Who is Interested in Participating in Participatory Budgeting?

Abstract

Inequalities in terms of who participates in politics yield policy outcomes that fail to reflect the interests of the broader public. Because these processes fail to engage the full citizenry in political decision-making processes, they are also markers of an anemic civic culture. Advocates of participatory budgeting (PB) – a process implemented at the subnational level in thousands of cities in the United States and beyond that invites residents to participate directly in the process of allocating public resources for local projects – argue that it can alleviate these inequalities. They argue that features of the PB process make it ripe for engaging new participants in the political process and weaving a more inclusive social fabric. We examine the correlates of interest in participating in PB using a survey of Cook County residents. We also consider the extent to which the policy priorities of those who are interested in participating diverge from those who are less interested. Although we find evidence that the process is particularly appealing to younger respondents and those who identify as Latine or Black (as opposed to White), we also find that interest is higher among those with higher socioeconomic status and those who perceive conditions in their neighborhood to already be good. Our evidence also suggests that inequalities in who is interested in participating may not radically affect policy outcomes. However, those who decline to participate cannot reap the broader social and political benefits advocates hope the PB process can foster.

Introduction

The fact that patterns of political participation – particularly in electoral politics – are unequal in the United States is well-documented (e.g. Lijphart 1997; Verba, Schlozman and Brady 1995). Furthermore, the preferences of those who choose to participate are better represented in policy-makers’ behaviors and policy outcomes than those who do not (Griffin and Newman 2005; Hill and Leighley 1992). This poses a normative problem in a democratic polity that purportedly rests on a principal of all voices being counted equally: if some residents do not make their voice heard, their interests are less likely to be prioritized by elected officials.

Advocates argue that participatory budgeting (PB) – where residents offer proposals for how public funds should be spent in their local community, debate the merits of those proposals, and ultimately vote directly on identified projects that prove feasible to implement – offers a promising avenue for addressing these inequalities. These advocates claim that the process allocates funds in ways that are more democratic and that reduce disparities in access to public resources (Hagelskamp et al. 2018). Beyond this, PB may draw new participants into the political process and yield broader social benefits. By inviting residents to identify needs in their communities and work with others to address those needs, PB has the potential to educate politically disengaged segments of the public about budgeting, public administration, and other political processes, to revitalize civil society in American cities, and perhaps even improve community health outcomes (Biron 2020; Colin 2022; Hagelskamp et al. 2018; Touchton, Sugiyama and Wampler 2017; Wampler, McNulty and Touchton 2021).¹ In short, PB has the potential to not only affect substantive policy outcomes – how resources are allocated in communities – but also facilitate an array of favorable social outcomes by drawing new participants into the political process (Baiocchi 2005; Wampler 2010).

However, in order to reap these broader benefits, people must choose to participate. In this article we consider two research questions. First, what factors are associated with interest in participating in the PB process? We are particularly attentive to whether the democratic, community-focused features of the PB process are enough to appeal to a different pool of residents than those who engage in other forms of political participation. Our second research question asks, to the extent that interest in participating is unequally distributed, are these patterns likely to affect the substantive policy outcomes these processes produce?

Existing work on PB participation has relied heavily on surveys of people who have already chosen to participate (e.g. Ganuza and Franćes 2012; Herian 2011; Mundt 2017). Here we report findings from a descriptively representative survey of residents of Cook County, home to Chicago where the first PB processes in the United States were implemented in the city’s 49th Aldermanic Ward in 2009 (Gilman 2016; Knutson 2016). Currently, individual Chicago alderpersons can opt to use PB as a way to allocate funds earmarked for their ward. Approximately 10 of Chicago’s 50 aldermanic wards, as well as Evanston, IL – a Cook County suburb North of Chicago – have used or currently use PB to allocate a share of public funds. This makes Cook County an attractive context to study attitudes about PB: the prospect of expanded opportunities to engage in PB is very real. Additionally, although most residents have not had an opportunity to directly participate in PB – and a small share of those who could, actually do participate – many residents may have heard about the process through local media outlets. Indeed, 30 percent of our respondents reported having heard about PB, a share we suspect is far higher than would be found in a national survey. We return to these respondents in the final section of this article.

The survey included questions designed to gauge residents’ interest in engaging in various stages of the PB process. It also included an open-ended question that asked respondents how they would spend $1,000,000 in their community – a question that mirrors a message Chicago PB organizers told us they use in their mobilization efforts. Thus, in contrast to in-depth interviews or surveys of those who have already opted into the PB process, we offer insights into the broader “lay of the land.” Are the factors that lead to interest in participating in PB different from those that predict other forms of political participation? Do those who are particularly inclined – or disinclined – to participate have distinctive priorities when it comes to how public funds should be allocated?

