How socio-economic inequality affects individuals’ civic engagement: a systematic literature review of empirical findings and theoretical explanations

Abstract

Rising socio-economic inequality in many countries raises the question of how it affects individuals’ civic engagement in the forms of charitable giving, volunteering and non-profit membership. Though a growing body of multidisciplinary literature has started to address this issue, empirical results vary considerably and explanations about what underlies this relationship remain fragmentary. We thus conduct a systematic literature review to (a) provide a synopsis of empirical findings and (b) identify theoretical explanations and presumed mediating mechanisms underlying this relationship. Reviewing 70 studies, we find that higher inequality is most often negatively related to civic engagement, and that this relation is moderated by individual factors, for example, income and education. Furthermore, we map the proposed theoretical explanations into five key approaches. For each of these, we trace and identify the underlying mechanisms at both the societal and the individual level, and provide a conceptual framework that facilitates their empirical analysis in future.

1. Introduction

In the light of rising socio-economic inequality in many countries (Mann and Riley, 2007; Lakner and Milanovic, 2016; Alvaredo et al., 2018), the question of how inequality affects individuals’ civic engagement is becoming increasingly pressing. If inequality undermines civic engagement, a vibrant civil society and the manifold goods and services provided through individuals’ civic engagement might be at stake (Dunn et al., 2008; Mackerle-Bixa et al., 2009; Son and Wilson, 2012; Wiepking and Handy, 2015; Igalla et al., 2019). At the same time, civic engagement itself is often understood as a tool for alleviating socio-economic inequality and could thus be expected to increase as socio-economic inequality increases (Duquette, 2018).

While a growing body of multidisciplinary literature has started to address this issue, existing knowledge is quite fragmented. On the one hand, empirical results are sometimes contradictory, revealing both positive and negative relationships between socio-economic inequality and individuals’ civic engagement (Alesina and La Ferrara, 2000; Lancee and Van de Werfhorst, 2012; Rotolo and Wilson, 2014; Payne and Smith, 2015; Duquette, 2018). On the other hand, a variety of different hypotheses to theoretically explain this relationship are suggested. Yet these explanations are often unconnected and incomplete, provide only elements of a coherent chain of reasoning and remain within disciplinary boundaries. For example, sociological literature highlights a negative relationship, arguing that rising inequality—in terms of income, wealth or social class—results in a decrease in social cohesion and generalized social trust, ultimately leading to reduced volunteering, charitable giving and non-profit membership (Gesthuizen et al., 2009; Collins and Guidry, 2018). Economic literature, in contrast, has argued that more unequal societies have a greater unmet need for public goods, and that individuals’ altruistic preferences (e.g. inequality aversion) motivate them to contribute voluntarily to the public good (Mastromatteo and Russo, 2017; Duquette and Hargaden, 2019). Still other schools of thought argue that socio-economic inequality results in power differentials, and that a lack of power prevents some people from participating in civic engagement (Duncan, 2010; Godfrey and Cherng, 2016; Filetti and Janmaat, 2018), or that socio-economic inequality might lessen the equality of opportunity to engage civically based on available resources (Lancee and Van de Werfhorst, 2012; Karakoc, 2013). Overall, elements of various arguments and mechanisms are brought forward, sometimes confused and often seemingly unaware of each other. What is more, they have at best only been indirectly tested in previous studies so far (Lancee and Van de Werfhorst, 2012).

Against this background, the first aim of this study is to provide an interdisciplinary overview of the empirical findings on the relation between socio-economic inequality and individuals’ civic engagement. The second and main aim, however, is to develop a conceptual framework that outlines and contextualizes the mechanisms commonly put forward to explain the effects of socio-economic inequality on civic engagement. This framework seeks to provide a basis to empirically test the various hypothesized mechanisms in future studies, and to investigate which of them in fact matter in different contexts, for different groups of people and for different types and purposes of civic engagement.

The corresponding research question reads: How does socio-economic inequality affect individuals’ charitable giving, volunteering and non-profit membership, and which mechanisms are at work here? To answer this, we conduct a systematic literature review (SLR). We (a) deliver an extensive overview of extant empirical evidence and (b) suggest a conceptual framework identifying the moderating and mediating mechanisms that serve to explain how inequality affects civic engagement and how various social groups are affected differently. Our analysis is based on a literature search using the ISI Web of Knowledge’s Social Sciences Citation Index (SSCI) database and retrospective reference list checking (Tranfield et al., 2003; Denyer and Tranfield, 2009; Brunton et al., 2017). We identify 70 studies examining the effect of inequality at the national, regional or local level on one or more types of civic engagement.

Regarding the forms of civic engagement, we focus on charitable giving and volunteering with a non-profit organization, labeled formal charitable giving and formal volunteering, as well as on non-profit membership. While we do not assume them to be an entirely homogeneous set of activities (Lee et al., 1999; Jones, 2006), they are usually regarded as one dimension of civic engagement encompassing ‘communal activities’, which aim at influencing circumstances in society that are of relevance to others, outside a group of family and friends (Adler and Goggin, 2005; Ekman and Amnå, 2012). Furthermore, previous research has suggested that, for these activities, similar mechanisms might be at work (Uslaner and Brown, 2005; Gesthuizen et al., 2009). We do not include formal ‘political activities’, such as voting, which are also included in broader definitions of civic engagement (Ekman and Amnå, 2012), because different motivations are likely to be at work for these (Uslaner and Brown, 2005).

As regards socio-economic inequality, we look for studies that understand inequality as a relational concept which describes a contextual factor influencing civic engagement. We opt for a broad understanding of socio-economic inequality that captures differences between individuals and groups which matter for their quality of life and general well-being (Lamont et al., 2014). Accordingly, we include studies that examine inequalities of income, wealth, education, status or class, and consider it on an aggregate level, for example, nation states, federal states or neighborhoods. Thus, we do not consider the extensive literature on how individual-level resources, such as income or education, impact civic engagement (Wilson and Musick, 1997; Bekkers and Wiepking, 2011; Wiepking and Bekkers, 2012; Wilson, 2012; Bekkers et al., 2016).

This study’s contribution is therefore a synopsis of research on the socio-spatial context effect of socio-economic inequality on civic engagement in the forms of volunteering, charitable giving and non-profit membership. It depicts the current state of empirical evidence, showing that higher levels of inequality are most often negatively related to civic engagement and discusses central limitations of the relatively narrow current empirical understanding. To address the major limitation in this field of research, namely, the lack of empirical investigation of the hypothesized mechanisms, this study provides a conceptual framework disentangling the various potential effects of inequality through mechanisms on the contextual and individual level. This framework may serve as a basis for testing competing but also complementary hypotheses in a systematic manner in future studies, which is an important step for moving the research field forward.

