Have COVID-19 Stimulus Packages Mitigated the Negative Health Impacts of Pandemic-Related Job Losses? A Systematic Review of Global Evidence from the First Year of the Pandemic

Abstract

Social protection can buffer the negative impacts of unemployment on health. Have stimulus packages introduced during the COVID-19 pandemic mitigated potential harms to health from unemployment? We performed a systematic review of the health effects of job loss during the first year of the pandemic. We searched three electronic databases and identified 49 studies for inclusion. Three United States-based studies found that stimulus programs mitigated the impact of job loss on food security and mental health. Furloughs additionally appeared to reduce negative impacts when they were paid. However, despite the implementation of large-scale stimulus packages to reduce economic harms, we observed a clear pattern that job losses were nevertheless significantly associated with negative impacts, particularly on mental health, quality of life, and food security. We also observe suggestive evidence that COVID-related job loss was associated with child maltreatment, worsening dental health, and poor chronic disease outcomes. Overall, although we did find evidence that income-support policies appeared to help protect people from the negative health consequences of pandemic-related job loss, they were not sufficient to fully offset the threats to health. Future research should ascertain how to ensure adequate access to and generosity of social protection programs during epidemics and economic downturns.

In early 2020, the COVID-19 pandemic plunged the global economy into recession, driving unprecedented disruption in global labor markets. By April 2020, 81 percent of the global workforce lived in countries with either required or recommended workplace closures. ¹ Projections suggest that global economic volatility, triggered in part by the pandemic, is likely to persist, even across high-income countries, until 2025. ²

An extensive—and longstanding—body of empirical literature has linked economic downturns, and particularly their impact on employment, to worse health.^3–8 The adverse impact of employment loss during recessions, for example, is clearly documented for mental health outcomes and suicide.^7,9,10 Unemployment is linked to poorer health outcomes as a result of several pathways, including psychosocial (e.g., the stigma of job loss), behavioral (e.g., higher rates of alcohol consumption among the unemployed), and—most importantly—material (lower income as a result of job loss affecting nutrition, heat, housing, etc.). ¹¹

Social protection policies have been found to mediate the impacts of recession-induced unemployment on health. The health effects of the 2008–2009 economic recession, dubbed the Great Recession, for example, have been found to vary according to the strength of welfare state policies, and particularly by state spending on active labor market policies that aim to maintain jobs and quickly re-integrate workers who have lost their jobs back into the workforce.^6,12 Reduced social spending and other austerity measures undertaken during the Great Recession have been associated with a number of negative health outcomes that did not materialize in countries where social spending was maintained.^5,13

However, there has yet to be a systematic review of the health effects of COVID-related unemployment or of the potential health buffering impact of COVID-related social protection policies. Many of the countries that launched austerity measures during the Great Recession discharged large economic stimulus packages in response to early waves of the pandemic. It could thus be that COVID-related stimulus responses are alleviating, or tempering to some degree, what would have been otherwise largely negative health consequences.

Against this backdrop, there is an urgent need to understand how the unemployment consequences of the COVID-19 pandemic are affecting health. This can help countries support those most affected and provide valuable information on the health benefits of stimulus support measures. The aim of our study was therefore to provide a global picture of the individual-level health effects of unemployment during the first year of the COVID-19 pandemic.

Methods

We conducted a systematic review of studies on the individual-level health effects of unemployment during the first year of the global COVID-19 pandemic. The review was registered with PROSPERO (CRD 42021255760) and structured in accordance with recommendations from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (see Appendix 1). We did not publish a protocol for this review.

Search Strategy and Study Selection

We used a combination of title and abstract searches across PubMed, Medline (Ovid), Embase (Ovid), and PsycINFO (Ovid). The databases were searched on April 30, 2021, and we included search terms from previous systematic reviews of unemployment, supplemented with terms relevant to the pandemic such as “furlough,” “layoff,” “coronavirus,” and common variations. The full search strategy is available in Appendix 2. The screening process was conducted by CM, with a 10 percent check by DS, with any conflicts resolved by consensus.

We defined our inclusion criteria a priori in terms of Population, Intervention/Exposure, Comparison, Outcome, and Study Design (Table 1). In summary, we focused on the retrieval of peer-reviewed quantitative studies assessing the health impact of COVID-related unemployment in any country. Our definition of health was broad to capture a wide range of impacts related to physical health, mental health, nutrition (including food security), and changes in health behaviors. To be included, a study must have analyzed employment loss that took place during the COVID-19 pandemic and individual-level health outcomes. Studies were excluded if they measured unemployment that could have preceded the pandemic or by population/area-level. We limited our inclusion of articles to those that were published in English from January 1, 2020–April 30, 2021.