We find that individuals with higher incomes and levels of educational attainment report greater interest in participating in the PB process – a pattern that mirrors the relationship between socioeconomic status and more traditional forms of political participation including voting, attending public meetings, and contacting elected officials. This said, after accounting for socioeconomic status, other groups that are often underrepresented in political processes – young, Black, and Latine respondents – express greater interest in participating than their older counterparts or respondents who identified as White.

We also assess the extent to which people who view their communities as in need of improvements report greater interest in participating in a process of allocating resources to community projects. Contrary to expectations, even after controlling for respondents’ demographic characteristics, we find that those who view their neighborhood as in better shape are more interested in participating in PB. Finally, we consider the extent to which residents are likely to view their neighbors as desirable decision-making partners is associated with interest in engaging in PB. Surprisingly, those living in areas where many of their neighbors share their ethnoracial identity report lower levels of interest in participating in PB. When we, instead, rely on reported levels of trust in one’s neighbors to measure social affinities, we find that trust is a potent predictor of interest in PB.

Taken together, these findings suggest that PB advocates’ goals of recruiting a representative pool of participants are unlikely to be realized by an “if you build it, they will come” approach. The process does not appear to be inherently more attractive to economically disadvantaged residents, those who view their neighborhoods’ needs as particularly acute, or those living in communities where many of their neighbors share their descriptive characteristics. The patterns we report underline the challenges of breaking cycles of anemic civic engagement. Attracting participants to novel participatory opportunities may be easiest in contexts where conditions are most favorable: where individuals have time and resources to participate and already harbor positive feelings about their neighborhood’s physical and social landscape.

We emphasize that our findings do not demonstrate that advocates’ goals cannot be realized. Rather, they are consistent with the notion that democratizing participation will require intentional, directed efforts to mobilize groups who are not already inclined to participate or living in contexts that already foster engagement – efforts that existing work demonstrates can work in the domain of PB processes (Gilman 2016; Godwin 2018; Gordon, Osgood Jr, and Boden 2016; Kasdan and Markman 2017).

We also leverage responses to an open-ended question to offer preliminary evidence regarding the extent to which patterns of interest in PB are likely to affect the substantive outcomes the PB process yields. Our tentative evidence offers some reassurance in that those who are interested in participating in PB have similar priorities in terms of the broad types of projects they think should be funded. However, many PB advocates argue that the greatest potential benefits of PB flow from the experience of participating, rather than the substantive policy outputs the process yields.

Costs and benefits of participating in participatory budgeting

As discussed above, PB advocates argue that the process can disrupt patterns of inequality that emerge with other forms of political participation. Existing research offers mixed evidence regarding the extent to which PB processes have broadened patterns of participation by attracting participants who better represent the community as a whole. One study fielded in Belgium found evidence that groups that, historically, participated at lower rates are more interested in engaging in the PB process (Talukder and Pilet 2021). Others have found evidence that, at least in some circumstances, PB can draw in participants from traditionally marginalized groups (Kasdan and Markman 2017). For example, poor women living in fairly homogeneous contexts in Brazil appear to have been effectively drawn into the PB process – perhaps helped along by high levels of social trust (Touchton, Sugiyama, and Wampler 2017; Wampler 2007; 2015).² Encouragingly for advocates, there is also evidence that those who participate in the PB process become more aware of issues affecting their communities, attuned to how local government processes work, and trusting of local government (Collins 2021; Crum, Sallnas, and Weber 2013; Gregořcǐc and Krǎsovec 2017).

However, there is also reason to doubt that the process has truly democratized participation. Drawing on interviews with PB practitioners, Pape and Lerner (2016) find that efforts to increase the equity of the PB process face a number of challenges. There are instances where people who usually participate at lower rates do not show particular interest in PB and the design of a PB process can leave out vulnerable populations (Ganuza and Franćes 2012; Wampler and Touchton 2019; Wampler, McNulty, and Touchton 2021). For example, in Spain, as in many other European countries, the process moved to a digital format. This change attracted the youth at higher rates but left out the elderly (Franćes, Carratalá, and Ganuza 2018).