The paper proceeds as follows. Section 2 describes the method used for locating, selecting and analyzing relevant studies. Section 3 summarizes the results of 49 empirical studies on how socio-economic inequality affects civic engagement. Section 4 maps the scattered theoretical explanations brought forward in all of the included 70 studies into a conceptual framework of five explanatory mechanisms. Section 5 presents a discussion and conclusion.

2. Method: SLR

Our study builds on a SLR as this method enables the aggregation of empirical results as well as a configuration of underlying theoretical explanations (Sandelowski et al., 2012; Gough et al., 2017). In addition, this replicable research process aids transparency, inclusiveness and the explanatory value of literature reviews (Tranfield et al., 2003; Gough and Thomas, 2017). It consists of three broad steps: (1) locating, selecting and evaluating relevant studies; (2) analysis and synthesis and (3) reporting and using the results (Tranfield et al., 2003; Denyer and Tranfield, 2009).¹

2.1 Locating, selecting and evaluating studies

We commenced our SLR with a scoping study by which we identified the keywords depicting socio-economic inequality on the one hand, and civic engagement on the other. After initially extracting a larger number of keywords regarding inequality, we selected those displayed in Table 1 to strike a balance between precision and sensitivity (Brunton et al., 2017). The search terms relating to inequality were combined with those relating to each of the three forms of civic engagement using search strings (see Table 1). We searched for relevant studies in the ISI Web of Knowledge’s SSCI database because it covers a wide array of disciplines and is thus well suited for searches on topics spanning multiple disciplines. Our search was limited to studies written in English and published before May 2019.

Table 1

Combination of key words for search strings

Inequality		Civic engagement	N
Inequality OR ‘income disparit*’ OR segregation	AND	(Membership OR member*) AND (non-profit OR nonprofit OR ‘non-profit’ OR nongovernment OR ‘nongovernmental organi*’ OR charit* OR ‘civil society’ OR npo OR ngo OR ‘third sector’)	106
		Volunteer OR volunteering	265
		‘Civic engagement’ OR ‘civic participation’	219
		Philanthrop* OR donat* OR ‘charitable giving’ OR ‘prosocial behavi*’ OR ‘public good provision’ OR ‘provision of public good*’	400
Sum of studies			990
Number of studies after removing duplicates			945

Note: N, number of studies identified.

Out of the 945 studies identified, we included only those that meet the following inclusion criteria, which resulted in a sample of 32 publications.

• Studies must explicitly discuss and/or empirically test the effect of socio-economic inequality on one or more types of formal civic engagement.

• Studies must perceive socio-economic inequality as a relational concept.

• Studies must focus on formal civic engagement.²

Furthermore, we included five studies meeting our inclusion criteria that were found in our scoping study and that did not appear in our database search (three journal articles, one unpublished study and one book chapter).

Next, we applied retrospective snowballing (Jalali and Wohlin, 2012; Brunton et al., 2017) to these 37 studies as well as to the studies meeting our inclusion criteria as a result of the snowballing process. For retrospective snowballing, all references in articles and book chapters were screened by title in the reference list, while references in books were assessed when they seemed relevant based on their citation in the text (Brunton et al., 2017). This led to identifying another 33 studies meeting our inclusion criteria,³ leaving us with a total of 70 studies—both theoretical and empirical—for our analysis. Figure 1 provides an overview of the literature search process.

2.2 Analysis and synthesis

We first set up a database containing information on the studies’ metadata, its methodology and, if available, empirical results. This database served as the basis for our description and the synthesis of the empirical results in line with our study’s first aim. Second, we coded the theoretical explanations and mechanisms underlying the relationship between inequality and civic engagement using a concept matrix (Webster and Watson, 2002). In doing so, we applied an inductive open coding approach (Sutcliffe et al., 2017), in which concepts emerge in an iterative process of reading and re-reading studies. Hence, coding and analysis were integrated, not separated. To ensure a common understanding of the mechanisms, working definitions were agreed upon between authors as new mechanisms were identified (Sutcliffe et al., 2017). We then followed a Framework Synthesis approach (Thomas et al., 2017), configuring arguments from the literature into an emergent framework of mechanisms to explore theory and advance theory-building for these mechanisms.

2.3 Reporting and using the results

Results are first described by aggregating the empirical studies’ results in a tabular overview. There are several limitations to the ‘vote counting’ approach of summarizing the results of studies, which are mostly related to varying methodological quality and statistical power between studies (Thomas et al., 2017). We mitigate these weaknesses by reporting the studies’ empirical strategy and sample size. Furthermore, we do not exclude results based on statistical significance, but rather report it and use it as one piece of information to assess the strength of empirical evidence (McShane et al., 2019). Following the aggregation of empirical results, we present a conceptual framework that disentangles and clearly identifies the various theoretical explanations and mechanisms that connect socio-economic inequality and civic engagement and in turn describe each mechanism.

3. Overview of results of empirical studies

Altogether, we identified 70 studies spanning numerous disciplines, including administrative science, economics, political science, psychology and sociology. Out of these studies, 49 are relevant to our first research question because they include empirical analyses on the relationship between socio-economic inequality and civic engagement. Below, we report their results according to the type of civic engagement examined to account for the fact that these are not necessarily equally affected by socio-economic inequalities. Furthermore, we distinguish the effects of inequality on the incidence versus the amount of charitable giving, volunteering and non-profit membership, that is, whether one engages at all versus to what extent one engages. Finally, we report individuals’ characteristics that have been found to moderate the relationship (see Table 2).

Table 2

Empirical studies’ findings on the effect of socio-economic inequality on charitable giving, volunteering, and non-profit membership