Table 1.

Inclusion Criteria.

Population	Human subjects of working age
Intervention/exposure	Unemployment during the COVID-19 pandemic at the individual level
Comparator	Studies with and without comparator
Outcome	Studies reporting at least one health outcome not limited to morbidity, mortality, prevalence, and incidence of conditions and life expectancy
Study design	Quantitative, peer-reviewed studies undertaken in 2020–2021, from any setting: low-, middle-, and high-income countries

Data Extraction, Quality Appraisal, and Data Synthesis

From included studies, we extracted data about study design, study sample, unemployment measure, comparator, health outcome, statistical result, and key summary text. Extraction was conducted by CM and checked by DS (and consensus developed). In some cases, not all of the findings of a study were relevant to our scope; in such cases, we only included the relevant findings. We appraised the quality of studies using the Joanna Briggs Institute standardized critical appraisal instrument for cross-sectional and longitudinal studies. Study quality was assessed alongside data extraction.

Studies were considered to be too heterogeneous for meta-analysis and were therefore summarized narratively. To manage heterogeneity, we grouped the studies first by health outcome and then by country economic status (i.e., high-income vs. middle- to-low-income, following the World Bank list of 2020).

Results

We identified 1,034 potentially eligible records. After abstract and title screening, 270 were identified for full-text review and 49 met the inclusion criteria for systematic review (Figure 1). A list of reasons for exclusion at the full-text screening stage is provided in Appendix 3.

Figure 1.

PRISMA flowchart.

Appendix 4 summarizes the methodological characteristics of the 49 included studies, whereas a more detailed extraction table can be found in Appendix 5.

Overall Study Characteristics

The vast majority of studies (n = 45) were cross-sectional. Thirty-seven were conducted using data from high-income countries (HICs), while 10 were conducted using data from middle- to low-income countries (LMICs) (See Figure 2). Two were conducted using data from a mix of HICs and LMICs. In total, data was synthesized from 495,935 individuals; however, study samples varied in size from n = 68 to n = 122,133.

Figure 2.

Coverage of countries included in the review.

All studies were undertaken in early stages of the pandemic from late February 2020 to September 2020, with the exception of two studies: one that began in April 2020 and followed up with participants every two weeks until November 2020 ¹⁴ and another that collected data on participants through mid-January 2021. ¹⁵

Most studies (n = 36) focused on COVID-related employment loss within a general population (e.g., aged 18 and older), whereas 13 focused on specific groups, including those aged 55 and older, ¹⁶ those aged 20–35, ¹⁷ post-natal women, ¹⁸ college students, ¹⁹ those taking care of children,^20,21 those earning less than $75,000 a year, ¹⁴ the trans community, ²² and those with specific chronic diseases.^15,23–25 One study focused exclusively on hourly service workers with children in large U.S. cities. ²⁶

In terms of unemployment measures, 33 studies examined employment loss during early stages of the pandemic, four examined furlough, and eight examined both employment loss and furlough, either separately (n = 4) or as a combined variable (n = 4). One study examined job loss due to COVID-19 infection. ²⁵

A number of employment groups were used to compare the health of people who experienced COVID-related unemployment loss with those who did not (three did not use comparators). For example, some studies compared the health of those experiencing COVID-related employment loss with those working from home (WFH) during the pandemic or those working remotely before and during the pandemic. Other studies compared the health of those experiencing COVID-related employment loss simply with those who did not lose their jobs.

Regarding the quality of studies, the majority (n = 32) were rated high. These studies clearly defined their inclusion criteria, described the participants and setting in adequate detail, and adjusted for confounders. The most common weaknesses among lower-quality studies were not identifying confounders and not accounting for the non-representativeness of samples. Quality ratings of included reviews are reported in Appendix 5, with more detailed quality appraisal results provided in Appendix 6.

In terms of health outcomes, 30 studies examined at least one mental health outcome. Eight studies examined at least one change in health behavior, seven studies examined a well-being/quality of life outcome, seven studies examined an outcome related to food insecurity, and five studies examined other health outcomes that did not fit into one of these categorizations. Most articles studied more than one health outcome. We narratively synthesize the results with respect to these categories of health outcomes (i.e., mental health, health behaviors, well-being/quality of life, food insecurity, and other health outcomes), looking first at the evidence in HICs followed by LMICs.