Although there is some evidence that those who participate in more labor-intensive planning and discussion phases of the PB process are descriptively representative of the pool of eligible residents, those who vote on PB proposals tend to be more affluent, better educated, and more likely to be homeowners (Lerner 2011; Pape and Lim 2019). Thus, the “usual suspects” – as Gilman (2016) refers to more broadly politically active residents – may often dominate the PB process (Holdo 2016). At a more basic level, rates of participation in the PB process are often quite low. In Spain, participation rates are between 1 and 3 percent (Ganuza and Franćes 2012) – similar to participation rates reported in Chicago wards. Averaged over a six-year period, less than 2 percent of eligible New York City residents voted on PB proposals (Kuenneke and Scutelnicu 2021) – a far lower rate of participation than found in municipal elections.³

One possibility is that, like other forms of participation, patterns of interest in – and engagement with – the PB process can be understood by considering the resources needed to participate, as well as the benefits of participating (Verba, Schlozman, and Brady 1995). Engaging in a PB session is certainly less costly than running for an office where one might wield substantial control over the budgeting process. However, it is not cost-free and, importantly, the costs of participation are likely to be experienced differently by different groups. Those with less flexible or more demanding work schedules, parents who cannot afford child care, and those with limited access to transportation may find it difficult to participate in lengthy PB sessions – especially if they are held at venues that may be difficult to reach (Muthomi and Thurmaier 2021; Wampler, McNulty, and Touchton 2021). In short, the costs of participation in the PB process may be particularly onerous for many of the groups PB advocates are eager to mobilize.

Researchers and advocates recognize the challenges these costs may create. For example, Stewart et al. (2014) suggest that providing childcare and transportation may be necessary to mobilize lower-income residents. Similarly, there is evidence that people report greater interest in participating in PB sessions if the meetings would be held online (Abbott and Touchton 2022; Wampler, McNulty, and Touchton 2021).

These costs of participation may be counterbalanced by the distinctive benefits of participating in the PB process. From an instrumental perspective, classic models of the “calculus of voting” (Riker and Ordeshook 1968) draw attention to the poorly defined benefits of election results. Consider the benefits a voter might expect to reap if their preferred candidate wins a congressional race. The power that elected official has to implement policy is likely to be sharply constrained. They must work with others in their chamber to find a policy that not only they agree on but can also pass the other chamber, be signed into law by the president and, potentially, survive judicial review. More cynically, once in office, elected officials may fail to act on their campaign promises. Further compounding the murky connection between a congressperson’s stated policy goals and policy outputs is the fact that national policies are typically complex and the connections between policy proposals at the national level (e.g., changes to the tax code or environmental regulations) and an individual voter’s well-being are often unclear. Even at the local level, it is often difficult to draw a straight line between one candidate’s electoral victory and expected policy outcomes. In short, from an instrumental perspective, it is often unclear what exactly a voter can expect to “get” from voting in an election.

In contrast, PB processes typically focus on local issues of immediate and concrete concern to participants, including maintenance of infrastructure, improvements to parks, and public safety efforts. What is at stake in these sessions is likely to resonate with many residents who may care a great deal about whether sidewalks on their block are repaired or local parks are attractive sites for children to play. Additionally, the outcomes of PB processes tend to lead to policy outcomes that are more predictable than those associated with the outcome of a given electoral campaign. Indeed, in some cases decisions participants make are binding on policy-makers. Thus, the instrumental benefits of participation in this process are relatively clear and may come in the form of something immediately observable like a renovated park or newly paved alley.⁴

Existing research demonstrates the importance of the nature of the instrumental benefits PB processes promise. For example, in line with Abbott and Touchton (2022), Collins (2021) finds that creating a space where binding decisions can be made increases the likelihood that residents will participate in PB meetings, compared to simply offering an advisory type of input. In a similar vein they find that respondents reported higher levels of interest in participating when the size of the budget PB participants could access was larger.

Beyond the enhanced instrumental benefits of participating, the PB process may offer distinctive expressive and solidary benefits. Common participatory acts like signing a petition, contributing to a candidate for office, and voting are primarily solitary acts. In contrast, PB sessions are inherently communal, offering participants an opportunity to express their views regarding the most pressing needs of their communities in a social setting with their neighbors. Indeed, advocates often point to these features of the PB process as crucial, noting that these collaborative processes are likely to build social capital within communities and stimulate a virtuous cycle where participants become more willing to work with their neighbors to take steps to improve their communities (e.g. Cabannes 2015).

Who is likely to be interested in participating?

In our analysis, we leverage the framework discussed above to answer our first research question by examining the correlates of interest in engaging in the PB process. We consider demographic correlates (focusing on those often associated with inequalities in more traditional forms of political participation), material self-interest, and social affinities with others in one’s area.