Author(s) (year)	Region (analysis scale)	N	Type of inequality (indicator used)	Charitable giving		Volunteering		Non-profit membership		Moderators
				Inc.	Am.	Inc.	Am.	Inc.	Am.
Studies solely on charitable giving
Bielefeld et al. (2005)	USA (states)	4200	Income inequality (95/5 ratio)	+a	- (ns)a
Evers and Gesthuizen (2011)	Europe and USA (individuals, countries)	33 474 (20)	Income inequality (Gini index)	-/+ (ns)b,c
Payne and Smith (2015)	Canada (FSAs)	3964	Income inequality (Gini index, Theil index, 90/10 ratio, Top 1% income share)	-(ns)a,d	+a					Indiv. income, education
Oto-Peralías and Romero-Avila (2017)	Spain (municipalities)	570	Wealth inequality (historical land inequality)	-a
		234	Income inequality (Gini index)	- (ns)a
Duquette (2018)	USA (states)	7977	Income inequality (top income shares)		-a					Indiv. income
Duquette and Hargaden (2019)	USA (individuals, groups)	2880 (144)	Economic inequality (unequal endowment with tokens tied to real money)	-b	-b					Indiv. endowment
Longhofer et al. (2019)	USA (counties)	9324	Income inequality (Gini index of household income)		-a
Studies solely on volunteering
Oliver (1999)	USA (individuals, municipalities)	1543 (518)	Income inequality (index of qualitative variation)			+b,e
Clark and Kim (2012)	New Zealand (meshblocks of ∼100 people and area units of ∼2,000 people)	49 600 (meshblocks)	Income inequality (Gini index of nominal household income)			+/- (ns)a,f
		3504 (area units)
Lancee and Van de Werfhorst (2012)	Europe (individuals, countries)	140 540 (24)	Income inequality (Gini index of disposable household income)			-b				Indiv. income
Rothwell (2012)	USA (MSAs)	183	Income inequality (Gini index)				-a
Smith (2012)	Canada (individuals, communities)	∼13 000 (140)	Income inequality (Gini index of household income)			- (ns)a
Rotolo and Wilson (2014)	USA (individuals, MSAs)	196 454 (248)	Income inequality (Gini index of household income)			-b
Godfrey and Cherng (2016)	USA (individuals, counties)	12 240 (137)	Income inequality (Gini index)			+b,e				SES of family, ethnicity
Mastromatteo and Russo (2017)	World (individuals, countries)	∼97 500 (56)	Income inequality (Gini index, top 10% income share)			+b
Veal and Nichols (2017)	Europe (countries)	23	Income inequality (Gini index, S80/S20 ratio and P90/P50 ratio of household income)				-g
Collins and Guidry (2018)	USA (individuals, MSAs)	20 271 (26)	Income inequality (Gini index of household income)			+ (ns)a
Filetti and Janmaat (2018)	Europe (individuals, countries	248 741 (29)	Income inequality (Gini index)			-b				Indiv. income
Fladmoe and Steen-Johnsen (2018)	Norway (individuals, municipalities)	5239 (61)	Income inequality (Gini index and 90/10 ratio)			+ (ns)b	- (ns)b,h
Studies solely on non-profit membership
Kawachi et al. (1997)	USA (states)	39	Income inequality (Robin Hood index)						-g
Kennedy et al. (1998)	USA (states)	39	Income inequality (Robin Hood index)						-g
Alesina and La Ferrara (2000)	USA (individuals, MSAs, PMSAs)	10 534 (∼200)	Income inequality (Gini index of family income)					-b
Rotolo (2000)	USA (individuals, towns)	1030 (10)	Socio-economic inequality (income, education and industry heterogeneity)						-b,i
Eckstein (2001)	Colombo District (individuals)	-	Class heterogeneity						-j
Gold et al. (2002)	USA (states)	39	Income inequality (Gini index of household income)						-g
La Ferrara (2002)	Tanzania (individuals, villages)	581 (-)	Wealth inequality (Gini index of household assets)					-a		Indiv. wealth and perceptions thereof
O’Connell (2003)	Europe (countries)	13	Income inequality (90/10 ratio)						-g
Bühlmann and Freitag (2004)	Switzerland (individuals, municipalities)	1288 (56)	Socio-economic inequality (heterogeneity index (not specified))						+ (ns)b
van Oorschot and Arts (2005)	Europe (individuals, countries)	28 894 (23)	Income inequality (20/20 group ratio)						+a
McVeigh (2006)	USA (counties)	6187	Income inequality (Gini index of family income)						+a
			Educational inequality (Gini index between educational categories)						+ (ns)a
Rupasingha et al. (2006)	USA (counties)	6094	Income inequality (mean/median household income)						- (ns)a
Mackerle-Bixa et al. (2009)	Europe (individuals, countries)	22 022 (18)	Income inequality (Gini index)					-b
Duncan (2010)	Canada (individuals, urban areas)	16 207 (4326)	Income inequality (Gini index)					+ (ns)b	+ (ns)b	Indiv. income
Kesler and Bloemraad (2010)	Rich countries (individuals, countries)	∼700 000 (∼50)	Income inequality (Gini index)					+ (ns)b
Hooghe and Botterman (2012)	Flanders (individuals, communities)	2080 (40)	Income inequality (75-25 ratio after tax)					∼0b,k
Karakoc (2013)	Postcommunist countries (individuals, countries)	16 726 (11)	Income inequality (Gini index of household income)					-/+b,l	-/+b,l	Indiv. income
Clark, A. K. (2015)	USA (individuals)	17 906	Income inequality (Gini index of household income)						- (ns)a
Wichowsky (2019)	USA (individuals, MSAs)	87 459 (30)	Income inequality (Gini index)					-b		Indiv. income
Studies on multiple forms of civic engagement
Costa and Kahn (2003b)	USA (individuals, MSAs)	42 134 (-) (volunt.)	Income inequality (Gini index of wages of full-time full year men aged 21-64)			-a		-a,m		Age
		7320 (-) (memb.)
Costa and Kahn (2003a)	USA (individuals, MSAs)	42 134 (-) (volunt.)	Income inequality (Gini index of wages)			-a		-a		Age, sex
		7320 (-) (memb.)
Okten and Osili (2004)	Indonesia (individuals, communities)	5262	Income inequality (Gini index of consumption)	-a	-a	+ (ns)a	-a
Uslaner and Brown (2005)	USA (states)	37 (charit.)	Income inequality (Gini index)	- (ns)a		+ (ns)a
		41 (volunt.)
Garcia-Mainar and Marcuello (2007)	Spain (individuals, communities)	4252 (-)	Income inequality (Gini index)	-a,n		-a,n		-a,n
Gesthuizen et al. (2008)	Europe (individuals, countries)	23 861 (28)	Income inequality (20/20 ratio)		-b		-b		- (ns)b
Pichler and Wallace (2008)	Europe (individuals, countries)	∼27 000 (27)	Income inequality (Gini index)				-(ns)b		-b	Social class
Geshuizen et al. (2009)	Europe (individuals, countries)	21 324 (28)	Income inequality (20/20 ratio)		-b		-b		-b
Fieldhouse and Cutts (2010)	USA and UK	US: ∼30 000 (∼41)	Income inequality (Gini index (USA)/social class fragmentation (UK))			0 (ns)b,o		0 (ns)b,o
	(individuals, communities)	UK: ∼15 000 (∼7000)
Saunders (2010)	OECD (countries)	11 (charit.)	Income inequality (Gini index)		-(ns)g			- (ns)g
		26 (memb.)
Smith, P. B. (2015)	World (countries)	128	Income inequality (Gini index)	- (ns)g		+ (ns)g

^a

single-level regression analysis.

^b

multi-level analysis including individual-level and contextual-level data and controlling for (unobserved) heterogeneity on the level that inequality is measured at. ^cEvers and Gesthuizen (2011) find a negative and significant effect of inequality on donations to activist organizations, a positive but insignificant effect on donations to interest organizations, and a negative but insignificant effect on donations to leisure organizations.

^d

coefficient significant on the forward sortation area (FSA) level but not the Census division (CD) level.

^e

results of single-level model confirmed in multi-level analysis, but these are not reported.

