Sage Journals: Discover world-class research

Abstract

Successive social shocks have reinforced unemployment as a persistent issue, yet its impact on prosocial engagement remains underexplored. This study examines the relationship between job loss and prosocial behaviors, specifically volunteering and donating participation, hypothesizing that short-term unemployment acts as a negative personal shock with heterogeneous effects depending on individuals’ economic resources. Using nationally representative survey data from the United States collected during the COVID-19 pandemic, we employ propensity score weighting enhanced by a machine learning approach to address potential confounders. Our findings indicate that unemployment is associated with reduced formal volunteering participation among lower-income individuals and decreased participation in donating among higher-income individuals. However, we find no significant relationship between unemployment and informal volunteering. Furthermore, our analysis suggests that conventional methods, which fail to adequately control for confounders, tend to underestimate the negative impact of unemployment on prosocial behaviors.

Keywords

unemployment prosocial behaviors volunteering donating COVID-19

Introduction

Over the last few decades, unemployment has increasingly become a critical social issue. A series of societal shocks, following each other closely, have resulted in massive layoffs. The financial crisis, for instance, pushed many workers out of their jobs. Foreign trade shocks diminished positions, particularly in the manufacturing sector, and rapid technological shifts transformed the labor market, displacing or relocating workers from their previous roles (Acemoglu & Restrepo, 2019; Autor et al., 2022; Clark, 2014). Then came the COVID-19 pandemic which brought the economy to a near standstill, forcing many workers out of their jobs or on furlough (Beland et al., 2023). As the concept of lifetime work or permanent employment fades, attention has shifted toward how individuals cope with unemployment rather than preventing it outright.

A voluminous body of literature has highlighted the detrimental effects of unemployment on individual well-being across multiple dimensions. It is now well-documented that unemployment adversely affects physical and mental health, promotes social exclusion, undermines self-esteem, and threatens material well-being (Andersen, 2009; Blustein, 2019; Brand, 2015; Penny & Finnegan, 2019). More recent studies suggest that unemployment may also contribute to the recent decline in fertility rates and delays in first childbirth (Miettinen & Jalovaara, 2020).

Despite extensive research, the relationship between unemployment and prosocial engagement remains theoretically contested and empirically underexplored. While the effects of job loss on individual well-being have been widely studied, its broader social implications, particularly how it shapes civic and prosocial behavior, have received comparatively less attention. Existing theoretical perspectives offer conflicting predictions. On one hand, some scholars argue that job displacement may enhance prosocial engagement, as individuals gain more discretionary time, face lower opportunity cost of prosocial behaviors, and seek to rebuild self-esteem or gain different forms of resources through prosocial engagement. On the other hand, opposing views suggest that the psychological and economic strain of unemployment compels individuals to withdraw from prosocial engagement, focusing instead on personal recovery and job-seeking efforts. These competing hypotheses persist, yet rigorous empirical evidence capable of adjudicating between them remains limited (Brand, 2015; Bundi & Freitag, 2020; Holstein & Qvist, 2025; Lim & Sander, 2013; Piatak, 2016).

This paper explores the relationship between unemployment during the COVID-19 pandemic and prosocial engagements, such as volunteering and donating, in the United States. We conceptualize unemployment as a significant negative shock that affects individuals both economically and socially and investigate if and how it affects prosocial behaviors. Moreover, recognizing that there exist costs related to prosocial behaviors, in terms of time and money, and that relative costs and benefits of these behaviors may significantly vary among individuals with different economic resources, we explore discrepancies in the association of unemployment and prosocial engagement across different economic groups.

Our contributions to the literature are fourfold. First, in terms of its short-term outcomes, we provide evidence that unemployment should be viewed as a negative shock to prosocial engagements, contrasting with its portrayal in previous studies and popular media, which often suggest that unemployed individuals are more likely to engage in social activities like volunteering during periods of joblessness (“At least I can do something”) (Baines & Hardill, 2008; Bosman, 2009). Moreover, our evidence contributes to adjudicating between competing theoretical claims, offering a more nuanced understanding of how economic displacement can influence prosocial behaviors. Second, by employing a machine learning approach to mitigate selection bias between those unemployed and those not laid off, we emphasize the importance of methodological rigor in accurately quantifying this negative impact. Without such measures, the effects of unemployment could easily be confounded by socioeconomic factors or demographic attributes like age and gender, as suggested by life course theories (Strauß, 2008). Our analysis indicates that relying solely on parametric controls for extensive demographic features in a linear regression model may underestimate the negative short-term impact of unemployment on prosocial engagement. Third, we draw on data from a nationally representative sample regarding both before and during the COVID-19 pandemic, although our theoretical and empirical contributions are not confined to this specific period. Finally, based on our results, we provide theoretical and practical implications for this field of inquiry.

Background and Previous Studies

Unemployment and Formal Volunteering

Volunteering is often regarded as unpaid work (Jenner, 1981), and substantial research has explored its relationship with unemployment, especially focusing on how volunteering may facilitate (re)entry into the labor market (Hackl et al., 2007; Handy et al., 2010; Spera et al., 2015). Various mechanisms have been proposed through which volunteering can boost employment prospects. For instance, Spera et al. (2015) suggest three pathways: increased social capital, enhanced human capital, and signaling of personal characteristics. First, it is well-documented that leveraging networks can provide essential information and opportunities for job seekers (Calvo-Armengol & Jackson, 2004; Granovetter, 1974), suggesting that ties formed through volunteer activities can serve as valuable social capital aiding labor market re-entry. In addition, the skills and knowledge required in many formal volunteering roles overlap with those needed in the workplace, allowing volunteers to accumulate human capital that benefits their employment prospects. As Menchik and Weisbrod (1987, p. 162) noted, volunteering can positively impact employment chances and future earnings “by providing work experience and providing potentially valuable contacts.” While this human capital theory stresses that substantive development through volunteering invites better economic opportunities, those who put more weight on the signaling effect of volunteering focuses on the history of volunteering. They argue that such record can address the information asymmetry between employers and potential employees, acting as a signal of desirable personal attributes and thereby enhancing one’s résumé (Spence, 1973). In a competitive job market, these prosocial engagements can imply that applicants are “good organizational citizens” (Handy et al., 2010, p.500) who can put the organization’s interest before their own interests. Considering all these salutary effects of volunteering, the role of volunteering in facilitating re-employment is well-recognized so much so that it has been accorded the status of “well-known fact.” It seems to be universally acknowledged that “Everyone seems to ‘know’ that volunteer experience enhances one’s resume and leads to improved labour market opportunities” (Day & Devlin, 1998, p. 1180).

In contrast to the consensus that volunteering can be beneficial for solving unemployment, our understanding of how unemployment impacts volunteering remains ambiguous and has been consistently highlighted as under-researched (Lim & Sander, 2013; Piatak, 2016). As Bundi and Freitag (2020) noted, “Social scientists have long studied various forms of volunteering behaviour, but few have directly examined how economic hardship affects volunteering activities or civic engagement in general” (p. 293).

Two competing perspectives dominate this debate. One suggests that job displacement can increase volunteer participation. This view holds that unemployment lowers opportunity costs and increases available time, aligning the volunteering patterns of the unemployed with those of retirees (Piatak, 2016). Moreover, because unemployment is often accompanied by psychological distress (Lucas et al., 2004), volunteering may serve as a mechanism for preserving self-worth—what Penny and Finnegan (2019, p. 155) describe as “access to self-respect.” Beyond its immediate psychological benefits, volunteering may also reshape attitudes toward both paid and unpaid labor. Under dominant cultural norms that equate self-worth with employment, joblessness can create cognitive stress, as suggested by cognitive dissonance theory (Harmon-Jones et al., 2018). To resolve this tension, individuals may develop a more positive orientation toward unpaid work of volunteering, increasing participation as a means of restoring self-esteem. A similar conclusion can be drawn from social exchange theory, which conceptualizes social behavior as an exchange of material and non-material goods (Homans, 1958; Schlosser & Zinni, 2011). From this perspective, volunteering is not merely a prosocial act but also a strategic response to economic displacement. Individuals invest their time and effort with the implicit expectation of securing material benefits or social capital that may facilitate reentry into the labor market. Volunteers displaced from their workplaces may view volunteering as a means to fulfill valued roles, develop professional networks, acquire new skills, or even obtain direct job referrals (Piatak, 2016).