Mental Health

Thirty studies examined mental health and 27 used comparators, 24 of which found worse health among people reporting COVID-related employment loss for at least one mental health outcome. The most commonly examined mental health outcomes were depression, anxiety, stress, and loneliness. We first narratively synthesize studies employing these outcomes and then present studies examining suicide and other mental health outcomes.

Depression, Anxiety, Stress, and Loneliness

Among analyses of mental health outcomes that included comparators, 14 of 15 (depression), 9 of 12 (anxiety), 10 of 12 (stress, including psychological distress), and one of three (loneliness) reported worse outcomes for those who experienced COVID-related employment loss. Analyses were undertaken in 25 studies, 15 of which were based in HICs (United States = 6, Australia = 2, Israel = 2, Canada = 2, Russia = 1, Spain = 1, and United Kingdom = 1) and five of which were based in LMICs (China = 1, Peru = 1, Serbia = 1, South Africa = 1, and Thailand = 1). Three were based on a mix of HICs, while two were based on data from a mix of HICs and LMICs.

With regard to studies in HICs, of the United States-based studies, Abrams and colleagues ¹⁶ (n = 6,264; quality = high) and McDowell and colleagues ²⁷ (n = 2,301; quality = medium) both found that, compared to participants whose work did not change, participants who lost their jobs reported higher symptoms of depression. Abrams and colleagues focused exclusively on individuals aged 55 and older while McDowell included participants aged 18 and older. Both studies also found no significant difference in symptoms of depression between those whose work remained unchanged and those who switched to WFH. McDowell and colleagues reported similar findings for both anxiety and stress, and Abrams and colleagues reported similar findings for loneliness. Abrams and colleagues additionally examined participants who were furloughed, finding higher symptoms of depression and loneliness than those whose work remained unchanged, although the greatest symptoms remained for those who lost their jobs completely. The authors also found job loss to be disproportionately common among Hispanic respondents, those with some college education, and workers in certain occupations (e.g., janitors, barbers, and childcare workers). Using a different comparator group, Jewel and colleagues ²⁸ (n = 1,083; quality = medium) found that individuals in the United States who were no longer working due to COVID-19 were more likely to report symptoms of depression and stress than those who were working remotely before and during the pandemic. However, results were attenuated and significance was lost when additional covariates—related, for example, to financial and food access concerns—were added to the model. The authors did not find an initial association with anxiety. Focusing on custodial grandparents during the pandemic, Wu and colleagues ²¹ found that grandparents who lost their jobs during the pandemic had significantly higher odds of reporting parenting stress than those who did not lose their jobs. Finally, two studies directly examined COVID-related unemployment and the mediating of income-protection policies in the United States. Berkowitz and Basu ²⁹ examined the risk of depression and anxiety among individuals in the United States who lost employment during the pandemic, finding a lower risk among those in households that received unemployment insurance versus those that did not (n = 68,911; quality = high). The authors also found that, in particular, Hispanic individuals and those with less education were both more likely to lose their jobs and more likely to not receive unemployment insurance. Berkowitz and Basu ³⁰ (n = 122,133; quality = high) also found that that receiving unemployment benefits was associated with lower risk for depressive and anxiety symptoms versus not receiving those benefits. However, the authors further observed that risk for depression and anxiety increased when the amount of benefits was reduced.

Two studies were based in Canada. Smith and colleagues ³¹ (n = 3,305; quality = high) found a statistically higher prevalence of depression and anxiety among those who lost their jobs compared to those who switched to working remotely, but not compared to those who remained working on site. By contrast, Pagé and colleagues (n = 2,423; quality = high) found no statistically significant difference in moderate to severe psychological distress among those who experienced COVID-related job loss and those who remained employed. ³²

Two studies were based in Australia. In a study by Dawel and colleagues ³³ (n = 1,296; quality = high), initial univariate tests showed a statistically significant relationship between COVID-related job loss and depression and anxiety. However, job loss did not have a significant independent association with mental health after accounting for financial distress and other covariates. Looking at the trans community in Australia, Zwickl and colleagues ²² argue that a lack of association between COVID-related job loss and depression in their study is likely explained by the introduction of a wage subsidy that provided individuals with payments equivalent to 70 percent of the national median wage (n = 1,019; quality = high).

Turning to the two studies based in Israel, Achdut and Refaeli ¹⁷ (quality = high) examined 389 participants from the ages of 20 to 35 and found worse stress among those who worked before the pandemic but were fired, furloughed, or had to close their business. Mimoun and colleagues (n = 29; quality = low) examined a broader population group (i.e., ages 18–84) and found that those on furlough tended to report higher distress levels in comparison with those who were unemployed prior to the pandemic and those who continued to work full-time. By contrast, no difference was found in comparison with those who continued to work part-time.