^f

positive coefficient on meshblock (∼100 people) level but negative (insignificant) coefficient of inequality on the area unit (∼2000 people) level.

^g

bivariate correlation.

^h

Fladmoe and Steen-Johnsen (2018) also report a negative relationship to the number of arenas, e.g. types of organizations.

ⁱ

when education heterogeneity, industry heterogeneity, income inequality and ethnic heterogeneity are included simultaneously only the coefficient for education heterogeneity is statistically significant.

^j

qualitative case study.

^k

Hooghe and Botterman (2012) additionally find no effect of income inequality on passive memberships and an insignificant positive effect on active memberships. ^lKarakoc (2013) find a negative effect for most levels of inequality, and a slight positive effect for very high values of inequality (Gini index above 0.39).

^m

Costa and Kahn (2003b) find a significant negative coefficient based on the GSS data which covers the years from 1974-1994; the coefficient is still negative but not statistically significant based on the American National Election Survey (ANES) which covers the years from 1952 to 1972.

ⁿ

Garcia-Mainar and Marcuello (2007) also report a negative relationship to the number of arenas, e.g. different types of organizations.

^o

coefficients not reported.

Note: N = number of observations on the individual and contextual level (in parenthesis); Inc. = incidence; Am. = amount; FSA = forward sortation area; MSA = metropolitan statistical area; PMSA = primary metropolitan statistical areas; SES = socio-economic status.

Most of the empirical studies are based on survey data and use regression analysis to estimate the effect of socio-economic inequality on civic engagement. Thus, one has to be cautious when interpreting the results of these studies, since effects might be confounded by numerous contextual factors, for example, redistributive measures taken by the welfare state or additional forms of inequality, which are not always controlled for. Therefore, our overview also indicates whether studies control for potential (unobserved) heterogeneity at the level that inequality is measured. If results from multiple model specifications are reported, we present the results of the full (largest) model. After summarizing the results of the empirical studies, we discuss limitations of the extant empirical understanding.

3.1 Charitable giving

Fourteen studies report on the relation between socio-economic inequality and charitable giving, including nine on the incidence and nine on the amount given. Almost all studies focus on monetary donations; only one examines the relationship between inequality and donating blood. All studies employ measures of income inequality (such as, e.g. Gini Index, Theil Index or 90/10 ratio), which are measured at the level of countries, federal states, municipalities or urban areas. They most often find a negative, albeit not always statistically significant, relationship regarding the incidence as well as the amount of giving to charities, except for three studies which report a positive relationship to the incidence of giving (Bielefeld et al., 2005; Evers and Gesthuizen, 2011) and the amount given (Payne and Smith, 2015) to charity. Oto-Peralías and Romero-Ávila (2017) additionally observe a negative relationship between persistent historical wealth inequality and the proportion of people donating blood in Spanish municipalities.

The suggested effects of socio-economic inequality on charitable giving are moderated by individuals’ income and education. Two studies revealing a negative main effect find that, in the USA, inequality particularly negatively affects charitable giving by better-endowed individuals (Duquette, 2018; Duquette and Hargaden, 2019). Payne and Smith (2015) find that the positive effect of inequality on giving which they observed is primarily driven by higher income and higher education neighborhoods.

3.2 Volunteering

Evidence regarding the relationship between socio-economic inequality and volunteering is provided by 22 studies and appears to be more mixed. Seventeen studies examine the effect of inequality on the incidence of volunteering, while seven studies examine the number of organizations one volunteers in. All studies employ measures of income inequality at various geographical scales as an indicator for socio-economic inequality.

Out of the seventeen studies analyzing the effect of inequality on the incidence of volunteering, seven report a negative relationship and nine studies report a positive relationship (one study does not report coefficients). Tellingly, most of the studies reporting a positive relationship report insignificant coefficients, which suggest that inequality’s effect on the incidence of volunteering is most likely small and thus hard to assess accurately. While the three cross-country studies analyzing European countries find a negative relationship, no clear pattern of results emerges from cross-country studies outside Europe. Similarly, there is little evidence of a clear geographic pattern in within-country studies, which often focus on within-country inequality in the USA, and show mixed results. Interestingly, however, most of the studies which do reveal a significant positive relationship analyze inequality on a small geographical scale, that is, at the municipal level (Oliver, 1999; Clark and Kim, 2012; Godfrey and Cherng, 2016), showing that different mechanisms might operate on different geographical scales. Regarding the amount of volunteering, all seven studies, out of which five analyze European countries, find a negative relationship with income inequality both across and within countries. In conclusion, evidence for a negative relationship across European countries seems rather established for the incidence and amount of volunteering, as indicated by several large multi-level studies. Results on within-country inequality and cross-country inequality outside Europe, though, are mixed.

As regards moderators, the studies identify income, age, ethnicity and sex as such. Several studies find that the participatory gap between the rich and the poor increases with higher economic inequality (Pichler and Wallace, 2009; Lancee and Van de Werfhorst, 2012; Filetti and Janmaat, 2018). Godfrey and Cherng (2016) find no evidence that the socio-economic status of families moderates the effect of economic inequality on youth volunteering, while under higher inequality, Asian-American and Black youth seem slightly more likely to volunteer than their White peers. Furthermore, the relation between inequality and volunteering is moderated by the volunteer’s age and sex. While inequality seems to affect volunteering among 25- to 54-year-olds, it does not seem to affect volunteering among those aged 65 or above (Costa and Kahn, 2003a, 2003b). Moreover, increases in inequality seem to affect men more than women (Costa and Kahn, 2003b).

3.3 Non-profit membership

In the assessment of inequality’s relationship to non-profit membership, the measures of socio-economic inequality are more diverse, including inequality of income, education, wealth and class. Regarding the incidence of non-profit membership, eight studies report a negative relationship with inequality (one of which is statistically insignificant), while two studies find a positive but insignificant relationship, one study finds a null effect and one study does not report the coefficients that are statistically insignificant. On examining the number of memberships, 11 studies find a negative relationship (three of which are statistically insignificant) and four report a positive relationship (three of which are statistically insignificant). Overall, the majority of studies reveal a negative relationship between socio-economic inequality and the incidence and number of non-profit memberships.

Karakoc (2013) provides further insight into this relationship by estimating the tipping point of a U-shaped curvilinear relationship between income inequality and the incidence and number of memberships. In their sample of post-Communist countries, this point lies at a Gini index of 0.39. While this insight is not able to explain the divergence of results we see between studies on non-profit membership, it demonstrates the importance of considering nonlinear effects and the absolute level of inequality.

Regarding moderators, Duncan (2010), La Ferrara (2002) and Wichowsky (2019) find that higher inequality particularly reduces non-profit membership among individuals with higher incomes, wealth or perceptions of higher wealth, while Pichler and Wallace (2009) and Karakoc (2013) find the opposite regarding social class and income. Inequality seems to affect membership among 25- to 54-year-olds, but not among people aged 65 or above, as also observed for volunteering (Costa and Kahn, 2003a).