Conversely, an alternative theoretical perspective posits that unemployment leads to a decline in volunteering. The dominant status model (Smith, 1983) argues that individuals occupying culturally and institutionally valued positions are more likely to engage in civic life including prosocial engagement. Unemployment, by diminishing social standing, may weaken this engagement. A parallel line of reasoning underscores the role of situational factors in shaping prosocial behavior. Smith (1994) highlighted how such behavior is often driven by immediate contextual cues, what sociologists describe as symbolic interaction and psychologists as cognitive assessment of situations. Among these situational factors, social contacts play a pivotal role. For instance, simply being asked to volunteer can be one of the strongest predictors of participation (Smith, 1994). Similarly, research in network science suggests that the diffusion of social behaviors can occur most readily in environments where interaction is recurring (Centola, 2018). Given that many personal networks are workplace-based, job displacement reduces opportunities for interaction and, in turn, limits exposure to volunteering (Musick & Wilson, 2008). From this perspective, unemployment is not merely an economic disruption, but it undermines the very mechanisms that sustain or initiate volunteer participation.

Moreover, the idea that those who are displaced from jobs have lower opportunity costs and greater time to engage in “good deeds” may be overly simplistic. It’s true that every individual has fixed and limited time to allocate to different activities, and employment implies that a significant portion of time is dedicated to workplaces. This explains why volunteer recruitment often strategically targets retired people or college students. However, job displacement does not necessarily lead individuals to allocate time to non-market activities (Taniguchi, 2006) and if individuals are concerned about making a living due to job loss, they will likely dedicate their time to finding new employment. Expecting the unemployed to be concerned about the welfare of others is a naïve extrapolation (Putnam, 2000).

Building on these insights, we conceptualize unemployment as a negative personal shock—one that affects individuals economically, emotionally, and socially (Bundi & Freitag, 2020). While unemployed individuals may engage in volunteering to seek social and psychological support, extensive research consistently reports declines in well-being following job loss (Brand, 2015; Lucas et al., 2004; Ruhm, 1991). Thus, any positive correlation between unemployment and volunteering is more likely to emerge in the long term, once individuals adjust to prolonged joblessness and the initial trauma of job loss subsides (Munyon et al., 2019).

During times of widespread societal shock or disaster communal and societal solidarity can temporarily override negative individual experiences, driving collective engagement in volunteering. For example, Lim and Laurence (2015) found that during the 2008 global financial crisis, personal experiences of economic hardship, such as unemployment, had less predictive power for volunteering behaviors than community-level factors like civic infrastructure and trust. Nevertheless, in the context of the COVID-19 pandemic, we expect the solidarity channel to have been less salient than in other disaster-related volunteerism contexts. Unlike previous social shocks characterized by widespread layoffs, the pandemic constrained volunteering opportunities not only through individual willingness but also through structural limitations—most notably, the reduced availability of volunteer roles due to public health restrictions such as social distancing mandates. For example, Lebenbaum et al. (2024) reported a nearly 50% decline in volunteer position postings in Canada during the pandemic. Similarly, Cnaan et al. (2022) found that approximately one-quarter of formal volunteers in the United States ceased their participation following the outbreak.

Finally, we also challenge the simplistic assumption that unemployed individuals face lower opportunity costs for volunteering. In this context, opportunity costs extend beyond forgone income to include time, cognitive load, and mental resources. Individuals without financial reserves or household support may experience higher relative costs, not only in terms of direct financial constraints but also in the time and effort required for job searching. As Lim et al. (2016) argued, job searching is time-intensive and cognitively demanding, often leading to burnout that may further diminish engagement in prosocial activities. Therefore, while individuals with sufficient financial reserves may have the economic and cognitive capacity to sustain their prosocial engagement, assuming that those without such resources face lower opportunity costs neglects the significant non-monetary burdens associated with unemployment. Based on these insights, we construct our first set of hypotheses as follows:

Hypothesis 1: Individuals who were recently unemployed will have a lower likelihood of engaging in formal volunteering compared with those who remained employed.

- Hypothesis 1-A: Among individuals with higher household income, those who were recently unemployed will not exhibit a lower likelihood of formal volunteering compared with those who remained employed.

- Hypothesis 1-B: Among individuals with lower household income, those who were recently unemployed will exhibit a lower likelihood of formal volunteering compared with those who remained employed.

Unemployment and Informal Volunteering

It is important to distinguish between formal and informal volunteering (Cnaan & Amrofell, 1994; Taniguchi, 2012). While both forms involve engagement in social, charitable, or civic activities, they differ fundamentally in structure and reciprocity. Formal volunteering is institutionally embedded, occurring within organizations such as schools, churches, hospitals, and shelters. Informal volunteering, by contrast, operates outside of formal structures, often within personal networks, encompassing activities like babysitting, cooking for a neighbor, coaching a team, or providing emotional support. As Hodgkinson et al. (1992) noted, “Formal volunteering involves regular work with an organization; informal volunteering involves helping neighbors or organizations on an ad hoc basis, such as babysitting for free or baking cookies for a school fair” (p. 2).

Informal volunteering, being more deeply embedded in routine social interactions, has a greater potential for reciprocity than its formal counterpart, where assistance is often unidirectional. Empirical research supports this distinction. Ramaekers et al. (2024) demonstrate that individuals are more likely to engage in informal volunteering—or “informal helping”—when they anticipate reciprocal support. This suggests that for individuals experiencing economic hardship, the costs of informal volunteering may be lower than those of formal volunteering, particularly when participation yields mutual benefits (Baines & Hardill, 2008). Desmond’s (2012) ethnographic study of poverty-stricken communities offers a parallel insight. Although he does not explicitly use the term informal volunteering, his concept of “disposable ties” closely aligns with its core principles. Disposable ties refer to relations characterized by “accelerated and simulated intimacy, a high amount of physical copresence (time spent together), reciprocal or semireciprocal resource exchange, and (usually) a relatively short life span” (Desmond, 2012, p. 1311). At the heart of both concepts lies informal, mutual helping of people.

Building on the conceptualization of unemployment as a negative personal shock, it is reasonable to expect that such disruptions heighten individuals’ reliance on mutual support systems, including informal volunteering. This perspective is consistent with research suggesting that economic hardship strengthens reciprocal aid networks (Small & Gose, 2020). Observations from the COVID-19 pandemic further support this view: receiving informal support consistently emerged as the strongest predictor of individuals’ engagement in informal volunteering (Cho et al., 2022; Haller et al., 2022; Sin et al., 2021).

Moreover, just as financial resources moderate the impact of short-term unemployment on formal volunteering, they are also likely to shape engagement in informal volunteering. Echoing Desmond’s (2012) concept of disposable ties as “network-based survival strategies” (p. 1331), if such ties function as mechanisms for navigating personal hardship, then individuals facing greater financial precarity should be more reliant on informal volunteering as a buffer compared with those with greater economic stability. Meanwhile, those with ample financial reserves may feel less need to actively reinforce mutual support networks following an unemployment shock. This reasoning leads to our second set of hypotheses:

Hypothesis 2: Individuals who were recently unemployed will have a higher likelihood of engaging in informal volunteering compared with those who remained employed.

- Hypothesis 2-A: Among individuals with higher household income, those who were recently unemployed will not exhibit a higher likelihood of informal volunteering compared with those who remained employed.

- Hypothesis 2-B: Among individuals with lower household income, those who were recently unemployed will exhibit a higher likelihood of informal volunteering compared with those who remained employed.

Unemployment and Donating

While volunteering can be understood as the charitable giving of time, donating represents the charitable giving of money or material resources. In the United States, research on charitable donations primarily focuses on formal giving—monetary contributions to charitable organizations and nonprofits—rather than informal donations, which include non-monetary support for friends, family, or neighbors (Reed & Selbee, 2000). For the purposes of this study, we focus exclusively on formal donations, referring to them broadly as “donating” or “donations.”

