better mental health (1–6 scale). Assessed annually (Nov/Dec 2018, 2019, 2020). - Financial stress: Psychological Inventory of Financial Scarcity (PIFS; alpha = 0.93), capturing appraisals (money shortage, lack of control) and responses (worry/rumination, short-term focus) on 1–7 scale; higher = more financial stress. Administered Apr 2018, Feb 2020, Aug 2020. - Economic variables: Net monthly household income (2018–2020), aggregated to annual and equivalized by dividing by square root of household size (OECD). Included baseline income and within-person income changes across waves. Savings: liquid assets at 12/31/2018 (category midpoints used if categorical). Debts: consumer credit (excluding mortgages and student loans), total as of 12/31/2018 (category midpoints used if categorical). - Controls: age, gender, education (6 levels; primary school as reference), household composition (no partner/no children; no partner/with children; partner/no children; partner/with children), Big Five subscales (emotional stability, conscientiousness, extraversion; alphas 0.77, 0.89, 0.87). Variance related to controls was parsed from independent variables to assess unique associations. Statistical analysis: - Linear mixed-effects model with random intercepts to model mental health trajectories and their relation to financial stress dynamics. Time included as a predictor and as moderator of financial stress. Unstructured covariance across time points. Numeric variables standardized for interpretability. Mediation model specified with mental health as outcome, financial stress as mediator, and income, savings, debts as independent variables. Equations provided in-text. - Missing data and outliers handled via: (a) Multiple imputation using MCMC with the jomo R package; age and gender included as auxiliary variables; 101 imputations chosen based on fraction of missing information and Von Hippel’s guidance; results pooled with Rubin’s rules. (b) Robust estimation of mixed models using the robustimm package (smoothed Huber loss with iterative reweighting) to mitigate influence of outliers. - Multiple testing: p-values adjusted using Benjamini–Yekutieli FDR control. Mediation (indirect) effects quantified with the distribution-of-the-product method (RMediation approach).

Descriptives: Average mental health remained stable across measurements (MHI-5 means approximately 4.14, 4.13, 4.17), while average financial stress was low and slightly declined by the third measure (means approximately 1.78, 1.76, 1.63 on 1–7). Despite stable means, there was substantial heterogeneity: between t=0 and t=2, mental health increased for 39% of respondents, decreased for 40%, and was unchanged for 21%. Financial stress showed similar heterogeneity. Autocorrelations across waves were moderate-high (mental health ~0.7; financial stress ~0.6–0.8). Regression and mediation: - Financial stress and mental health: Increases in financial stress were associated with decreases in mental health (standardized β = -0.119, t(667)=5.25, p<0.001), supporting Hypothesis 1. The strength of this association did not differ across time points (no significant time × financial stress interactions). Mean mental health did not change over time (no significant time effects). - Predictors of financial stress: Lower pre-pandemic savings predicted higher financial stress during COVID-19 (β = -0.141, t(122)=-3.53, p=0.005), and higher pre-pandemic consumer debts predicted higher financial stress (β = 0.0912, t(240)=3.31, p=0.008), supporting parts of Hypothesis 2. Income level just before the pandemic and income changes during the pandemic were not significant predictors of financial stress or mental health changes after correction. Average financial stress did not change over time. - Mediation: Changes in financial stress mediated the associations between (a) lower savings and (b) higher debts and decreases in mental health (95% CI for the indirect effect approximately [0.00662, 0.0292]), supporting Hypothesis 3. There was no supported indirect effect of income on mental health (95% CI approximately [-0.04, 0.003]). - Covariates: Younger age was associated with greater increases in financial stress (age estimate ~ -0.0928, t=3.11, p=0.013, indicating higher stress among younger respondents). The lowest education group (primary school) experienced more financial stress than higher education groups (education levels 2–6 had negative coefficients relative to primary school). Conscientiousness and emotional stability were negatively associated with financial stress increases (both p<0.001). Of covariates, only emotional stability showed a strong direct association with mental health (β ≈ -0.501, p<0.001, given reverse-scored orientation). Additional mediation analyses indicated that increases in financial stress mediated associations between age, gender, and education level (and personality traits conscientiousness and emotional stability) and mental health changes (several indirect effects with CIs not spanning zero as reported).

The study demonstrates that dynamics in financial stress help explain the substantial heterogeneity in mental health changes observed during the first six months of COVID-19, even as average mental health remained stable. Financial stress increased when households had lower pre-pandemic liquid savings and higher consumer debts, and these stress increases were linked to declines in mental health. Contrary to expectations, income level and income changes did not explain financial stress or mental health changes, possibly due to government income support programs limiting income volatility during the study window. The strength of the financial stress–mental health association did not change across time points. Younger individuals and those with lower education experienced greater increases in financial stress, and personality traits (higher conscientiousness, higher emotional stability) were protective against stress increases. These findings underscore the importance of financial buffers and manageable debt burdens in mitigating stress-related mental health risks during economic shocks. The results suggest that targeting financial stress mechanisms (perceived control, worry/rumination, short-term focus) could be a key lever in mental health interventions during crises.

Main contributions: (1) Using pre- and during-pandemic data, the study shows that changes in financial stress, rather than income level or changes, account for individual differences in mental health changes during early COVID-19. (2) Pre-pandemic financial vulnerability—lower liquid savings and higher consumer debt—predicted higher financial stress during the pandemic, which in turn related to worse mental health. (3) The average stability of mental health masked considerable heterogeneity that financial stress dynamics help explain. Policy and practice implications include: safeguarding financial security for vulnerable households during and after crises; ensuring mental health services reach financially vulnerable groups; incorporating financial counseling/coaching to enhance control and self-efficacy in mental health programs; and promoting buffer savings and avoidance of unnecessary consumer debt to enhance resilience. Future research should examine longer-term trajectories (as mental health effects may peak later), broaden outcomes to specific symptoms (PTSD, insomnia, loneliness), and investigate physical health sequelae. Continued work is needed to disentangle causal pathways and the roles of disease, containment measures, and economic stressors.

- Temporal scope: Data cover the first year/first six months of the pandemic; longer-term mental health consequences may emerge later, so effects may be underestimated. - Causality: Although temporal ordering was leveraged (pre-pandemic savings/debts), the observational design precludes strong causal inferences; unmeasured confounding may remain. - Measurement scope: Focused on general mental health (MHI-5) and overall financial stress; other specific mental health outcomes (e.g., PTSD, insomnia, loneliness) were not analyzed here. - Sample attrition and missingness: Substantial missingness and attrition were addressed via multiple imputation and robust methods, but bias cannot be fully ruled out. - Income dynamics: Limited variability in income during the observation window (possibly due to government supports) may have reduced power to detect income–stress effects.