Putting the results of all empirical studies together, there is prevailing evidence for a negative relation between socio-economic inequality and civic engagement. This result is rather consistent for the incidence and amount of charitable giving and non-profit membership. For volunteering, the negative relationship seems to hold for European countries, while there is less (consistent) evidence for countries outside of Europe. Two-thirds of the empirical studies provide a snapshot of the relationship through cross-sectional analyses, while about one-third of the studies analyze the relationship over time. On aggregate, both approaches reveal a negative relationship between socio-economic inequality and civic engagement, suggesting that countries and regions with higher inequality have lower levels of civic engagement, and that increasing socio-economic inequality similarly decreases civic engagement over time. As regards the moderators (income and education for charitable giving; income, age, sex and ethnicity for volunteering and income, wealth, social class and age for non-profit membership), we cannot assert a clear tendency with regard to the direction of their influence. For instance, while some studies find that, under higher inequality, high-resource individuals engage more, others find the opposite.

For several reasons, however, the empirical understanding provided by these empirical studies is rather narrow. First, about half of the empirical studies included in our review are based on data from the USA; another third is based on data from individual or multiple European countries and only few studies take an internationally comparative approach or focus on other countries, namely Canada, Indonesia, New Zealand and Tanzania. This bias is comparatively small given that about 60–80% of the overall published research on civic engagement stems from the USA (Shier and Handy, 2014; Ma and Konrath, 2018). Nevertheless, this distribution reveals a strong imbalance toward countries in the Global North, indicating low generalizability beyond these countries.

Second, another issue of geographical focus arises from the modifiable areal unit problem (MAUP). The MAUP refers to the fact that the scale and zonation of geographical units affect results when analyzing contextual effects (Petrović et al., 2019). Thus, careful consideration of the scale that the effect of socio-economic inequality might operate at should precede the selection of the analysis scale. In practice, however, most analyses are guided by data availability and consider only one level of geographical scale. An interesting exception is the analysis by Clark and Kim (2012), who discuss the difficulty in selecting appropriate geographical boundaries to capture an effect of inequality and provide some guidance on how to approach this question empirically.

Third, the majority of these empirical studies were published after 2000, which reflects the general surge in quantitative research into context effects during the late 1990s, spurred by the increasing availability of computing power and micro data (Petrović et al., 2019). Accordingly, most studies rely on data collected between 1990 and 2010. A notable exception to the limited time-coverage is the analysis by Duquette (2018), who provides a historical analysis of the relation between inequality and philanthropy in the USA between 1917 and 2012.

Fourth, the measures of socio-economic inequality used in empirical studies are almost entirely limited to measures of income inequality and the Gini index as a measure for income inequality. Thus, we do not know which part of the income distribution matters for the relationship between economic inequality and civic engagement. A related limitation is that many studies do not report whether their measure of income inequality captures inequality before or after redistributive measures are taken into account.

Lastly, a crucial limitation of many empirical studies is that they do not analyze the potential mechanisms underlying the relationship. Yet the influence of inequality on civic engagement ultimately depends on the mechanisms at work. In the next section, therefore, we work out the mechanisms that are commonly brought forward to explain this relationship.

4. Mechanisms explaining how inequality affects civic engagement

Almost all of the 70 studies reviewed offer some theoretical explanations for the effect of socio-economic inequality on civic engagement, and refer to a variety of factors and mechanisms rooted in different disciplines. Often, however, these explanations remain incomplete, confound elements of different schools of thought, or refer only to mechanisms at either the societal or the individual level. Through coding and condensing these explanations, we mapped them into five key theoretical approaches for explaining the relation between levels of socio-economic inequality and civic engagement (see Online Appendix B). We labeled these approaches (1) social disintegration hypothesis, (2) conflict hypothesis, (3) relative power hypothesis, (4) resource hypothesis and (5) inequality aversion hypothesis. The theoretically assumed pathways of each approach are depicted in the conceptual framework in Figure 2. Three approaches explain mechanisms for a negative relation between the level of inequality and civic engagement, while two approaches explain mechanisms that might underlie a positive relation. The five mechanisms consider the effects of socio-economic inequality at the societal or group level as well as the individual level, since both of these levels can be affected by inequality. Because charitable giving, volunteering and non-profit membership are not a completely homogenous set of behaviors, the mechanisms might apply to the three types of engagement to varying degrees, which we address where applicable.

Figure 1

Systematic literature search process.

Figure 2

Overview of mechanisms explaining the effect of inequality on civic engagement.

4.1 Social disintegration hypothesis

The link between socio-economic inequality and civic engagement is by far most frequently explained by building on sociological theories of social integration and social structure (Tomeh, 1969; Blau and Blau, 1982; Kennedy et al., 1998; Park and Subramanian, 2012; Collins and Guidry, 2018). According to this approach, resource inequalities give rise to status differentials between individuals, which increase social distance, social disorganization and segregation between groups, and in turn reduce social cohesion among community members. In other words, ‘inequalities undermine the social integration of a community by creating multiple parallel social differences which widen the separations between […] social classes, and […] [they create] a situation characterized by much social disorganization and prevalent latent animosities’ (Blau and Blau, 1982, p. 119). Accordingly, the effect of socio-economic inequality is not only considered a direct result of the distribution of economic and cultural resources, but is also seen as an outcome of processes of identification and group formation that shape individuals’ perceptions and motives.

Across the reviewed studies, the concept of (generalized social) trust is commonly used to capture how social disorganization, latent animosities and diminished feelings of community manifest on the individual level. Trust is assumed to be more likely to develop between people who can identify with each other and therefore to be found more often in homogeneous than in heterogeneous groups (Alesina and La Ferrara, 2000; Rothstein and Stolle, 2003; Armony, 2004; Uslaner and Brown, 2005; Rupasingha et al., 2006; Hommerich, 2015; Fladmoe and Steen-Johnsen, 2018). These explanations invoke ideas of homophily, that is, that people tend to associate with those with whom they share similarities (Rotolo, 2000; McPherson et al., 2001; Gesthuizen et al., 2009) because of an ‘aversion to heterogeneity’ (Alesina and La Ferrara, 2000). Moreover, inequality manifests and is reproduced in structural arrangements, for example, in the form of socio-spatial segregation into neighborhoods and schools (McVeigh, 2006). Such segregation reduces opportunities for inter-group contact, which results in fewer opportunities to build trust (Putnam, 2007; Rothwell, 2012; Crowley and Knepper, 2019). Lower generalized social trust, in turn, leads to lower levels of civic engagement (Rothstein and Stolle, 2003; Uslaner and Brown, 2005; Crowley and Knepper, 2019).⁴

Though almost all reviewed studies refer in some way to the approach of social disintegration when explaining the link between inequality and civic engagement, they emphasize different aspects of this chain of reasoning and use different terms to describe the mechanism. Studies that focus on arguments at the societal level make use of concepts such as ‘social disorganization’ (Collins and Guidry, 2018), ‘social distance’ (Tomeh, 1969; Duquette and Hargaden, 2019), ‘social segregation’ (McVeigh, 2006; Mastromatteo and Russo, 2017) and ‘social barriers’ (Gesthuizen et al., 2009). Studies focusing on arguments at the individual level, on the other hand, make use of concepts such as the ‘homophily principle’ (Gesthuizen et al., 2009; Fladmoe and Steen-Johnsen, 2018), ‘latent animosities’ (Collins and Guidry, 2018) or ‘status anxiety’ (Wilkinson and Pickett, 2009; Veal and Nichols, 2017), which are often assumed to result in lower levels of ‘(generalized social) trust’ (Uslaner and Brown, 2005; Gesthuizen et al., 2009).