As discussed earlier, during the COVID-19 pandemic, the volunteering landscape was notably constrained: even aside from individuals’ willingness to volunteer, the number of available volunteer positions shrank dramatically compared with normal times. In contrast, opportunities for donating remained relatively resilient to external shocks. Although in-person fundraising events became difficult to organize, the use of digital platforms for donating expanded significantly. Using a time-series model based on pre-COVID-19 data for forecasting and normalizing the daily number of campaigns by the mean observed up to the end of February 2020, Martin and Schlereth (2024) reported that the daily number of fundraising campaigns on social media more than quadrupled by late March 2020. This pattern suggests that, in explaining participation in donating activities, individual-level factors—pertaining to the supply rather than the demand side of giving—carry greater explanatory weight.

Bekkers and Wiepking (2011), in their review of the literature, identify eight mechanisms that drive charitable giving: awareness of need, solicitation, costs and benefits, altruism, reputation, psychological benefits, values, and efficacy. Among these, psychological benefits, values, and efficacy can be understood as non-material returns to donating, paralleling the social exchange perspective on formal volunteering. However, there is a crucial distinction. Whereas the benefits of volunteering, such as skill-building and information from social networks, can be more easily converted into tangible resources, the returns to donating remain less readily transferable. For example, while charitable giving may enhance one’s reputation or social standing within a community,¹ it is unlikely to provide the material buffer necessary to mitigate the economic shock of unemployment. Instead, the most salient factor influencing donation behavior following job loss is likely to be its direct cost.

This brings us back to economic resources as a moderating factor. For individuals with sufficient financial reserves, it is less clear whether the cost of donating would surpass a critical threshold that prompts withdrawal from charitable giving. Empirical research offers mixed findings. While Lee and Farrell (2003) found statistically significant differences in donation patterns between the employed and unemployed, Bennett (2021) reported that, despite unemployment rising to 8.6% in 2020, average charitable donations in the United States increased by 2.2% between 2020 and 2021. Bennett (2018) also suggested that recently unemployed individuals do not immediately reallocate or reduce their monetary donations.

Nevertheless, for individuals with scarce financial resources, it is reasonable to expect that unemployment would lead to a reevaluation of donation behaviors. Unlike volunteering, where the costs are often implicit, donating directly reduces financial capital, making its economic burden immediately visible. For those already engaged in donating, job loss may prompt withdrawal from charitable giving, whereas for those not previously engaged, it may deter initiation altogether. Based on this reasoning, we construct the final set of hypotheses as follows. For the general population and individuals with greater economic resources, we follow previous empirical studies that report low sensitivity in withdrawing charitable giving immediately after economic shocks. On the other hand, for individuals with scarce economic resources, we expect that unemployment will likely discourage donation participation.

Hypothesis 3: There will be no significant difference in the likelihood of donating between individuals who were recently unemployed and those who remained employed.

- Hypothesis 3-A: Among individuals with higher household income, there will be no significant difference in the likelihood of donating between individuals who were recently unemployed and those who remained employed.

- Hypothesis 3-B: Among individuals with lower household income, those who were recently unemployed will be less likely to donate compared with those who remained employed.

Method

Data

Data collection for our study was conducted in collaboration with the survey company Social Science Research Solutions (SSRS), with the questionnaire being developed, administered, and checked from January 1 to April 18, 2022. Our respondents were sourced from the SSRS Opinion Panel, a nationally representative, probability-based web panel that allows findings to be statistically projected to the general adult population. Participants were recruited either through mailed invitations to a random sample from an address-based sample or via a dual-frame random digit dial sample.

The questionnaire underwent pre-testing, which included 11 cognitive interviews to refine questions based on participant feedback concerning their relevance and clarity. After these pretests, the survey was finalized and made available in both English and Spanish. To enhance participation, panelists were offered a modest incentive. Weighting procedures were applied to adjust for non-response and other potential biases, ensuring that our sample’s demographic profile accurately reflects that of the U.S. adult population. The process involved applying base weights, adjusting for non-internet propensity, and calibrating to national population parameters. This approach ensured that our data are representative across key demographic variables, closely aligning with benchmarks from the Current Population Survey, the Census Planning Database, and the Pew Research Center’s National Public Opinion Reference Survey. Details of key survey questions used in our analysis are provided in the Online Supplementary Information.

Empirical Strategy

Although our goal in this paper does not extend to establishing strict causal claims, it is of utmost importance to minimize bias as much as possible. Results from previous analyses that do not effectively address the confounding factors which surround unemployment and prosocial engagement outcomes should not provide strong evidence. Yet merely adding these attributes as controls is insufficient to solve the issue. This is because standard parametric adjustments through regression often fall short when a model is not accurately specified (Rubin, 1979), potentially leading to results that may not truly reflect the effect of being laid off on prosocial behaviors but rather other individual attributes confounding the results.

To address this challenge, we utilized propensity score weighting (PSW), a methodology extensively applied to observational data (González Canché, 2014; Ridgeway et al., 2015). While many studies rely on parametric logistic regression to estimate propensity scores, we opted for a gradient boosted model (GBM), a machine learning approach that effectively captures higher-order terms and nonlinear relationships across a broad set of covariates (McCaffrey et al., 2004; Ridgeway et al., 2023). We direct readers interested in the methodological details of our GBM approach to the Online Supplementary Information.

Although this PSW approach does not directly adjust for unobservable attributes, it significantly improves observable balance between the groups. Table 1 lists the variables used for the regression tree and the main regression analysis, serving also as descriptive statistics for the two groups, reflecting sampling weights. The descriptive statistics for the total sample², which also includes individuals who were not employed prior to the pandemic, are available in Supplementary Table 1.

Table 1.

Balance of the Treatment and Comparison Groups.

Variable	Laid off	Not laid off: PSW	Not laid off: raw	Std effect size
Age	40.71	40.58	42.915	0.01
Gender_ Female	0.491	0.507	0.433	−0.032
Gender_ Male	0.491	0.476	0.534	0.031
Race_ White	0.524	0.594	0.611	−0.14
Race_ Asian	0.041	0.042	0.069	−0.003
Race_ Black	0.148	0.133	0.136	0.043
Race_ Hispanic	0.194	0.192	0.144	0.006
Race_ Other	0.054	0.021	0.021	0.146
Marital_ Single	0.455	0.463	0.353	−0.015
Marital_ Divorced	0.103	0.088	0.075	0.049
Marital_ Married	0.392	0.4	0.53	−0.015
Marital_ Separated	0.032	0.037	0.022	−0.027
Marital_ Widowed	0.018	0.013	0.019	0.036
Religious_ Attended worship less than once a month or none	0.668	0.66	0.593	0.018
Religious_ Attended worship once a month or more	0.332	0.34	0.407	−0.018
Family experiences_ Never	0.247	0.248	0.258	−0.003
Family experiences_ Sometime	0.476	0.488	0.534	−0.024
Family experiences_ Frequently	0.274	0.262	0.206	0.028
Parent_ No	0.638	0.606	0.642	0.066
Parent_ Yes	0.334	0.377	0.354	−0.091
Education_ High school or less	0.435	0.411	0.316	0.047
Education_ College or more	0.565	0.589	0.684	−0.047
Financial_ Didn’t have enough to meet basic expenses	0.038	0.031	0.025	0.038
Financial_ Could just meet basic expenses	0.254	0.21	0.123	0.1
Financial_ Met basic expenses with a little left over for extras	0.439	0.442	0.424	−0.005
Financial_ Lived comfortably	0.269	0.317	0.428	−0.108
Employment_ Employed full time	0.693	0.749	0.81	−0.121
Employment_ Employed part time	0.307	0.251	0.19	0.121
Volunteering before pandemic_ No	0.463	0.473	0.523	−0.021
Volunteering before pandemic_ Yes	0.537	0.527	0.477	0.021
Donating before pandemic_ No	0.315	0.313	0.335	0.004
Donating before pandemic_ Yes	0.685	0.687	0.665	−0.004
Informal volunteering before pandemic_ No	0.22	0.224	0.267	−0.011
Informal volunteering before pandemic_ Yes	0.78	0.776	0.733	0.011
Sample size (without sampling and PSW adjusted weights)	308	913
Effective sample size	171.07	370.02

We incorporated a broad array of covariates acknowledged for their association with both the key variable of unemployment during the pandemic and outcomes related to prosocial behaviors (Brodeur et al., 2021; Chambré, 2020; Couch et al., 2020; Musick & Wilson, 2008; Smith, 1994; Yasenov, 2020). Regarding our analytical approach, the lasso method selectively retained the most relevant main effects and interactions among the covariates, assigning weights to each unit accordingly. Despite allowing for up to 10,000 iterations, we identified an optimal model—showcasing the lowest balance measure (mean standardized effect size)—at the 2,496th iteration, as depicted in Figure 1. The figure also suggests no further iterations would have improved the model significantly.