Demographics

A large body of research has documented the relationship between socioeconomic status and political participation in the United States. For example, individuals who report higher levels of educational attainment or income tend to vote at higher rates than those with less formal education or lower incomes. These patterns are often attributed to inequalities in the experienced costs of participating. Those with higher levels of educational attainment and income may tend to have more flexible work schedules, the ability to pay for child care while they participate, or easier transportation options. Absent special efforts (e.g., provision of cost-free child care during sessions), PB processes are unlikely to mitigate these costs. In fact, some stages of participation – for example, serving on a PB committee – are likely to be far more costly than voting.

Higher socioeconomic status is also associated with greater access to political information and improved confidence in one’s ability to make sensible judgments about often-complex political debates. Consistent with this, some studies find that residents with high levels of educational attainment are notably more likely to participate in PB than less-educated residents (Mundt 2017). However, because the PB process focuses on priorities for spending on concrete projects in local communities, information disparities may be an unlikely avenue whereby socioeconomic status could affect eagerness to participate in PB processes. We note that the benefits associated with the PB process may also mitigate the relationship between socioeconomic status and interest in participating in PB. Quality infrastructure, parks, and other public goods may be particularly important to individuals with limited personal resources. Similarly, to the extent that PB stimulates more equitable redistribution of resources, it may attract lower SES participants (Wampler 2012). This said, we expect to find a positive relationship between socioeconomic status and interest in participating in the PB process. With the exception of voting on final proposals, engagement in the PB process is more costly than turning out to vote or signing a petition (or perhaps even taking on the responsibility of circulating a petition). These costs are likely to be felt in a particularly pronounced way by lower-income and less-educated individuals.

H_1A: Higher socioeconomic status is associated with greater interest in participating in PB.

A second demographic factor we consider is ethnoracial identity. Pape and Lim (2019) find that Hispanic and African American residents are underrepresented among Chicago residents who vote in the PB process. Similarly in the case of PBNYC, representation of non-White residents is no better than it is in elections (Kuenneke and Scutelnicu 2021; Lerner and Secondo 2012; Stewart et al. 2014). However, we note that ethnoracial identity remains closely related to socioeconomic status in the United States – a pattern that may help to explain aggregate differences in participation rates across groups.⁵ We expect that once we account for this relationship, non-White residents will be particularly drawn to the PB process as it serves as an alternative to traditional forms of (primarily electoral) participation that have historically disadvantaged non-White Americans.

H_1B: After accounting for other demographics, non-White residents report greater interest in participating in PB.

A final demographic characteristic we consider is age – an important characteristic to consider, given that some PB processes invite residents who are too young to vote in regular elections to participate. Although older individuals are substantially more likely to vote in the United States, young people appear to be eager to participate in other ways (e.g. Strama 1998). There is also some evidence that young people are particularly attracted to PB (e.g. Talukder and Pilet 2021). This attraction to alternative forms of participation may be rooted in young people’s cynicism about politics (e.g. Southwell 2003: but see Zukin et al. 2006). In short, participatory budgeting may be particularly appealing to young people because it shifts power away from “untrustworthy” elites. Thus, once we control for other demographic characteristics, we expect young people to report higher levels of interest in PB than older residents.

H_1C: After accounting for other demographics, younger people report greater interest in participating in PB.

Material self-interest

In addition to demographic characteristics, we consider the role material self-interest may play in shaping interest in PB.⁶ Scholars have offered philosophical arguments that self-interest has an important role to play in political deliberation (Mansbridge et al. 2010). However, researchers have found surprisingly mixed evidence regarding the extent to which self-interest shapes political outcomes. Early work found that symbolic considerations (e.g., personal values) tend to outweigh material interests in shaping political attitudes (e.g. Sears et al. 1980; Sears and Funk 1990), but more recent studies have demonstrated that when the stakes of a policy are clear and immediate, self-interest can substantially affect attitudes (Erikson and Stoker 2011; Margalit 2013). Additional evidence finds that, even when self-interest does not affect attitudes, those with a material stake in a policy are often more likely to take actions like participating in protests tied to that policy (Green and Cowden 1992).

Participatory budgeting invites residents to engage in a process where the material stakes of the outcome are direct and fairly immediate. These processes may yield smoother roads, new or refurbished parks for children to enjoy, and other projects that affect material conditions in one’s neighborhood. Although all residents – regardless of baseline conditions – can expect the process to yield improvements, we posit that these improvements are likely to be experienced as particularly consequential by those who perceive conditions in their neighborhood to be poor. Repaving a road that is in extremely poor condition is more consequential than more cosmetic repairs; the prospect of building a new playground may feel like a bigger deal than updating equipment at an already functioning playground. In short, we expect that, after controlling for demographic characteristics that may affect the costs of participation in PB, those who report that conditions in their neighborhood are particularly poor will indicate greater levels of interest in engaging in the PB process. This said, we acknowledge that the opposite pattern may emerge. If favorable local conditions reflect residents’ ongoing efforts to tend to community needs, participatory budgeting may be particularly attractive to those living in areas where conditions are good.