While the arguments provided by the social disintegration hypothesis are used in most articles on volunteering and non-profit membership, they are less frequently used in articles on charitable giving. This difference stands to reason considering that the former types of civic engagement are more likely to involve personal contact with others and thus are more likely affected by individuals’ trust in others. In fact, Uslaner and Brown (2005), who were the only ones to test the proposed mechanism through generalized social trust, found this mechanism to be at work for volunteering but not charitable giving. Duquette and Hargaden (2019), however, also confirm the negative effect of inequality on charitable giving in an experimental setting that can be read as supporting the social disintegration hypothesis, as it renders other, more macro-oriented explanations unlikely.

4.2 Conflict hypothesis

About a dozen of the reviewed studies refer to (parts of) the conflict hypothesis, also referred to as the ‘conflict school’ (Karakoc, 2013, p. 198) or ‘conflict theory’ (Solt, 2008, p. 49). It assumes that higher levels of inequality cause higher rates of civic engagement because socio-economic inequality is translated into class conflicts and beliefs of injustice, which in turn drive civic engagement (Oliver, 1999; McVeigh, 2006; Karakoc, 2013; Fieldhouse and Cutts, 2010).

The conflict hypothesis argues that an increase in inequality increases divergences in political preferences between the haves and the have-nots. Such divergences most notably refer to redistributive measures. These become more attractive to the have-nots because rising inequality means that the have-nots become poorer compared with their fellow citizens, while redistributive measures simultaneously become more costly to the well-off individuals (McVeigh, 2006; Solt, 2008). This increase in incompatible political preferences particularly encourages the have-nots to organize (e.g. in a non-profit organization) and mobilize against the policies that favor the haves, aiming to bring about a change in society (Duncan, 2010; Karakoc, 2013). At the same time, it encourages the haves to enforce their opposition against any redistributive measures (Solt, 2008).

The mechanism linking these distributional conflicts to civic engagement on the individual level might be an increase in perceived injustice. Individuals, first and foremost the less affluent, find that political institutions do not represent their interests but instead consider these institutions responsible for socio-economic disparities, and realize that they have to become active themselves in order to create societal change (Karakoc, 2013). Thus, the conflict hypothesis proposes that an increase in socio-economic inequality will particularly increase more activist forms of engagement (e.g. volunteering or becoming a member in civil rights non-profits), which ‘draw together individuals with common interests and place them in confrontation with the state or with other opponents in society’ (McVeigh, 2006, p. 515).

In line with the conflict hypothesis, McVeigh (2006) has shown that the number of activist organizations in US counties increased with increasing income inequality. Among the reviewed studies, however, there is no direct empirical support for the whole mechanism suggested by the conflict hypothesis. Such evidence could be provided by showing that socio-economic inequality translates into perceptions of injustice and increased civic engagement in turn.

4.3 Relative power hypothesis

The second most frequently cited argument to explain the link between socio-economic inequality and civic engagement refers to power differentials. In short, the relative power hypothesis argues that resource inequalities result in power differentials (Blau and Blau, 1982; Uslaner and Brown, 2005; Collins and Guidry, 2018; Filetti and Janmaat, 2018). The haves become more powerful because resources can be used to influence others. Higher levels of power in the hands of the well-off, in turn, allow them to engage with issues relevant for them and to prevent issues relevant for the less well-off from becoming part of public debate (Solt, 2008). The have-nots, in contrast, withdraw from civic engagement due to a state of powerlessness. According to the relative power hypothesis, the negative effect on civic engagement is therefore expected to be stronger for the have-nots. At the other end of the spectrum, growing inequalities (particularly in wealth) place power in the hands of the rich, which these might also exert through civic engagement (Duquette, 2018; Reich, 2018).

The mechanism linking power differentials and civic engagement at the individual level is perceived powerlessness. If individuals cannot make their voices heard, it ruins people’s aspirations and confidence in political institutions and the law—it puts them in a state of (perceived) powerlessness (Blau and Blau, 1982; Armony, 2004; Solt, 2008). They eventually realize that the political system is incapable of defending their interests, and decide to abandon their civic engagement (Putnam, 1993; Filetti and Janmaat, 2018) because ‘the system is stacked against them’ (Uslaner and Brown, 2005, p. 876). In other words, they ‘opt out of civic and social engagement, so as not to waste their time and energy on an unfair system’ (Godfrey and Cherng, 2016, p. 2221).

Owing to its origin in political science (Schattschneider, 1960; Solt, 2008), the relative power hypothesis is most often used to explain the relation between inequality and volunteering or non-profit membership, which are often seen as adjacent to formal political engagement (Uslaner and Brown, 2005). Nevertheless, the mechanism has not been empirically tested in any of the studies reviewed.

4.4 Resource hypothesis

A total of 16 of the reviewed studies draw on elements of the resource hypothesis. The resource hypothesis also suggests a negative relation between socio-economic inequality and civic engagement, arguing that more equal societies also provide opportunities for civic engagement more equally across its members (Solt, 2008; Lancee and Van de Werfhorst, 2012; Karakoc, 2013).

In contrast to the other mechanisms, the focus of the resource hypothesis is not on processes that shape individuals’ perceptions and motives, but on the distribution of economic and cultural resources themselves. The distribution of these resources is primarily seen as the outcome of the organization of welfare states, such as labor market institutions and educational institutions, and therefore as the post-tax and post-transfer distribution of resources (Rothstein and Stolle, 2003; Lancee and Van de Werfhorst, 2012). These arrangements provide access to public goods, such as education in public schools or better employment conditions, in a more egalitarian manner, which increases opportunities to civically engage for the many (Duncan, 2010; Lancee and Van de Werfhorst, 2012; Rotolo and Wilson, 2014; Godfrey and Cherng, 2016; Veal and Nichols, 2017). The resource hypothesis has also been applied to the community level, with the argument that, with an increase in inequality, groups and organizations in communities with lower endowments have comparatively fewer resources—such as financial support, links to centers of power and leadership skills—to rely upon to foster civic engagement (Armony, 2004; Duncan, 2010).