Figure 1.

Balance Measure and GBM Iterations.

Table 1 delineates the mean values for these variables across both the laid-off group and those not laid off. For individuals not experiencing layoff, namely the control group, we distinguished between PSW-adjusted samples (“Not laid off: PSW”) and the raw control group with only sampling weights applied (“Not laid off: raw”). Before PSW, individuals who were not laid off during the pandemic were older, more likely to be male and White, less likely to be single, more religious, more educated, in better financial situation, and less likely to be engaging in formal volunteering before the pandemic. However, as a result of PSW, we can notice PSW effectively reduced discrepancies across variables examined. The standardized effect size column, calculated as the difference between the means of the treatment and control groups divided by the standard deviation of the treatment group, demonstrated no values exceeding 0.15 in absolute terms. This indicates minimal differences in means between the groups, underscoring the effectiveness of our methodological approach in achieving balance.

In this study, the pandemic served as a negative shock for both groups. As a result, the impact of COVID-19 on prosocial behaviors would be expected to be largely uniform across both groups, unless there is strong evidence suggesting that social distancing measures or the risk of contagion disproportionately affected opportunities for prosocial engagement between them. Even if some residual discrepancies existed, the flexible control of extensive covariate sets through GBM-based PSW would likely account for and balance these differences.

Visual assessments of PSW were also conducted. In Appendix, Figure A1 illustrates the concept of common support, revealing varying degrees of skewness between the two groups yet a substantial overlap in the density distribution of propensity scores. At the same time, contrary to the methods of propensity score matching and stratification, where significant overlap is crucial for effective balance, weighting is recognized for its resilience to minimal overlap areas (Ridgeway et al., 2023; Whatley & González Canché, 2022). Figure A2, with closed circles denoting statistically significant differences in standardized effect sizes between the treatment and control groups, demonstrates a significant reduction in these differences.

Using the weights from propensity scores, we employ linear probability model as well as logistic regression to examine the effect of unemployment to the participation of formal volunteering, donating, and informal volunteering. We used the same set of covariates for PSW and for control variables in our main regression analysis. Because we have balanced the distribution of characteristics for the treatment group and the control group, having these control variables in the regression analysis is not necessary as the weights already account for differences in covariates. However, having these control variables present several advantages in that this addition can control for remaining unbalances which are relatively small. More importantly for our case, we started with a relatively moderate number of unweighted sample size (n = 1,221, treatment group = 308, control group = 913), applied sampling weights, and adjusted for PSW (effective sample size = 541, treatment group ESS = 171, control group ESS = 370). This is related to the fact that applying weights yield greater sampling variance, which would likely to lead to smaller statistical power. To address this, as with randomized trial cases, we included covariates as it can improve the precision of estimates.

Finally, a key variable in our model is the unemployment indicator. As detailed in the Online Supplementary Information, respondents were asked whether they lost their job, were laid off, or were placed on furlough since the onset of the COVID-19 pandemic. Consequently, our measure captures both permanently displaced individuals and those on temporary furlough. Because our study focuses on the effects of job loss rather than temporary employment suspension, this classification may introduce measurement error. Two mechanisms could bias our estimates. First, classical measurement error occurs if the misclassification of furloughed individuals as unemployed is uncorrelated with their true employment status. This would introduce attenuation bias, leading to an underestimation of the true effect size. Second, systematic misclassification arises if furloughed individuals differ from the truly unemployed in individual characteristics and prosocial engagement. Echoing with how we conceptualize short-term unemployment as a negative personal shock, furloughed individuals likely experience a less severe disruption than those fully displaced. As a result, they may be less likely to withdraw from costly prosocial behaviors (i.e., formal volunteering and charitable donations) and less likely to increase reliance on reciprocal social support (i.e., informal volunteering). In this scenario as well, the misclassification would bias our estimates downward, attenuating the observed association between unemployment and prosocial engagement. Therefore, in both cases, our estimates would be conservative in predicting the relation between unemployment and prosocial engagement.³

Findings

Main Results

Our main statistical results are presented in Table 2. In columns 1, 2, and 3, we used linear probability modeling to estimate how becoming unemployed at the onset of the pandemic affected the likelihood to participating in formal and informal volunteering and donating behaviors during the pandemic. The results suggest that being laid off during COVID decreases the likelihood of formal volunteering participation by 8 percentage points. For donating, the negative effect was about 6 percentage points. On the other hand, our analysis did not show any association between unemployment and informal volunteering. While the point estimate showed a negative value, the p-value was 0.66. This is related to the fact that overall informal volunteering was relatively stable before and during COVID as shown in Supplementary Table 1.

Table 2.

Regression Analysis.

Variable	LPM			Logistic
	Volunteer	Donating	Informal	Volunteer	Donating	Informal
	(1)	(2)	(3)	(4)	(5)	(6)
Laid off	−0.084** (0.038)	−0.055* (0.031)	−0.013(0.029)	−0.647** (0.281)	−0.654** (0.307)	−0.127(0.352)
Age	0.001(0.002)	−0.001(0.001)	0.002(0.001)	0.007(0.011)	−0.014(0.013)	0.022* (0.013)
Gender_ Male	−0.023(0.038)	−0.016(0.030)	−0.047(0.029)	−0.201(0.278)	−0.275(0.319)	−0.559(0.359)
Race_ Asian	−0.035(0.058)	−0.058(0.055)	−0.022(0.051)	−0.425(0.489)	−0.623(0.592)	−0.115(0.549)
Race_ Black	0.051(0.054)	−0.059(0.057)	−0.025(0.032)	0.411(0.357)	−0.697(0.528)	−0.287(0.461)
Race_ Hispanic	−0.029(0.059)	−0.108* (0.056)	−0.123** (0.063)	−0.155(0.384)	−1.133** (0.547)	−1.372** (0.538)
Race_ Other	0.039(0.157)	−0.081(0.146)	−0.003(0.028)	0.204(0.865)	−0.793(0.908)	1.314* (0.716)
Marital_ Divorced	−0.046(0.087)	−0.082(0.079)	−0.001(0.054)	−0.389(0.556)	−0.794(0.618)	0.008(0.738)
Marital_ Married	0.013(0.056)	−0.033(0.043)	−0.026(0.033)	0.064(0.399)	−0.417(0.447)	−0.298(0.407)
Marital_ Separated	−0.018(0.099)	0.055(0.061)	0.033(0.046)	−0.312(0.773)	0.716(0.581)	0.552(0.569)
Marital_ Widowed	−0.063(0.132)	0.069(0.105)	−0.121(0.083)	−0.505(1.043)	0.827(1.204)	−1.574(1.037)
Parent	−0.040(0.046)	−0.028(0.039)	−0.011(0.035)	−0.271(0.308)	−0.305(0.378)	−0.137(0.391)
Attended worship before	0.051(0.044)	0.016(0.036)	0.045(0.034)	0.350(0.256)	0.205(0.358)	0.757(0.478)
Family experiences_ Sometimes	0.033(0.047)	−0.058* (0.034)	0.010(0.037)	0.350(0.351)	−0.670* (0.379)	0.109(0.406)
Family experiences_ Frequently	0.045(0.054)	−0.047(0.041)	0.051(0.042)	0.371(0.402)	−0.578(0.415)	0.841(0.575)
Financial_ Just met basic	0.068(0.132)	−0.021(0.104)	0.045(0.065)	0.567(0.860)	−0.113(0.853)	0.706(0.959)
Financial_ A little left over for extras	0.064(0.121)	0.092(0.093)	0.039(0.065)	0.579(0.791)	1.073(0.792)	0.661(0.915)
Financial_ Lived comfortably	0.077(0.123)	0.084(0.094)	0.029(0.063)	0.577(0.808)	0.977(0.805)	0.532(0.930)
Employment_ Employed part time	0.005(0.040)	−0.041(0.036)	−0.007(0.033)	0.011(0.297)	−0.433(0.321)	−0.178(0.434)
Education_ College or more	0.050(0.047)	0.035(0.037)	0.022(0.035)	0.317(0.339)	0.399(0.347)	0.316(0.447)
Volunteering before	0.566*** (0.043)	0.079** (0.037)	−0.057* (0.033)	3.244*** (0.396)	0.789** (0.334)	−0.769* (0.418)
Donating before	−0.040(0.045)	0.741*** (0.038)	0.083* (0.047)	−0.236(0.347)	4.521*** (0.385)	0.871** (0.437)
Informal volunteering before	0.074(0.049)	−0.023(0.041)	0.680*** (0.048)	0.810** (0.397)	−0.315(0.419)	4.202*** (0.372)
Constant	−0.043(0.150)	0.220* (0.129)	0.145(0.103)	−3.981*** (1.150)	−1.243(1.136)	−2.745** (1.238)