In a study based in Russia (n = 23,756; quality = medium), Nekliudov and colleagues ³⁴ found that job loss due to the pandemic was associated with higher scores on an anxiety scale. WFH, by contrast, was associated with lower scores. Not working was also associated with lower scores, but not in a statistically significant way. In a study based in Spain (n = 407; quality = medium), Gunther-Bel and colleagues found a positive association between psychological distress and both COVID-related job loss and unemployment prior to the pandemic. The authors also found negative associations between psychological distress and unaffected employment and WFH. In a study based in the United Kingdom, Ferry and colleagues ³⁵ found that being permanently laid off was associated with psychological distress, even after adjustment for socioeconomic factors. However, they found no evidence of an association between being on furlough/paid leave and mental distress.

Turning to the three studies that used data from a mix of HIC countries, Schnell and Krampe (n = 1,527; quality = low) found that participants in Germany and Austria who were unemployed due to COVID-19 reported higher stress than those who were not. Wei (2020) (n = 112; quality = medium) examined individuals residing in Australia, Canada, the United Kingdom, and the United States and found that COVID-related job loss predicted higher loneliness only for a non-U.S. subset of participants. The authors however, did not find an association between COVID-related job loss and depression or anxiety. Finally, Witteveen and Velthorst ³⁶ included individuals aged 25–64 from the Czech Republic, Germany, Italy, Netherlands, Slovakia, and Spain (n = 1,012; quality = high). The authors found that losing employment during the peak pandemic lockdowns almost doubled the risk of feelings of depression. Moreover, experiencing job loss in the lowest and middle occupational groups yielded a significantly higher chance of reporting feelings of depression as compared to the highest occupational group. The significance tests for the relationship between the job loss and loneliness, however, failed to reject the null hypothesis.

Turning to LMIC-based studies, Antiporta and colleagues ³⁷ (country = Peru; n = 57,446; quality = high) and Posel and colleagues ³⁸ (country = South Africa; n = 2,213; quality = high) both found that, compared to participants who continued to work through the pandemic, participants who lost their job had a higher prevalence of depression and higher scores on depression scales. In their longitudinal analysis, Posel and colleagues (2021) additionally found that the mental health benefits of employment accumulated over time, underscoring the effect of unemployment duration on mental health. This study also found no mental health benefits to being furloughed (on unpaid leave), but that paid leave had a strong and significant positive effect on the mental health of adults. Ruengorn and colleagues ³⁹ (n = 2,303; quality = medium) found that individuals in Thailand who lost their jobs during the pandemic had a higher risk of perceived stress compared to those who maintained their jobs. Guo and colleagues ²³ and Stojanov and colleagues ¹⁸ both focused on specific populations within LMICs. Examining patients with skin diseases in China, Guo and colleagues found a higher risk of depression and anxiety among those who lost their employment due to COVID, versus those who did not (n = 506; quality = high). Stojanov and colleagues ¹⁸ examined 108 post-partum women in Serbia and found higher rates of postnatal depression among women who lost their jobs due to the pandemic versus those who did not (quality = medium).

Finally, two studies were based on data from a mix of HICs and LMICs. Mojtahedi and colleagues ⁴⁰ (n = 723; quality = low) examined participants from the United Kingdom, Ireland, North America, India, Brazil, and other non-specified countries and found that individuals who lost their jobs during the pandemic reported higher levels of depression, anxiety, and stress compared to those WFH. In this study, mental toughness, a psychological concept used to describe a set of enabling individual psychological resources, was found to moderate these relationships to some degree. Solomou and Constantinidou ⁴¹ (n = 1,642; quality = low) examined the Cypriot population residing in Cyprus and abroad (i.e., in North America, Europe, and Greece) and did not find an association between COVID-related job loss and depression, anxiety, or loneliness.

Suicide

Three studies (HIC = 2, LMIC = 1) examined suicide, and two of these found an association with COVID-related job loss. A study based in the United States (n = 500; quality = medium) found that pandemic-related job loss evidenced a significant zero-order correlation with suicide risk; however, pandemic-related employment loss was not uniquely associated with suicide risk when “perceived burdensomeness” (i.e., psychological perceptions of being a burden to others) was included in the model, nor was job-loss significantly associated with perceived burdensomeness, suggesting that relation of job loss to suicide risk is due to a shared association with perceived burdensomeness. ⁴² The second study, looking specifically at the transgender community in Australia, found that unemployment during the pandemic was not statistically associated with thoughts of self-harm or suicide. The authors relate this finding to a wage subsidy that provided individuals with 70 percent of the national median wage (n = 1,019; quality = high). ²² Finally, a study based in China (n = 56,679; quality = high) found a greater risk of suicide ideation among those who experienced COVID-related unemployment than those who did not.