With the lack of individual resources being the mechanisms at the individual level, the resource hypothesis primarily predicts a growing participatory gap between the haves and the have-nots (Verba et al., 1995; Karakoc, 2013). Those individuals with fewer resources find it increasingly difficult to engage, while professionalized associations increasingly target potentially wealthy contributors, which might increase their contributions (Skocpol, 2004; Reich, 2018). The resource hypothesis, however, has also been used to predict overall lower amounts of civic engagement, implicitly assuming that the lower engagement of have-nots cannot be offset by the higher engagement of better-endowed individuals. This explanation is therefore more likely to apply to volunteering and non-profit membership than charitable giving.

Lancee and Van de Werfhorst (2012) have demonstrated that the resource hypothesis indeed plays a role in explaining the negative relationship between socio-economic inequality and volunteering. However, studies that extensively control for the effects of contextual level as well as individual-level resources show that ‘something more’ must be going on (Alesina and La Ferrara, 2000; Gesthuizen et al., 2009; Lancee and Van de Werfhorst, 2012), such as the mechanisms proposed by the social disintegration or the relative power hypothesis.

4.5 Inequality aversion hypothesis

Among the reviewed studies, the inequality aversion hypothesis is the most common approach to explaining a positive effect of socio-economic inequality on civic engagement. It argues that higher inequality leads to a greater number of people in need, and that (some) individuals that are driven by inequality aversion, altruism or impure altruism (including altruism and the feeling of a warm glow) increase their engagement with the aim of addressing these needs (Andreoni, 1990; Batson et al., 1991; Fehr and Schmidt, 1999; Charness and Rabin, 2002; Mastromatteo and Russo, 2017; Nair, 2018; Duquette and Hargaden, 2019).

The point of departure for the inequality aversion hypothesis is that with higher levels of socio-economic inequality, even after redistributive measures are taken, economic and cultural resources are less likely to be provided in sufficient quantity or quality to meet the needs of society as a whole. It is then assumed that at least some individuals hold (impurely) altruistic preferences, such as inequality aversion or social welfare preferences (Fehr and Schmidt, 1999; Charness and Rabin, 2002; Derin-Gure and Uler, 2010; Nair, 2018; Duquette and Hargaden, 2019). Individuals holding these preferences would derive utility from civic engagement because they are concerned with addressing the given need for public goods (Andreoni, 1990; Smith, 2012; Mastromatteo and Russo, 2017). Individuals may then decide to contribute time or money to a non-profit organization (Garcia-Mainar and Marcuello, 2007; Crowley and Knepper, 2019) and particularly those who are above the average resource level (Nair, 2018; Duquette and Hargaden, 2019). Civic engagement should therefore increase with higher inequality.

The inequality aversion hypothesis assumes that individuals hold accurate views about the general (global, national or regional) distribution of resources and their own relative position within this distribution, which is often not the case (Gimpelson and Treisman, 2018; Nair, 2018). Further, to the extent that inequality materializes not only in resource heterogeneity but also group formation, the assumed positive effect is also dependent on the extent to which potential beneficiaries are perceived as in-group members (Batson et al., 1991; Nair, 2018): ‘most people are conditional altruists whose preferences are other-regarding to the extent that the other in question is a member of their in-group’ (Nair, 2018, p. 828). Against the backdrop of the social disintegration hypothesis, inequality aversion might therefore only be at work as long as inequality is not accompanied by segregation.

The inequality aversion hypothesis is considered by 18 of the reviewed studies, especially by those that focus on charitable giving, followed by those on volunteering. It is almost absent from the discussion of the relation between inequality and non-profit membership. This likely reflects expectations about the varying ‘redistributive component’ of these types of civic engagement, as well as the fact that economics, where this hypothesis originates, tends to focus on giving over volunteering and membership. While there is no direct test of this mechanism in the studies reviewed, the empirical studies show that the inequality aversion hypothesis does not seem to hold for civic engagement in general or that it is overpowered by other processes. In future studies, it might be worth examining this mechanism in relation to civic engagement for redistributive causes.

5. Discussion and conclusions

Although socio-economic inequality is commonly understood as an influencing factor for individuals’ civic engagement, there has been no synthesis of whether and through which mechanisms its effect might work. Reviewing a total of 70 studies on civic engagement in the forms of charitable giving, volunteering and non-profit membership, we have synthesized results from empirical studies and configured the suggested underlying theoretical explanations in an overarching conceptual framework.

5.1 Results of empirical studies

Our SLR reveals that higher levels of socio-economic inequality are most often negatively related to civic engagement. In most contexts, the incidence as well as the amount of charitable giving, volunteering and non-profit membership are negatively affected by inequality. Findings of a negative relationship are most homogenous for charitable giving, non-profit membership and the amount of volunteering, while evidence is less conclusive regarding the incidence of volunteering. These empirical findings augment critiques that question the purpose and power of civic engagement to reduce socio-economic inequality (Skocpol, 2003; Giridharadas, 2018; Reich, 2018). Instead of tackling inequality, it seems that civic engagement itself is undermined by increasing inequality, and that the manifold public goods provided through civic engagement are at stake.

In addition, our review shows that several individual factors appear to moderate the effect of socio-economic inequality on civic engagement. The most prominent factor is the position one takes in the structure of inequality, as measured by individual income, education or class. The findings regarding these moderators, however, are inconclusive. While some studies observe that high levels of individual income, education or class dampen the effect of inequality on civic engagement, others find the opposite.

Our review also demonstrates that the explanatory power and generalizability of the extant empirical knowledge is quite limited. One of its apparent limitations is the imbalance toward the analysis of countries of the Global North and a lack of comparative cross-country research. So far, we do not know why inequality might have a negative effect on civic engagement in some countries or regions but a positive effect in others. This might, for example, depend on interactions with other dimensions of inequality (e.g. ethnic heterogeneity and gender inequality) (Kesler and Bloemraad, 2010; Andreoni et al., 2016; Dale et al., 2018) and the absolute level of inequality (Karakoc, 2013). Further, many studies suffer from an atheoretical approach to the selection of the analysis scale, that is, the level that inequality is examined at. To explain the relationship, however, it is essential to consider which mechanisms are assumed to be working at the international, national or local level. Finally, since most studies focus on the relationship between inequality and civic engagement within one society, the vast inequalities that exist between countries certainly warrant further investigation of the relationship beyond local or national borders (e.g. Nair, 2018).

A common methodological limitation of the almost exclusively observational empirical studies is the existence of a vast number of potentially omitted factors that might confound the relationship between socio-economic inequality and civic engagement; a limitation which extends into our synthesis. These factors include, among others, a region’s absolute wealth, a country’s democratic history as well as redistributive measures taken by the welfare state. The most important limitation of previous research, however, is the sparse theoretical conceptualizations of potential mechanisms underlying the effect of socio-economic inequality on civic engagement.