Note. Columns 1 and 4 shows the result of regression analysis on formal volunteering. Columns 2 and 5 shows the result of regression analysis on donating. Columns 3 and 6 shows the result of regression analysis on informal volunteering. Linear probability model is employed in columns 1, 2, and 3, while logistic regressions is used for columns 4, 5, and 6. Propensity score weighting using a gradient boosted model was applied to balance the characteristics between the laid-off and non–laid-off groups. Sampling weights reflecting population representativeness were also applied.

***

p-val < .01, **p-val < .05, *p-val < .1.

As our outcome variables are binary indicators for formal volunteering, donating, and informal volunteering, we also conducted logistic regression to compare the results from the linear probability model. Columns 4, 5, and 6 in Table 2 suggest that findings from logistic regression approach were substantively identical to those from the linear probability model approach, whereas it showed a relatively higher precision. In addition, although not presented here, the negative effect of unemployment on prosocial behaviors was not moderated by whether the individual had been employed full-time or part-time prior to the pandemic. While most covariates showed statistically insignificant results, likely due to relatively small sample sizes in our analysis and the fact they were already utilized in estimating the propensity score, variables which indicated respondents’ previous engagement to volunteering, donating, and informal volunteering still showed high effect sizes.

To compare the results from the PSW approach and the results from the naïve unweighted OLS approach, we present the latter in Supplementary Table 2. This demonstrates that the negative impact of unemployment to volunteer participation can be significantly under-appreciated with the naïve approach. In contrast to our main estimation, the association of being laid off during COVID with the likelihood of volunteering showed no statistical significance. For donating, while the naïve approach did show significant result, the point estimate underestimated the magnitude of decrease. The disparities persisted for the logistic regression method as well.

To test hypotheses about the heterogeneous effects of unemployment across income groups, we categorized respondents into two groups based on household earnings. Respondents with annual household incomes of $50,000 or more—corresponding to the midpoint of the income categories used in our survey—were classified as higher-income, whereas those earning less than $50,000 were classified as lower-income. Table 3 shows that, as the hypotheses expected, among those laid off the higher income group did not experience a decrease of formal volunteering whereas the lower income group experienced a 12 percentage points decrease in their probability of formal volunteering. Both groups did not show any significant change in informal volunteering. However, for donating, while the higher household income group experienced a significant drop in the propensity to donate, the lower household income group did not show any change. Results for the other remaining covariates are shown in Supplementary Table 3.

Table 3.

Subgroup Analysis.

Income group	Volunteer	Donating	Informal
Income group	(1)	(2)	(3)
a. Higher income	−0.049(0.049)	−0.121** (0.044)	−0.059(0.045)
b. Lower income	−0.120* (0.051)	−0.039(0.041)	0.028(0.040)

Note. Propensity score weighting using a gradient boosted model was applied to balance the characteristics between the laid-off and non–laid-off groups. Sampling weights reflecting population representativeness were also applied. All covariates used in the regression analysis in Table 2 were included both in the computation of the propensity scores and as control variables in the linear probability model.

***

p-val < .01, **p-val < .05, *p-val < .1.

Sensitivity Analysis

One potential limitation of employing PSW arises when statistical outcomes are disproportionately influenced by observations with extremely high or low propensity scores and their corresponding weights. Specifically, in the calculation of the ATT without adjusting for sampling weights, each comparison unit is assigned a weight of $w_{i} = \frac{p (x_{i})}{1 - p (x_{i})}$ . Consequently, units with exceedingly high propensity scores might unduly sway the estimation process. To assess the robustness of our analysis against such cases, we adhered to established precedents in the literature by trimming our sample based on propensity scores (Whatley & González Canché, 2022). We generated two subsamples, one excluding observations within the top and bottom 1% of propensity scores (total exclusion of 2%) and another excluding the top and bottom 2% (total exclusion of 4%), and then re-conducted our linear probability modeling analysis. The outcomes of this refined approach are presented in Table 4.

Table 4.

Sensitivity Analysis.

Trimming specification and income group	Volunteer	Donating	Informal
Trimming specification and income group	(1)	(2)	(3)
a. Both ends 1% trimmed	−0.067* (0.038)	−0.061** (0.029)	−0.011(0.028)
b. Both ends 2% trimmed	−0.071* (0.038)	−0.062** (0.030)	−0.010(0.028)
Higher income group
c. Both ends 1% trimmed	−0.047(0.049)	−0.123** (0.044)	−0.059(0.045)
d. Both ends 2% trimmed	−0.024(0.047)	−0.130** (0.047)	−0.065(0.048)
Lower income group
e. Both ends 1% trimmed	−0.102* (0.050)	−0.026(0.043)	0.035(0.040)
f. Both ends 2% trimmed	−0.100* (0.052)	−0.051(0.040)	0.034(0.039)

***

p-val < .01, **p-val < .05, *p-val < .1.

While the p-values for estimates related to volunteering in panels a and b of Column 1 slightly increased to 0.077 and 0.066, respectively, the consistency of all point estimates with our principal findings underscores the resilience of our analysis to potential distortions from extreme propensity score values. Trimming the outliers within the subgroups of the higher income group and the lower income group also did not change the results, as illustrated in panels c to f. Overall, we find no indication that our results are driven by observations with outlying propensity score induced weights.

Discussion and Conclusion

Theoretical Implications

Contrary to the portrayals often seen in media and academic literature, our evidence reveals that individuals with lower household income did not change their donating behaviors, and those with higher household income did not modify their volunteering activities after being unemployed. Neither changed their involvement in informal volunteering. This suggests that the less affluent were more sensitive in reallocating their time use for formal volunteering, whereas the more affluent were more careful in reallocating their monetary expenditures. These findings highlight the need to disaggregate the effects of unemployment across different economic strata. While prior studies have examined how non-market time is reallocated in response to job loss (Aguiar et al., 2013; Krueger & Mueller, 2012), few have systematically analyzed how these shifts vary by economic background. Even fewer have attempted to formally theorize this heterogeneity. Although our study does not directly address the mechanisms driving these patterns, it underscores an important empirical puzzle, one that future research should further seek to explain.

One possible explanation for the significant decrease in volunteering among lower household income groups is that these individuals could not afford to reallocate their increased non-market time to leisure activities. Given their limited resources to buffer against economic shocks, their time after being laid off may have been consumed by efforts to reenter the labor market, such as job searching, participating in part-time or temporary work, or engaging in activities necessary for obtaining unemployment benefits or job training. This also suggests that the assumption of lower opportunity costs for non-market activities among the unemployed oversimplifies the complex decision-making processes they face. Therefore, it is critical to consider the individual’s economic context when analyzing the impact of unemployment on prosocial engagement. It is also possible that under the unprecedented restrictions of social distancing during the pandemic, those with fewer resources found leaving home or volunteer online too challenging.