H₂: After accounting for demographics, those who rate conditions in their neighborhood less favorably report greater interest in participating in PB.

Shared identities and trust

The final factor we consider in addressing our first broad research question pertains to respondents’ social contexts. We consider two measures that capture the extent to which people may anticipate that other PB participants will be constructive partners: shared ethnoracial identity and interpersonal trust.

People’s tendency to forge and maintain trusting social bonds with others who share their demographic characteristics is well-documented (McPherson, Smith-Lovin, and Cook 2001). Given the collective nature of the PB process, there is reason to suspect that people will have more favorable expectations about – and thus be more inclined to participate in – the PB process if they believe others who participate are likely to be “like them.” One approach we take in the analysis we present below focuses on homophily tied to ethnoracial identity – arguably the “biggest divide in social networks today in the United States” (McPherson, Smith-Lovin, and Cook 2001, 420).

Attraction to working with others who share one’s characteristics may be reinforced by characteristics of the PB process. For example, Wampler, McNulty, and Touchton (2021, 139) argue that “consensus-based voting is susceptible to elite domination while the secret ballot allows more vulnerable voters to express their preferences.” Thus, people’s tendency to seek out relationships and experiences with others who share their characteristics may be amplified by the open nature of the PB process. Below we assess whether those who live in areas where a larger share of other residents share their ethnoracial identity report greater interest in participating in the PB process.

H_3A: After accounting for demographics, those living in neighborhoods where a larger share of residents share their ethnoracial identity report greater interest in participating in PB.

An alternative way of testing these theoretical expectations is to use a measure of interpersonal trust. Although measures of shared identity have the advantage of being “objective,” they require us to posit that people will tend to be more inclined to trust and, thus, work with those who share their descriptive characteristics. Reported trust in one’s neighbors is a more subjective measure. However, it arguably captures the concept at the heart of the dynamic we set out to test more directly: those who feel they can trust other prospective participants will be more inclined to participate in the PB process.

H_3B: After accounting for demographics, those who say they trust their neighbors report greater interest in participating in PB.

Interest in participatory budgeting and project priorities

A growing body of work has examined how PB affects policy outcomes. Findings on this front have been mixed. Many studies find that the effects of PB on patterns of government spending are more limited than advocates may hope. For example, Spada (2009) finds that the policy consequences of PB are concentrated in the first few years after it is implemented. Kuenneke and Scutelnicu (2021) find that PB did not yield higher levels of redistribution in districts with lower socioeconomic backgrounds. Calabrese, Williams, and Gupta (2020) studied the PBNYC program and argue that the potential policy effects of PB processes may only manifest if decision-making power is truly in the hands of the public – a condition not met in New York where PB decisions are not binding and final spending decisions are made by legislators. Under these conditions, the authors find little evidence that PB yields identifiable differences in spending, compared to patterns observed in districts without PB.

Below we build on this line work. Rather than considering whether PB leads to different outcomes than would be realized via more traditional processes, our second research question considers whether those who express greater interest in participating report different spending priorities than those who express less interest. If, for example, higher SES individuals are more interested in participating and prioritize different types of projects, this would suggest that PB processes are susceptible to the same inequalities in representation found elsewhere in the political system in the United States (Bartels 2016; Gilens and Page 2014; Page, Bartels, and Seawright 2013). In contrast, if interest in participating is unrelated to spending priorities, inequalities in who participates in PB processes may be less problematic.

H₄: Interest in participating in PB predicts spending priorities.

Data and analysis

Our data are from a survey of residents of Cook County, Illinois, the second most populous county in the United States and home to Chicago – a pioneer in PB.⁷ Survey respondents were recruited by Dynata from their opt-in panels to complete the survey online, with the goal of achieving a sample with demographic characteristics that mirror those of the Cook County population – a process that was broadly successful (see Supplementary Table S1 for summary statistics).⁸ We restrict our analysis to the 1,446 (out of 1,491) respondents who provided responses to all of the items used in our analysis.

The battery of questions at the heart of our analysis asked respondents to rate their level of interest in participating in each of four stages of PB – submitting project ideas; participating on a planning committee; attending a town hall meeting; voting on which projects to fund – on a 4-point scale ranging from not at all interested to very interested. Full question wording is provided in Supplementary material S2. We report the distribution of responses to each question in Table 1.