5.2 Theoretical approaches and future research

The second part of our study has mapped and put into context the arguments put forward to explain how socio-economic inequality affects civic engagement, resulting in five main explanatory approaches. While each of these explanations delineates a specific pathway through which socio-economic inequality may act, their functioning is not mutually exclusive. As can be seen from Figure 2, three of these explanations propose a negative relation between inequality and civic engagement, namely the social disintegration hypothesis, the resource hypothesis and the relative power hypothesis. The social disintegration hypothesis argues that socio-economic inequality leads to a decrease in social cohesion, which in turn lowers the generalized social trust of individuals, and subsequently leads to reduced civic engagement. The resource hypothesis claims that people are better able to civically engage in more equal societies, which is particularly relevant for the group of the have-nots. The relative power hypothesis, on the other hand, refers to power differentials in society and supposes that increasing power differentials leave have-nots in a state of powerlessness, which undermines their civic engagement. Taken together, these three hypotheses would predict that inequality strongly reduces civic engagement by the have-nots, and that lower social integration might also reduce civic engagement by the haves.

The remaining two approaches, the conflict hypothesis and the inequality aversion hypothesis, suggest a positive relation. The conflict hypothesis proposes that inequality results in power differentials (and distributional conflicts), which manifests in perceptions of injustice on an individual level, and leads people to get active themselves and increase their civic engagement. Finally, the inequality aversion hypothesis suggests that inequality triggers civic engagement because some individuals are driven by altruistic concern for the welfare of others. The conflict and inequality aversion hypotheses thus complement each other in predicting that inequality increases engagement for redistributive causes by the have-nots (particularly volunteering and membership) and by the haves (particularly charitable giving), respectively.

On an aggregate level—without considering heterogeneity in how individuals might be affected by inequality—there are thus only two contradictions in these mechanisms’ predictions. Firstly, the conflict hypothesis’ predictions of overall increased engagement contradict those by the other hypotheses, except for the inequality aversion hypothesis. Second, the inequality aversion hypothesis is at odds with the social disintegration hypothesis, although the former might be better applicable to charitable giving and the latter to volunteering and non-profit membership. On an individual level, however, we are likely to observe several mechanisms to be at work at the same time, for example, a lack of resources and low engagement by some, and inequality aversion and high engagement by others. If future studies were to find these mechanisms to be working in parallel, they would sketch more unequal societies as characterized by decreasing civic engagement by the many, but potentially increased engagement of groups that are in confrontation with each other and potentially increased ‘elite philanthropy’ by the haves.

By configuring the mechanisms through which socio-economic inequality affects civic engagement, our theoretical framework enables further research to overcome several limitations of current empirical research. First, future analyses will have to consider underlying mechanisms to design their empirical approach and to overcome the MAUP. Only if we know what mechanism we are examining can we adapt our analysis to identify spatial scales, specific groups and causes of engagement for which these mechanisms might operate (Falleti and Lynch, 2009; Clark and Kim, 2012; Petrović et al., 2019). For example, while the mechanism of social distance and distrust is likely to operate at a local scale, the mechanisms proposed by the resource hypothesis are likely to operate at the level of nation states. Considerations of spatial scale, however, are absent from almost all included studies. Second, identifying these underlying mechanisms will be key to understanding how particular types, fields and activities of civic engagement might be affected differently. It might seem trivial to note that determinants of civic engagement vary with the specific type and field of engagement in question, since this has been confirmed for numerous individual factors (Ackermann, 2019; Neumayr and Handy, 2019). However, this is of particular importance with regard to the impact of socio-economic inequality because it transforms the very structure of society that gives rise to civic engagement in the first place. For example, while the relative power hypothesis and the resource hypothesis would predict that inequality generally decreases civic engagement, the conflict hypothesis and inequality aversion hypothesis primarily predict more engagement for (or against) redistributive causes. Third, since most of the mechanisms assume that inequality shapes individuals’ perceptions and motives, a conceptualization of proposed individual-level mechanisms serves as a basis for future research to integrate these assumptions into their empirical design (Kenworthy and McCall, 2007; Nair, 2018). Lastly, if we know what mechanisms might be at work, we can tackle the question of the circumstances in which socio-economic inequality might have a positive or negative impact on civic engagement. For example, if the conflict hypothesis is right in predicting that inequality increases (specific forms) of civic engagement because of increases in perceived injustice, the effect of inequality should depend on the extent to which it is perceived to be based on meritocracy (McVeigh, 2006).

5.3 Concluding remarks

While this study has mapped the theoretically assumed pathways in a conceptual framework that synthesizes the hypothesized effects of socio-economic inequality on both the contextual and individual level, none of these mechanisms have been thoroughly tested by any of the studies in our review. Going forward, testing these mechanisms will be the major challenge for empirical studies in this field of study. Though socio-economic inequality may affect individuals’ civic engagement through more than the five mechanisms we present, our framework provides an essential basis for future studies by indicating the pathways and variables necessary to study underlying mechanisms. In our view, it will be key for future research to (a) pinpoint the contextual factors that the hypotheses assume to be affected by socio-economic inequality, (b) assess the accuracy of the mechanisms’ microfoundations—that is, assessing to what extent these factors actually shape individuals’ perceptions and, accordingly, their motives for engaging in different causes and (c) account for how engagement by different groups and for different causes are affected differently. To acquire insights into the workings of each of the proposed mechanisms, our framework calls for future research including qualitative studies, longitudinal multi-level studies incorporating cross-level mechanisms and experimental studies examining real-world civic engagement.

Footnotes

1

The review process was guided by a review protocol (see Online Appendix A), specifying each of these three steps following Briner and Denyer (2012).

2

This implies that studies focusing on broader concepts (e.g. social capital or public goods provision) were only included when at least one of the three types of formal civic engagement is examined explicitly.

3

Many of these studies have their main focus on another topic than inequality and civic engagement (e.g. they only include inequality as one of many control variables) and were thus identified via snowballing rather than the database search.

4

Assumptions diverge on whether these mechanisms diminish out-group trust but foster in-group trust and accordingly civic engagement, or whether both are negatively affected by inequality (Putnam, 2007; Gesthuizen,et al., 2009; Park and Subramanian, 2012; Crowley and Knepper, 2019).

Acknowledgements

The authors wish to thank the editor, two anonymous referees and Astrid Pennerstorfer for their constructive and insightful comments. They also acknowledge valuable suggestions from participants at the 2019 Conference of the European Research Network on Philanthropy (ERNOP), the 2019 Annual Conference of the Association of Research on Nonprofit Organizations and Voluntary Action (ARNOVA) and participants of the Philanthropy Research Workshop hosted by the Lilly Family School of Philanthropy at Indiana University in fall 2020.

Supplementary material

Supplementary material is available at SOCECO Journal online.