Conversely, the expectation that time allocated for market and social engagement activities should follow a strict zero-sum pattern—where a decrease in one lead to an increase in the other—may not hold true in situations of personal shocks like unemployment. This can result in a reduction in both types of engagements as individuals grapple with inconsistency and unpredictability in their lives. For instance, many years ago, Bakke observed the daily routines of unemployed individuals in Greenwich, London, around the Great Depression period noting:

My impression from reading the diaries of the unemployed is that they are not idling their time away. Their time is fully occupied, for the most part at useful tasks. The extra time which is on their hands after time spent looking for work is deducted, is spent by most of the men at home, not in the “pubs” or on the streets. What time is lost and cannot be accounted for is lost due to the irregularity of the daily programme, the lack of routine. (Bakke, 1934, p. 201)

Bakke’s old observations suggest that the additional non-market time is not necessarily directed toward increased social engagements, as economic pressures and lack of routine may limit the capacity for such engagements. Furthermore, the finding that individuals with higher household incomes did not significantly reduce their volunteering during a period of economic uncertainty raises important questions for future research. However, we also acknowledge the possibility that this result may reflect insufficient statistical power to detect the negative coefficients suggested by our analysis, potentially due in part to our use of the PSW approach.

The divergence in donating behaviors following unemployment presents another intriguing area for future investigation. Specifically, it is essential to explore why and how individuals with higher household incomes more promptly reduce their donating. Could this be attributed to their generally higher household expenditures, or perhaps supporting other unemployed family members and thereby prompting a more cautious redistribution of their financial resources in the first year of the pandemic? Or does it relate to superior financial literacy and management skills?

The donating behaviors may reveal underlying strategies that influence broader social engagement decisions. The finding that people with lower household income kept donating as before being unemployed may indicate that the occurrence of unemployment, with regards to donating behaviors, is less dramatic for them. They may have experienced unemployment before and become familiar with the demands of job searching, or they may know their household budget and have experience managing it promptly in both good and bad times. It might also be that in the period of recent joblessness, lower income individuals find time more costly than their small financial gifts to help others.⁴ For people with higher household income on the other hand, the shock of unemployment may mean inability to pay their mortgage or their children private school. In such times they enter a state of financial austerity that includes reducing non-critical expenses such as donations. These possible explanations call for a qualitative study of people soon after being unemployed.

In addition, although our findings indicate that unemployment does not significantly impact informal volunteering overall, variations in specific types of informal volunteering and their frequency post-displacement warrant further exploration. Given the reciprocal nature of informal volunteering, understanding why individuals with fewer economic resources do not increase their participation in such engagements despite negative personal shock could provide valuable insights. One possible explanation is that concerns about virus transmission suppressed the increased demand for direct, in-person assistance outside of one’s household, particularly among lower-income or less-educated individuals, who may have perceived a higher risk of exposure. This suggests that in other crisis contexts, where health risks are not a primary concern, the informal volunteering behaviors of lower-income unemployed individuals may differ.

Most studies examining the relationship between unemployment and prosocial engagement treat unemployed individuals as a homogeneous group, without accounting for the duration of their unemployment or their individual characteristics. This approach overlooks the nuanced experiences of unemployment. Our study addresses this gap by investigating changes in prosocial behavior shortly after unemployment, distinguishing between individuals based on household income, and categorizing them as either poor or better off. While this approach offers valuable insights, it should not be generalized to our understanding of the relationship between longer-term unemployment and prosocial engagements (Piatak, 2016). For instance, although Munyon et al. (2019) did not examine specific prosocial behaviors in relation to unemployment, they found that individuals experiencing long-term unemployment with stronger attachment to their community tended to remain unemployed longer due to the “stickiness” of community embeddedness. This raises the question of whether this stickiness affects people with varying economic means in the same way. Clearly, this area of research deserves further exploration.

Practical Implications

Nonprofit organizations rely on volunteers and donors to sustain their initiatives, yet participation in these roles is influenced by economic conditions. Volunteering is often recommended to unemployed individuals as a way to enhance job prospects and maintain self-worth. However, this approach may not be suitable for everyone, particularly those who have recently lost their jobs or are facing financial hardship. These individuals may require time to stabilize their personal and economic situations before engaging in volunteer work. Similarly, charitable donations may recover slowly among higher-income individuals following economic downturns, requiring nonprofit organizations to exercise patience and adaptability.

Governments are also increasingly relying on nonprofit, charitable organizations, and informal entities to enhance the provision of welfare services, from poverty alleviation to social care. This growing trend highlights the importance of prosocial engagement and volunteering, which are central to these institutions. As Haski-Leventhal et al. (2018, p. 1152) emphasize, policymakers must “strive to increase volunteering rates through social policy, legislation, and awards” to ensure these organizations can effectively support welfare services.

Our findings indicate that the volunteer pool can significantly shrink during crises marked by widespread unemployment, such as the COVID-19 pandemic. We suggest that these challenging periods tend to recur with quasi-regularity, necessitating proactive strategies to adapt, rather than relying on delayed, ad hoc responses. There are numerous channels through which policies and institutions can strengthen this sector. For example, in Australia, volunteers are protected under worker health and safety laws. Governments can also stimulate social engagement and prosocial behavior by reducing the administrative burdens faced by charitable organizations.

For recently unemployed individuals, replacing ineffective or outdated job training requirements tied to unemployment benefits with volunteering opportunities could provide a valuable alternative. In line with Biermann et al. (2024, p. 3), who advocate for greater attention to “volunteering using specialized occupational competencies and resources,” greater emphasis should be placed on coordinating opportunities for unemployed individuals to apply their skills within their local communities. In addition to these institutional efforts, nonprofits can play a crucial role by developing skill-based volunteer roles and offering more flexible engagement opportunities. Furthermore, they may consider expanding low-commitment volunteer positions and diversifying fundraising strategies to adapt to shifting donor behaviors, particularly during massive economic downturns.

Conclusion

Unemployment has emerged as a persistent social issue, exacerbated by ongoing societal shocks. Despite this, its relationship with social engagement remains underexplored. In this study, we specifically examine prosocial engagement by analyzing nationally representative survey responses collected during the COVID-19 pandemic. This survey allows us to focus on individuals who were recently unemployed. In addition, we explore how household income among the newly unemployed is associated with prosocial engagement. To address potential confounders intertwined with unemployment and its impacts, we employed PSW, estimated via a GBM. This machine learning approach allowed us to incorporate non-linearities and interactions among covariates, effectively minimizing bias between those who were unemployed and those who were not during the pandemic.

Our findings indicate that unemployment is associated with a decrease in formal volunteering among individuals with lower household income and a reduction in donating participation among those with higher household income. In addition, we observed that informal volunteering does not appear to be influenced by unemployment status. These results remained robust after the sensitivity tests, and naïve estimation of the effect of unemployment tended to underappreciate the negative impact of unemployment.

Our findings primarily pertain to short-term unemployment and are situated within the context of the COVID-19 pandemic. While we acknowledge the limited generalizability of our results, we believe they offer relevant insights into the immediate consequences of job displacement, particularly in cases where mass layoffs occur within short timeframes. This is partly due to the fact that our analysis extensively balanced observable characteristics between unemployed and non-unemployed individuals, using a model that incorporates higher-order terms and complex interactions via machine-learning-based iteration. Consequently, some distinct features of COVID-19-induced unemployment, such as the contraction of prosocial engagement opportunities due to social distancing, can be at least partially accounted for in our estimation process.

A promising avenue for future research is the study of long-term unemployment. For instance, advances in artificial intelligence and automation are expected to reshape labor markets, leading to persistent rather than temporary job displacement. These technological shifts are likely to reflect structural changes in skill demand, potentially resulting in longer unemployment durations than before. Some studies suggest a positive correlation between unemployment duration and both formal and informal volunteering (Holstein & Qvist, 2025; Munyon et al., 2019). Beyond identifying the long-term effects of unemployment, distinguishing between short-term and long-term unemployment also warrants substantial scholarly attention. While the distinction remains conceptually challenging and varies by research focus, empirical studies exploring the cutoff point in unemployment duration where prosocial engagement patterns diverge could provide valuable insights to the literature.

In addition, while our study takes an important first step in examining the relationship between unemployment and prosocial engagement, our approach relied on a single income-based cutoff to differentiate between lower- and higher-income groups. This method, while informative in providing a broad, nationally representative picture, does not fully capture regional variations in the cost of living. Future research could refine this approach by incorporating multiple income thresholds to better account for local economic conditions, thereby offering more precise insights into smaller-scale populations. Taken together, our findings contribute to understanding of the relationship between unemployment and prosocial engagement, offering empirical evidence derived from a more rigorous methodological approach than conventional studies. As labor markets continue to evolve, particularly amid recurring disruptions, further inquiry into both the short-term and long-term consequences of unemployment for social and civic life remains essential.