Other Mental Health Outcomes

Three studies examined other mental health outcomes (HIC = 2, LMIC = 1) and two found worse health among those with COVID-related job loss. A longitudinal study based in the United Kingdom (n = 60,449; quality = high) examined changes in common mental disorders (CMD) between those who were furloughed and those who were self-employed with a business that was not affected by the pandemic. The authors found that furlough was not associated with developing a CMD. By contrast, employees who became unemployed or were made redundant or whose work hours were reduced were over two times as likely to develop a CMD compared to the self-employed with a business that was not affected by COVID-19. A United States-based study among 352 hly service workers with children in large cities found that household adult layoff was significantly associated with parental negative mood, but not parental negative sleep quality, child uncooperative behavior, or child worry. ²⁶ Finally, a repeated cross-sectional study based in Brazil examined maternal mental disorder among 577 mothers before and during the pandemic and found no statistically significant difference between those who lost their jobs (either formal or informal) during the pandemic and those who remained employed. ⁴³

Health Behaviors

Eight studies analyzed the relationship between COVID-related job loss and a change in health behavior. Of these, six found a change in at least one health behavior among people reporting COVID-related employment loss. Overall, however, the results were mixed with regards to whether the change was in a health-positive or -negative direction.

Five studies examined COVID-related unemployment in relation to changes in alcohol consumption. All of these studies were undertaken in HICs (Australia = 2, Norway = 1, Belgium = 1, and United Kingdom = 1). One Australia-based study (n = 2,307; quality = medium) found that COVID-related job loss was not associated with a change in alcohol consumption, nor was WFH. ⁴⁴ However, the authors’ dichotomous measure of job loss included not only respondents who lost their jobs, but also those who were furloughed or whose breadwinning partner lost their job or was furloughed. Another Australian study (n = 5,158; quality = high) found that those who had lost their jobs as a result of COVID-19 reported drinking more than those who retained their employment. ⁴⁵ In a study based in Belgium (n = 3,632; quality = medium), those reporting COVID-related unemployment were 36 percent more likely to report that they were drinking more during the pandemic compared to respondents WFH. ⁴⁶ A Norway-based study (n = 25,708; quality = high) found that being temporarily laid off was associated with decreased alcohol consumption compared to those who were not temporarily laid off. ⁴⁷ Economic worries, however, a variable that captured, in part, perceived worry about job loss, was associated with increased alcohol consumption. A United Kingdom-based study (n = 2,777; quality = high) looked at three alcohol consumption measures and found that being furloughed (compared to not being furloughed) was positively associated with increases in the frequency of heavy episodic drinking in lockdown, both among women and men, and with increases in the frequency of drinking, but only among men. By contrast, no significant association was found between furlough and increases in units consumed per drinking occasion in lockdown, neither among women nor among men. ⁴⁸

Two studies examined COVID-related unemployment in relation to changes in smoking behavior. A study based in Italy (n = 6,003; quality = high) found that pandemic-related job loss was associated with smoking reduction but not with smoking cessation. ⁴⁹ In a Belgium-based study (n = 3,632; quality = medium), those reporting COVID-related unemployment were 63 percent more likely to report that they were smoking more during the pandemic compared to respondents working from home. ⁴⁶

Two studies examined COVID-related unemployment in relation to changes in cannabis use. One was based in the Netherlands (n = 120; quality = high) and the other was based in Belgium (n = 3,632; quality = medium).^46,50 Neither found a statistically significant association.

One study, based in the United States (n = 2,303; quality = medium), found that compared to those whose employment remained unchanged, participants who lost their jobs during the pandemic reported greater sitting and screen time than those whose employment remained unchanged, but found no significant group differences for physical activity. ⁵¹

Quality of Life/Well-Being

Seven studies examined COVID-related unemployment in relation to quality of life or well-being. All of these found worse health among those who experienced pandemic-related job loss; however, statistical significance was lost in one United States-based study ²⁸ and in one Australian study ³³ after additional covariates—related, for example, to financial and food access concerns—were taken into account. Of the remaining five studies, two were based in HICs and three were based in LMICs. Ikeda and colleagues ⁵² (n = 7,994; quality = high) found that pandemic-related job loss in Japan was negatively associated with a quality of life utility score. Escudero-Castillo and colleagues ⁵³ (n = 1,050; quality = high) found worse self-perceived well-being among participants in Spain who experienced pandemic-related job loss or furlough than among those who remained in employment.