Table 1. Reported interest in participating in each stage of participatory budgeting process

Note. Cell entries are percentages.

The distributions of responses in Table 1 indicate that not all respondents view PB as an attractive process to engage with. Approximately half of respondents said they were either “not at all” or only “a little” interested in participating in the first three stages of the process. The first two of these stages – submitting ideas and working on committees – are where the agenda is set regarding which local projects will be considered. Roughly one out of four respondents said they were “very” interested in using their voice at these stages. Interest in the less costly act of voting on proposals was notably higher: 73 percent of respondents said they were either “somewhat” (29 percent) or “very” (44 percent) interested in voting on projects.

Demographic correlates of interest in participating

Which individual-level and contextual factors predict interest in participating in PB? We begin by reporting results from a series of Ordinary Least Squares (OLS) models, predicting interest in each form of participation with respondents’ educational attainment, family income (as well as an indicator for respondents who declined to report their income),⁹ ethnoracial identity, and age. We also control for respondents’ reported gender and whether the respondent’s ZIP code indicated that they reside in Chicago. We report estimates from these models in Table 2. Note that, with a handful of exceptions, the relationships that emerge are quite similar across these four outcome measures. The results contradict the notion that PB is distinctively appealing to lower socioeconomic status residents and support H _1A . The relationship between education and reported interest in participation is positive and statistically significant across all four models, as is the relationship between family income and interest in participating. The model estimates that, holding other variables in the model constant, those with a four-year college degree reported levels of interest that were between .20 and .28 units higher than those whose highest level of educational attainment is a high school diploma or equivalent on these 4-point scales.¹⁰ A two standard deviation increase in family income (7 unit increase on the 16-point scale) is associated with a similar .29 to .42 unit increase in interest across the four models.

Table 2. Predicting interest in participating in PB

Note. Cell entries are unstandardized OLS coefficients; robust standard errors in parentheses.

^∗ p < 0.05.

^∗∗ p < 0.01.

Turning to H _1B , the relationships tied to respondents’ ethnoracial identities offer some support for the notion that PB is particularly attractive to those from groups that are often underrepresented in the political process. After controlling for other variables in the models, respondents who identified as Latine or Black expressed greater interest in participating in each stage of the PB than respondents who identified as White (the reference category).¹¹ This said, respondents who identified as Asian Americans reported notably lower levels of interest – a pattern that is consistent with existing evidence that this group participates in local politics at lower rates (Hajnal and Trounstine 2005). We find suggestive evidence that those who identified with more than one ethnoracial group were more interested in PB than their White counterparts, but interpreting this pattern is difficult, given the heterogeneity of this group.¹² In short, with the exception of respondents who identified as Asian, our evidence supports H _1B .

The patterns tied to age also suggest that the PB process may be attractive to another group that is traditionally underrepresented in the political process: young people. The relationship between age and reported interest in participating is negative and statistically significant for each stage of the PB process except the final voting stage. A 30-year increase in age (a bit less than two standard deviations) is associated with approximately .3 units lower reported interest in participating in each of the first two stages of the process and .12 units lower interest in participation in the third stage (debating proposals at a town hall). Thus, broadly speaking, our findings support H _1C .

Finally, two demographic characteristics that we did not, a priori, expect to be related to interest in PB show statistically significant relationships in this regression model. First, respondents who identified as men reported higher levels of interest in participating in the agenda-setting stages of the process (submitting ideas and working on committees) than those who identified as women.¹³ One possibility is that this pattern is tied to gender differences – including inequalities in responsibilities for childcare and a tendency for women to underestimate their political capabilities – that other scholars have pointed to as factors in gender differences in political ambition (e.g. Lawless and Fox 2005; Fox and Lawless 2014).¹⁴

Second, respondents residing in Chicago expressed greater interest in PB. Here it may be that those living in smaller municipalities in Cook County feel better-positioned to influence local policies through traditional avenues than residents of Chicago, where city government may feel more complex and distant. This said, Chicago residents also reported higher rates of other forms of political participation than their suburban counterparts (see Supplementary Table S2). Another possibility is that Chicago residents are particularly cynical about local politics, given Chicago’s reputation as a hotbed of political corruption. Yet another possibility is that Chicago residents tend to be more drawn to PB because they are more likely to be familiar with the process – an explanation that finds some support in the fact that 37 percent of our survey respondents who reported living in a Chicago ZIP code indicated that they had heard of PB, compared with only 19 percent of suburban respondents.