Supplemental Material

sj-pdf-1-nvs-10.1177_08997640251348416 – Supplemental material for Unemployment and Prosocial Engagement: Behavior Changes During the COVID-19 Pandemic

Supplemental material, sj-pdf-1-nvs-10.1177_08997640251348416 for Unemployment and Prosocial Engagement: Behavior Changes During the COVID-19 Pandemic by Hwiyoung P. Lee, Daniel Choi, Ram A. Cnaan and Femida Handy in Nonprofit and Voluntary Sector Quarterly

Footnotes

Appendix

Author’s Note

Ram A. Cnaan is also affiliated with Kyung Hee University, Seoul, South Korea.

Data Availability

Data will be made available on reasonable request.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Generosity Commission of Giving USA and the Institute of Education Sciences under PR/Award #R305B200035.

ORCID iDs

Hwiyoung P. Lee

Ram A. Cnaan

Femida Handy

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Hwiyoung P. Lee is a PhD student at the School of Social Policy and Practice at the University of Pennsylvania and an IES predoctoral fellow. His research focuses on the causal determinants of social and civic engagement. He holds dual bachelor’s degrees in Social Welfare and Economics from Seoul National University, where he also earned an MA in Social Welfare.

Daniel Choi is a PhD student at the School of Social Policy and Practice at the University of Pennsylvania. His research focuses on social entrepreneurship, faith-based social welfare, and the role of volunteerism and philanthropy in society. He earned his Bachelor of Social Work from the University of Georgia in 2017 and his Master of Social Work from the University of Pennsylvania in 2018.

Ram A. Cnaan is a professor at the University of Pennsylvania and director of the Program for Religion and Social Policy Research. He is a global eminent scholar at Kyung Hee University, South Korea, and a fellow of the American Academy of Social Work and Social Welfare. His research focuses on volunteerism, faith-based organizations, criminal justice, and social policy.

Femida Handy is a professor of Social Policy at the University of Pennsylvania. Her research focuses on nonprofit economics, volunteering, philanthropy, environmental issues, and microfinance. Her recent books include Ethical Decision Making for Social Impact and The Palgrave Handbook on Global Philanthropy.

References

Acemoglu

Restrepo

(2019). Automation and new tasks: How technology displaces and reinstates labor. Journal of Economic Perspectives, 33(2), 3–30.

Aguiar

Hurst

Karabarbounis

(2013). Time use during the great recession. American Economic Review, 103(5), 1664–1696. https://doi.org/10.1257/aer.103.5.1664

Andersen

S. H.

(2009). Unemployment and subjective well-being. A question of class? Work and Occupations, 36(1), 3–25.

Autor

Dorn

Hanson

(2022). On the persistence of the China shock. Brookings Papers on Economic Activity, 2021(2), 381–476. https://doi-org.proxy.library.upenn.edu/10.1353/eca.2022.0005

Baines

Hardill

(2008). ‘At least I can do something’: The work of volunteering in a community beset by worklessness. Social Policy and Society, 7(3), 307–317.

Bakke

E. W.

(1934). The unemployed man: A social study. E. P. Dutton.

Bekkers

Wiepking

(2011). A literature review of empirical studies of philanthropy: Eight mechanisms that drive charitable giving. Nonprofit and Voluntary Sector Quarterly, 40(5), 924–973. https://doi.org/10.1177/0899764010380927

Beland

L. P.

Brodeur

Wright

(2023). The short-term economic consequences of COVID-19: Exposure to disease, remote work and government response. PLOS ONE, 18(3), Article e0270341. https://doi.org/10.1371/journal.pone.0270341

Bennett

(2021). Fewer jobs have been lost in the EU than in the U.S. during the COVID-19 downturn (April 15). https://www.pewresearch.org/fact-tank/2021/04/15/fewer-jobs-have-been-lost-in-the-eu-than-in-the-u-s-during-the-covid-19-downturn/

10.

Bennett

(2018). Financial charity giving behaviour of the working poor: An empirical investigation. Journal of Marketing Management, 34(17–18), 1587–1607.

11.

Biermann

K. Y.

Breitsohl

Meijs

L. C.

(2024). Occupation-related volunteering: A qualitative systematic literature review, conceptualization, and directions for future research. SAGE Open, 14(2), 21582440241255834.

12.

Blustein

D. L.

(2019). The importance of work in an age of uncertainty: The eroding work experience in America. Oxford University Press.

13.

Bosman

(2009, March 15). From ranks of jobless, a flood of volunteers. The New York Times. https://www.nytimes.com/2009/03/16/nyregion/16volunteers.html?

14.

Brand

J. E.

(2015). The far-reaching impact of job loss and unemployment. Annual Review of Sociology, 41(1), 359–375. https://doi.org/10.1146/annurev-soc-071913-043237

15.

Brodeur

Gray

Islam

Bhuiyan

(2021). A literature review of the economics of COVID-19. Journal of Economic Surveys, 35(4), 1007–1044.

16.

Bundi

Freitag

(2020). Economic hardship and social capital in Europe: A comparative analysis of 27 democracies. European Journal of Political Research, 59(2), 290–311.

17.

Calvo-Armengol

Jackson

M. O.

(2004). The effects of social networks on employment and inequality. American Economic Review, 94(3), 426–454.

18.

Centola

(2018). How behavior spreads: The science of complex contagions. Princeton University Press.

19.

Chambré

S. M.

(2020). Has volunteering changed in the United States? Trends, styles, and motivations in historical perspective. Social Service Review, 94(2), 373–421. https://doi.org/10.1086/708941

20.

Cho

Daley

R. T.

Cunningham

T. J.

Kensinger

E. A.

Gutchess

(2022). Aging, empathy, and prosocial behaviors during the COVID-19 pandemic. The Journals of Gerontology: Series B, 77(4), e57–e63.

21.

Clark

(2014). Hard times: The divisive toll of the economic slump. Yale University Press.

22.

Cnaan

R. A.

Handy

Marrese

Choi

Ferris

(2022). Generosity trends and impacts: Before and during the Covid-19 pandemic in the USA. University of Pennsylvania. https://sp2.upenn.edu/wp-content/uploads/2014/07/Generosity-Trends-and-Impacts.pdf

23.

Cnaan

R. A.

Amrofell

(1994). Mapping volunteer activity. Nonprofit and Voluntary Sector Quarterly, 23(4), 335–351.

24.

Couch

K. A.

Fairlie

R. W.

(2020). Early evidence of the impacts of COVID-19 on minority unemployment. Journal of Public Economics, 192, 104287. https://doi.org/10.1016/j.jpubeco.2020.104287

25.

Day

K. M.

Devlin

R. A.

(1998). The payoff to work without pay: Volunteer work as an investment in human capital. Canadian Journal of Economics, 31(5), 1179–1191.

26.

Desmond

(2012). Disposable ties and the urban poor. American Journal of Sociology, 117(5), 1295–1335. https://doi.org/10.1086/663574

27.

González Canché

M. S

. (2014). Is the community college a less expensive path toward a bachelor’s degree? Public 2-and 4-year colleges’ impact on loan debt. The Journal of Higher Education, 85(5), 723–759.

28.

Granovetter

M. S.

(1974). Getting a job: A study of contacts and career. University of Chicago Press.

29.

Hackl

Halla

Pruckner

G. J.

(2007). Volunteering and income–the fallacy of the good Samaritan? Kyklos, 60(1), 77–104.

30.

Haller

Lubenko

Presti

Squatrito

Constantinou

Nicolaou

Papacostas

Aydın

Chong

Y. Y.

Chien

W. T.

Cheng

H. Y.

Ruiz

F. J.

García-Martín

M. B.

Obando-Posada

D. P.

Segura-Vargas

M. A.

Vasiliou

V. S.

McHugh

Höfer

Baban

. . . Gloster

A. T.

(2022). To help or not to help? Prosocial behavior, its association with well-being, and predictors of prosocial behavior during the Coronavirus disease pandemic. Frontiers in Psychology, 12, 775032.

31.

Handy

Cnaan

R. A.

Hustinx

Kang

Brudney

J. L.

Haski-Leventhal

Holmes

Meijs

L. C. P. M.

Pessi

A. B.

Ranade

Yamauchi

Zrinscak

(2010). A cross-cultural examination of student volunteering: Is it all about résumé building? Nonprofit and Voluntary Sector Quarterly, 39(3), 498–523. https://doi.org/10.1177/0899764009344353

32.