With respect to studies in LMICs, Guo and colleagues ¹⁵ (n = 135; quality = high) examined participants in China with melanoma and found well-being scores were lower in patients who became unemployed during the pandemic versus those who had not. Scores were also significantly lower among those whose family members became unemployed during the pandemic. In contrast, the difference was not significant between those who were unemployed or retired before the epidemic and those who were unaffected. Similarly, Guo and colleagues ²³ focused on participants in China with skin diseases and found a higher likelihood of impaired quality of life among those reporting pandemic-related unemployment compared to those whose employment was unaffected. Finally, Zomalheto and colleagues ²⁵ (quality = medium) examined 68 patients with rheumatoid arthritis in Egypt and found that quality of life was worse among those who experienced pandemic-related temporary unemployment than those who did not.

Food Insecurity

Six studies examined food insecurity (United States = 5, Brazil = 1) and all found worse health among those who experienced COVID-related unemployment. Of the United States-based studies, Niles and colleagues ⁵⁴ (n = 3,219; quality = high) found the both job loss and furlough were associated with greater odds of being in a household experiencing food insecurity, whereby the odds were slightly higher for those experiencing job loss than those experiencing furlough. Berkowitz and Basu ²⁹ (n = 68,911; quality = high) examined the risk of food insufficiency among individuals in the United States who lost employment during the pandemic, finding a lower risk among those in households that received unemployment insurance versus those that did not. Relatedly, in their longitudinal analysis, Raifman and colleagues ¹⁴ (n = 2,319; quality = high) focus on those in the United States who lost their jobs in households earning less than $75,000 and found that receipt of unemployment insurance was associated with a 4.3 percentage point decrease in food insecurity and a 5.7 percentage decrease in eating less due to financial constraints. This was equivalent to a 35 percent relative reduction in food insecurity from the average of 12.3 percent during the full study period and a 47.9 relative reduction in eating less, from the mean of 11.9 percent. Further, more generous unemployment benefits were associated with greater reductions in food insecurity. Reductions in food insecurity were also greater immediately after receipt of benefits and declined over subsequent periods; the authors note this is likely due to decreased benefits levels that followed the suspension of a weekly $600 supplement program. Berkowitz and Basu ³⁰ (n = 122,133; quality = high) also found that that receiving unemployment benefits was associated with a lower risk for food insufficiency compared to those who lost their jobs but did not receive benefits. The authors further observed that risk for food insufficiency increased when the amount of unemployment benefits was reduced. The final United States-based study (n = 502; quality = high) examined food insecurity among college students at three campuses of a large, diverse, state-funded university in Texas. ¹⁹ The authors found that students who reported being laid off, being furloughed, losing part-time shift work, or experiencing other changes in income as a result of the COVID pandemic had higher odds of food insecurity compared to those whose income was not affected. The odds of reporting food insecurity were greatest among those students who lost part-time shift work followed by those who were laid off, those who experienced other changes in income, and those who were furloughed. Relevantly, the authors note that many college students do not qualify for federal and state safety net programs.

Turning to the one Brazilian study, Rocha and colleagues ⁴³ found an increased likelihood of food insecurity among those who lost a formal job during the pandemic, compared to those who remained employed. An increased risk of food insecurity was also found among those who were not working prior to the pandemic; however, there was no significant difference in food insecurity between those who lost informal employment due to the pandemic and those who remained working.

Other Health Outcomes

Five studies analyzed a measure of health that did not fit into the other categories. One study, based in the United States (n = 342; quality = high), found that parents of 4- to 10-year-olds who lost their jobs were more likely to psychologically maltreat their children than those who did not. This study also found an association between parental job loss and an increased probability of child physical abuse, depending on the manner in which parents reported habitually coping with stressful experiences. ²⁰ A study based in Spain (n = 5,220; quality = medium) found that participants who were taking part in the country's furlough scheme (the Temporary Redundancy Scheme) had poorer sleep quality than those who were not. ⁵⁵ A study based in Japan (n = 25,482; quality = high) found that individuals who lost their jobs or experienced a reduction in work were more likely to report dental pain. ⁵⁶ After adjusting for income change, these differences were only partly attenuated and were found to be mediated by psychological distress and by the postponement of dental care. A study based in Canada (n = 2,423; quality = high) found that job loss was associated with a lower likelihood of reporting worsened pain than being still employed or not in the workforce. ³² By contrast, individuals who worked were more likely to report worsened pain. Finally, Singh and colleagues ²⁴ examined individuals with diabetes or hypertension living in India (n = 1,734; quality = high) and found a worsening of disease symptoms among those who lost their employment during the pandemic compared to those who did not.