The role of neighborhood context

Next we consider whether neighborhood context is related to respondents’ interest in participating in PB. We consider the possibility that self-interest rooted in neighborhood conditions affects eagerness to participate, as well as the possibility that interest in participating is contingent on how an individual is likely to feel about other participants. Because our four measures of interest in participating are highly correlated, for the sake of simplicity, we combine them into a mean index (inter-item correlations range from .58 to .73; Cronbach’s α = .89) for this analysis.

Column (1) of Table 3 replicates the models reported in Table 2 using this summary outcome measure.

Table 3. Predicting interest in participating in PB (Index)

Note. Cell entries are unstandardized OLS coefficients; robust standard errors in parentheses.

^∗ p < 0.05.

^∗∗ p < 0.01.

Unsurprisingly, the patterns that emerge are similar to those reported in Table 2. Higher levels of income and educational attainment are associated with significantly greater interest, Latine and Black respondents report greater – and Asian American respondents report lower – interest than White respondents, age is negatively related to interest, men are more interested than women, and Chicago residents report greater interest than Cook County residents living beyond the city limits. Before proceeding, we note that these patterns are broadly similar to those that emerge when we consider the correlates of other forms of participation in local politics – especially protest participation (see Supplementary Table S2 for details). The most notable exceptions are that the relationships between both education and reported participation and gender and participation do not emerge for all participatory outcomes. Additionally, as expected given previous research on the correlates of turnout, age is positively related to reported turnout in local elections.

Desire to remedy poor conditions

In column (2) of Table 3 we begin to assess whether a self-interested desire to remedy poor neighborhood conditions is associated with interest in participating in the PB process after controlling for respondents’ demographic characteristics. Specifically, we add a measure of respondents’ ratings of their neighborhood. Respondents were asked, “How would you rate the quality of each of the following parts of your neighborhood?” They rated the “streets, alleys, and sidewalks,” “green spaces (for example parks, trees, gardens, parkways),” and “overall cleanliness” of their neighborhood on a 5-point scale ranging from poor to excellent. We combined responses to these three highly correlated items into a mean index (correlations range from .69 to .77; Cronbach’s α = .89).

Including this measure in the model does not substantially affect the coefficients on the other variables in the model. Contrary to the expectation (articulated in H ₂) that improved subjective assessments of one’s neighborhood would be demobilizing – and, conversely, that those who view conditions in their neighborhood as poor would be more inclined to engage in PB – those who rated their neighborhood more favorably reported greater interest in participating. A two unit (approximately two standard deviation) increase in ratings of one’s neighborhood is associated with a .29 unit increase in interest in engaging in the PB process. We emphasize that this pattern emerges after controlling for respondents’ socioeconomic status (education and income).

In column (3) we consider another factor that may serve as a suitable proxy for neighborhood conditions. Specifically, we leverage the fact that respondents provided their ZIP code to consider a measure of the poverty rate in the respondents ZIP Code Tabulation Area (ZCTA) as measured by the US Census. This measure has the advantage of being objective, rather than rooted in respondents’ subjective assessments. The coefficient on this variable is negative (suggesting that, after controlling for respondent demographics, those living in areas with higher rates of poverty are less interested in participating), but not statistically significant (p = 0.328).

Shared identity and trust

In column (4) of Table 3 we consider the possibility that eagerness to engage in PB is shaped by whether others in one’s neighborhood share one’s ethnoracial identity (H _3A ). We again use respondents’ reported ZIP code to match them to Census ZCTAs, allowing us to construct a measure of the percentage of residents within their ZCTA who share each respondents’ reported ethnoracial identity. We restrict our analysis to respondents who identified as White, Black, Latine, or Asian. The findings run counter to the expectation that people would be more inclined to engage in PB if others in their area share their identities. Instead, the pattern that emerges indicates that respondents living in areas where more people share their ethnoracial identity report lower levels of interest in participating in PB.¹⁵

In column (5) we use the more direct, but subjective measure of how people feel about others in their neighborhood (H _3B ). We construct a measure of trust in one’s neighbors that is the average of responses to items asking people “How much do you trust [your neighbors]?” and “How much can you rely on [your neighbors] for support?” In each case, responses were recorded on a 4-point scale ranging from “not at all” to “a lot.” Here we find clear support for the expectation that people who expect co-participants to be constructive partners would be more drawn to PB. A 2-unit (approximately two standard deviation) increase in trust is associated with a .37 unit increase in reported interest in participating.

Finally, in column (6) we include all variables in a single model. This approach yields conclusions that are substantively similar to those discussed above. Thus, our findings do not appear to be a product of idiosyncrasies in our model specifications.