Harmon-Jones

Armstrong

Olson

J. M.

(2018). The influence of behavior on attitudes. In Albarracin

Johnson

B. T.

(Eds.), The Handbook of Attitudes, Volume 1: Basic Principles (2nd ed., pp. 404–449). Routledge.

33.

Haski-Leventhal

Meijs

L. C. P. M.

Lockstone-Binney

Holmes

Oppenheimer

(2018). Measuring volunteerability and the capacity to volunteer among non-volunteers: Implications for social policy. Social Policy & Administration, 52(5), 1139–1167.

34.

Hodgkinson

V. A.

Weitzman

M. S.

Noga

S. M.

Gorski

H. A.

(1992). Giving and volunteering in the United States: Findings from a National Survey, 1992 Edition. Independent Sector.

35.

Holstein

E. L.

Qvist

H. Y.

(2025). Lagging behind by doing good: How volunteering prolongs unemployment. The British Journal of Sociology, 1468–4446. https://doi.org/10.1111/1468-4446.13182

36.

Homans

G. C.

(1958). Social behavior as exchange. American Journal of Sociology, 63(6), 597–606. https://doi.org/10.1086/222355

37.

Jenner

J. R.

(1981). Volunteerism as an aspect of women’s work lives. Journal of Vocational Behavior, 19(3), 302–314. https://doi.org/10.1016/0001-8791(81)90065-8

38.

Krueger

A. B.

Mueller

A. I.

(2012). Time use, emotional well-being, and unemployment: Evidence from longitudinal data. American Economic Review, 102(3), 594–599. https://doi.org/10.1257/aer.102.3.594

39.

Lebenbaum

De Oliveira

McKiernan

Gagnon

Laporte

(2024). COVID-19 pandemic, physical distancing policies, and the non-profit sector volunteer force. Nonprofit and Voluntary Sector Quarterly, 53(1), 274–288.

40.

Lee

Farrell

(2003). Buddy, can you spare a dime? Homelessness, panhandling, and the public. Urban Affairs Review, 38(3), 299–324.

41.

Lim

Laurence

(2015). Doing good when times are bad: Volunteering behaviour in economic hard times: Doing good when times are bad. The British Journal of Sociology, 66(2), 319–344.

42.

Lim

Sander

(2013). Does misery love company? Civic engagement in economic hard times. Social Science Research, 42(1), 14–30.

43.

Lim

V. K. G.

Chen

S. S. Y.

Tan

(2016). Unemployed and exhausted? Job-search fatigue and reemployment quality. Journal of Vocational Behavior, 92, 68–78. https://doi.org/10.1016/j.jvb.2015.11.003

44.

Lucas

R. E.

Clark

A. E.

Georgellis

Diener

(2004). Unemployment alters the set point for life satisfaction. Psychological Science, 15(1), 8–13.

45.

Martin

Schlereth

(2024). When a crisis has a silver lining: Social media P2P fundraising at the start of COVID-19. Nonprofit and Voluntary Sector Quarterly, 53(5), 1329–1342.

46.

McCaffrey

D. F.

Ridgeway

Morral

A. R.

(2004). Propensity score estimation with boosted regression for evaluating causal effects in observational studies. Psychological Methods, 9(4), 403.

47.

Menchik

P. L.

Weisbrod

B. A.

(1987). Volunteer labor supply. Journal of Public Economics, 32(2), 159–183.

48.

Miettinen

Jalovaara

(2020). Unemployment delays first birth but not for all. Life stage and educational differences in the effects of employment uncertainty on first births. Advances in Life Course Research, 43, 100320. https://doi.org/10.1016/j.alcr.2019.100320

49.

Munyon

T. P.

Madden

L. T.

Madden

T. M.

Vigoda-Gadot

(2019). (Dys)functional attachments? How community embeddedness impacts workers during and after long-term unemployment. Journal of Vocational Behavior, 112, 35–50. https://doi.org/10.1016/j.jvb.2019.01.005

50.

Musick

M. A.

Wilson

(2008). Volunteers: A social profile. Indiana University Press.

51.

Penny

Finnegan

(2019). Employability or self-respect? Justifying the use of volunteering to address unemployment. Voluntary Sector Review, 10(2), 151–165.

52.

Piatak

J. S.

(2016). Time is on my side: A framework to examine when unemployed individuals volunteer. Nonprofit and Voluntary Sector Quarterly, 45(6), 1169–1190.

53.

Putnam

R. D.

(2000). Bowling alone: The collapse and revival of American community. Simon & Schuster.

54.

Ramaekers

M. J. M.

van der Lippe

Derks

(2024). Why neighbors would help: A vignette experiment on reciprocity in informal helping. Nonprofit and Voluntary Sector Quarterly. Advance online publication. https://doi.org/10.1177/08997640241241323

55.

Reed

P. B.

Selbee

L. K.

(2000). Formal and informal volunteering and giving: Regional and community patterns in Canada. Statistics Canada. https://publications.gc.ca/Collection/Statcan/75F0048M/75F0048MIE2002005.pdf

56.

Ridgeway

Kovalchik

S. A.

Griffin

B. A.

Kabeto

M. U.

(2015). Propensity score analysis with survey weighted data. Journal of Causal Inference, 3(2), 237–249.

57.

Ridgeway

McCaffrey

D. F.

Morral

A. R.

Cefalu

Burgette

L. F.

Pane

J. D.

Griffin

B. A.

(2023). Toolkit for weighting and analysis of nonequivalent groups: A tutorial for the R TWANG package. Rand.

58.

Rubin

D. B.

(1979). Using multivariate matched sampling and regression adjustment to control bias in observational studies. Journal of the American Statistical Association, 74(366a), 318–328.

59.

Ruhm

C. J.

(1991). Are workers permanently scarred by job displacements? The American Economic Review, 81(1), 319–324.

60.

Schlosser

F. K.

Zinni

D. M.

(2011). Transitioning ageing workers from paid to unpaid work in non-profits: Transitioning from paid to unpaid work. Human Resource Management Journal, 21(2), 156–170.

61.

Sin

N. L.

Klaiber

Wen

J. H.

DeLongis

(2021). Helping amid the pandemic: Daily affective and social implications of COVID-19-related prosocial activities. The Gerontologist, 61(1), 59–70.

62.

Small

M. L.

Gose

L. E.

(2020). How do low-income people form survival networks? Routine organizations as brokers. The ANNALS of the American Academy of Political and Social Science, 689(1), 89–109. https://doi.org/10.1177/0002716220915431

63.

Smith

D. H.

(1983). Synanthrometrics: On progress in the development of a general theory of voluntary action and citizen participation. In Smith

D. H.

Van Til

(Eds.), International perspectives on voluntary action research (pp. 80–94). University Press of America.

64.

Smith

D. H.

(1994). Determinants of voluntary association participation and volunteering: A literature review. Nonprofit and Voluntary Sector Quarterly, 23(3), 243–263.

65.

Spence

(1973). Job market signaling. The Quarterly Journal of Economics, 87(3), 355–374. https://doi.org/10.2307/1882010

66.

Spera

Ghertner

Nerino

DiTommaso

(2015). Out of work? Volunteers have higher odds of getting back to work. Nonprofit and Voluntary Sector Quarterly, 44(5), 886–907.

67.

Strauß

(2008). Volunteering and social inclusion: Interrelations between unemployment and civic engagement in Germany and Great Britain. Springer-Verlag.

68.

Taniguchi

(2006). Men’s and women’s volunteering: Gender differences in the effects of employment and family characteristics. Nonprofit and Voluntary Sector Quarterly, 35(1), 83–101.

69.

Taniguchi

(2012). The determinants of formal and informal volunteering: Evidence from the American Time Use Survey. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 23, 920–939.

70.

Whatley

González Canché

M. S.

(2022). A robust estimation of the relationship between study abroad and academic outcomes among community college students. Research in Higher Education, 63(2), 271–308.

71.

Yasenov

V. I.

(2020). Who can work from home? IZA Discussion Paper No. 13197. https://doi.org/10.2139/ssrn.3590895

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.38 MB