Discussion

The aim of our review was to provide a first picture of the health effects of job loss during the first year of the global COVID-19 pandemic. Overall, we found a clear pattern suggesting that the initial COVID-19 economic downturn had a negative impact among those losing their employment, particularly in terms of mental health, well-being, quality of life, and food security. We also observed associations between COVID-related unemployment and physical health outcomes, including increased alcohol consumption, child maltreatment, poor sleep quality, worse dental health, and poorer chronic disease outcomes. While fewer studies examined these outcomes, the breadth of negative health outcomes associated with COVID-related unemployment is suggestive that an individual experiencing job loss in the context of the pandemic could be encountering multiple health-related hardships.

While we did not aim directly to study the mediating impact of social protection measures, three studies included in our review did. All used quasi-experimental mediation studies, all were rated of high quality, and all were based in the United States.^29,30,57 The evidence from these studies suggests that income-support policies can play an important role in offsetting the negative health consequences of COVID-related job loss, particularly with respect to poor mental health and food insecurity. Importantly, these studies also highlight different characteristics of social protection policies that may affect their ability to counter the negative health effects of job loss. First, the generosity of income-replacement factors seems to make a difference, where more generous policies are associated with greater health benefits ⁵⁷ and where reductions in generosity are associated with worsening health. ²⁹ Issues of access and eligibility also appear to be limiting factors, with many of those reporting job loss also reporting that they did not receive benefits. ³⁰

Our review captured multiple types of pandemic-related job loss other than losing employment completely, including unpaid and paid furlough. Examining findings from this perspective also points to the health importance of social protection as, overall, a graded effect was observed, whereby the health impact of COVID-related unemployment was worst among those who lost their employment completely, followed by those who were on unpaid furlough. Paid furlough, by contrast, was often not associated with worse health.

Despite our best efforts to ensure quality and comprehensiveness, our findings, however, should be interpreted in light of some limitations. First, more than 70 percent of the studies synthesized from this review were based on data from HICs, and a third of these were based on data from the United States. Of the studies based on data from LMICs, a third used data from China. The evidence is thus extremely limited in terms of geographical focus, indicating an important gap for future research. We are also aware of a number of additional primary evaluations that have been conducted since we undertook our search in April 2021. In the interest of providing a timely “first look” at the evidence, we decided against repeated searches. Relatedly, all studies included in this review assess COVID-related job loss with respect to early waves of the pandemic and document largely short-term health impacts. It is likely, however, that the health effects of COVID-related job loss will change over time, especially as the socio-political context of countries’ responses to COVID-19 change, but also depending on the duration of an individual's unemployment, where a longer period of unemployment is likely to be associated with a greater deterioration in health.^58,59 Further, the heterogeneity of the study populations and the different ways authors measured pandemic-related job loss and health outcomes limited the comparability of results. While there were fewer studies from LMICs than from HICs, our analysis did not show substantial variation in outcomes between HIC and LMIC settings, at least with respect to the direction and statistical significance of the relationship between job loss and health. We also could not exclude the risk of publication bias against studies reporting negative findings. This could be addressed in future work through an examination of gray literature and government reports, which may be more likely to publish negative findings. Several limitations also arose at the individual study level. The majority of included studies, for example, were cross-sectional. This makes it difficult to account for a “health selection” hypothesis that suggests that unhealthy people are more likely to lose their jobs than those who are healthy. While this effect might explain some part of the relationship between COVID-related unemployment and health, the majority of unemployment was undoubtedly a result of pandemic-related lockdowns and can be treated as exogenous to the individual. Further, while some studies utilized a distinct comparator, many treated the non-unemployed as a standard comparator group. This is problematic as all individuals were affected by the pandemic and those who continued to work (e.g., in health care and other essential sectors) are likely to have done so in very precarious conditions. Future work should account for this complexity in terms of both study design and data interpretation. Most studies also relied on internet-based survey methods, which would have biased samples against people with limited access to the internet. This type of selection bias is particularly important in countries where access to the internet varies significantly and systematically by socioeconomic status and location. In these settings, effect sizes are likely to be underestimations. Finally, relatively few studies examined the differential health effects of COVID-related job loss. This represents a significant gap in the evidence base as it is widely understood that the economic fallout of the pandemic is disproportionately affecting groups that suffer from poorer health, have fewer health protective resources, or were less able to work from home, such as women; those in precarious employment; ethnic minorities; older workers; and those within lower socioeconomic strata.^60–62

Notwithstanding these limitations, our findings provide a first look at how COVID-related unemployment has affected the health of individuals across the world. Our findings are consistent with previous work documenting the negative health impact of recession-induced unemployment and studies that document the buffering role of income-support policies. Future work should focus on understanding more precisely the moderating role of social protection and which policy characteristics provide the greatest health benefit, especially with respect to vulnerable populations and a wider set of country and sociopolitical settings. In policy debates about designing and modifying both COVID-related economic support policies and emergency/pandemic preparedness plans, it will nevertheless be important to remember that social protection policies can provide vital health benefits.

Author Biographies

Courtney McNamara (PhD) is a Lecturer of Public Health within the Population Health Sciences Institute at Newcastle University. Her research focuses on the global political economy of health. She has written extensively on the health impacts of international trade; the social determinants of health; and the links between COVID-19, labor markets, and health. She currently leads a large research project funded by the Norwegian Research Council. This project asks whether trade shocks cause health to suffer and investigates which policies could help prevent any potential harms. She is an affiliated researcher at the Centre for Global Health Inequalities Research (CHAIN) in Trondheim, Norway, and Chair of the Trade and Health Forum within the American Public Health Association.

Virginia Kotzias (BS, BA, MPP) is a Ph.D. candidate in the Department of Sociology and Political Science at the Norwegian University of Science and Technology (NTNU). Prior to NTNU, she worked as a health policy analyst for the RAND Corporation and as a public health administrator for Maryland Medicaid. Her research portfolio includes qualitative and mixed-methods work in patient experience of care, health services research, mental and behavioral health, telemedicine, emergency department and acute care, and long-term services and supports. The work in her doctoral studies focuses on the intersection of economic and social factors that affect maternal and child health.

Clare Bambra (PhD, FAcSS) is a Professor of Public Health. Her research focuses on understanding and reducing health inequalities. She is a National Institute for Health and Care Research (NIHR) Senior Investigator and has leadership roles in several large research collaborations: Centre for Global Health Inequalities Research (CHAIN), NIHR Policy Research Unit in Behavioural Science, SIPHER: UKPRP consortium on Systems Science in Public Health, NIHR Applied Research Collaboration - North East and North Cumbria (NE-NC-ARC), NIHR School for Public Health Research, and Fuse: the Centre for Translational Research in Public Health. She also leads a Wellcome Trust grant investigating regional health inequalities in England and co-leads an NIHR grant on the mental health impacts of universal credit. She has extensively analyzed COVID-19 and health inequalities, she led the independent SAGE report on COVID-19 and health inequalities and the NHSA report on COVID-19 in the North, and she co-leads a Health Foundation Grant on COVID-19 and health inequalities. She is a member of the Institute for Public Policy Research's Commission on Health and Prosperity and works regularly with various policy and practice organizations, including the National Health Service, Local Authorities, Office for Health Improvement and Disparities, the European Union, and the World Health Organization. She has published widely, including several award-winning books, and her research is regularly covered by the media.

Ronald Labonté is Professor Emeritus and former Distinguished Research Chair in Globalization and Health Equity in the School of Public Health and Epidemiology, University of Ottawa. For the past 25 years, his research has focused on the health equity impacts of diverse globalization processes, many of which form the content of his recent book, Health Equity in a Globalizing Era (Oxford University Press), awarded Book of the Year by the British Medical Association in 2021. He is co-Editor-in-Chief of the BMC journal Globalization and Health, active with the People's Health Movement (PHM), a frequent contributor to its flagship publication Global Health Watch, a co-editor of its 6th edition, and a member of the PHM Steering Council.

David Stuckler, PhD, MPH, Hon.MFPH, FRSA, is a Professor of Social and Political Sciences at the University of Bocconi in Milan. He directs a World Health Organization Collaborating Centre on Social Protection and Governance for health. He has written over 300 peer-reviewed scientific articles on global health in The Lancet, British Medical Journal, and Nature, in addition to other major journals. His book about the global chronic disease epidemic, Sick Societies, was published by Oxford University Press in 2011. He is also an author of The Body Economic, published by Penguin Press in 2013 and translated into over 10 languages. His work has been featured on covers of The New York Times and The Economist, among other venues. Foreign Policy named him one of the top 100 global thinkers of 2013.