Do interested individuals have different priorities?

We conclude by considering our second research question regarding how variation in interest in participating in PB may shape the substantive outcomes the PB process yields. To what extent are those who are interested in participating in the PB process inclined to prioritize different types of projects than those who are less interested? We rely on open-ended responses to a question that asked respondents “If you were given $1 million to improve public spaces in your community (such as streets and sidewalks, street lighting, parks), what would you spend it on?” A team of three coders flagged each response for mentions of various types of projects. Each response was considered by two independent coders. A third coder then resolved any coding discrepancies. Note that, although many responses (48 percent) referred to only one category (e.g., “repair the streets”), responses ranged in length from 0 words (11 cases in our sample¹⁶) to 131 words, with a median of 6 words (average = 9.7 words). Thus, many were flagged for multiple codes. For example, “roads parks and schools” was coded for three categories: Streets and Traffic, Parks/Recreation, and “Education/Youth Programs.” Excluding unusable responses, the median number of codes responses were flagged for was 2, with an average of 1.78. We provide further details about the coding process in Supplementary material S1.

We summarize the final coding scheme, present example responses, and report intercoder agreement (prior to resolution by the third coder) and the percent of responses flagged with each code in Table 4. The most commonly cited projects pertained to streets and traffic (38 percent), parks and recreation (27 percent), lighting (20 percent), and sidewalk repairs (19 percent). Note that, because we anticipated that few respondents would be familiar with the types of projects the PB process can fund, we offered broad examples in the question we asked (“such as streets and sidewalks, street lighting, parks”). This may partially explain why these categories were mentioned so often. However, this is less of a concern when we assess how the probability of mentioning a particular type of project varied across respondents – analysis which we turn to now.

Table 4. Open-ended coding categories

Note. Intercoder agreement: % of responses that two independent coders coded identically. Responses flagged: % of all responses where each code was flagged.

In Table 5 we report the relationships between our index of interest in engaging in PB and whether each open-ended coding category was flagged. For presentation purposes we collapse the interest index into quartiles and report the percentage of participants in each quartile whose response was flagged for each code. We also report the bivariate correlation between the (uncollapsed) index and whether each code was flagged.

Table 5. Relationships between interest in participating in participatory budgeting and open-ended responses

Note. Cell entries indicate the percentage of responses offered by people in each “interest in participatory budgeting” quartile mentioning a given type of project. Correlation row indicates the point biserial correlation between the continuous measure of interest in participatory budgeting and whether a type of project was mentioned.

* p < 0.05.

The patterns that emerge in Table 5 fail to support H ₄. We find scant evidence that interest in PB is related to the types of projects respondents mentioned. The only relationship that reaches conventional thresholds of statistical significance is tied to responses that cited needs to improve streets, alleys, or traffic safety. In that case, as interest in PB increases, the probability of mentioning this type of project decreases.¹⁷

Discussion

The causes and consequences of inequalities in political participation in democracies are well-documented. Participatory budgeting advocates argue that the PB processes – implemented in a growing array of subnational political units in the United States and beyond – offer a path toward remedying these inequalities. They posit that by bringing new participants into the political process, PB can not only alter how public funds are allocated in urban areas but also serve as a forum for forging community ties and a healthier culture of civic engagement. However, in order for these benefits to be realized, eligible residents – especially those who decline to participate in other forms of participation – must opt to engage in this process. To date PB participation rates are low and questions remain regarding the extent to which PB is likely to attract a more representative pool of participants than other forms of political engagement.

In this article we leveraged a survey of Cook County residents to examine the correlates of interest in participating in the PB process to shed light on a basic, but crucial, question: who finds the PB process appealing? The relationship we find between respondents’ socioeconomic status and interest in participating cuts against advocates’ hopes that the PB process might, by its nature, overcome social inequalities by drawing traditionally marginalized groups into the political process. However, our evidence also suggests that the process may be particularly attractive to people of color and younger Americans – groups whose voices have also been underrepresented in politics.¹⁸

We find little evidence that respondents who we might expect to have a particularly pronounced material stake in PB outcomes – those who view their neighborhood as particularly desperate for the types of projects the PB process tends to fund – are more interested in participating. In fact, even after accounting for respondents’ demographic characteristics, we find the opposite: those who view conditions in their neighborhood as more favorable reported higher levels of interest in participating.

Turning to social features of the neighborhood, our analysis yields seemingly mixed findings. When we rely on administrative data to calculate the extent to which respondents’ neighbors share their ethnoracial identity, we find that greater rates of shared identity are associated with lower levels of interest in participating. In contrast, a more direct measure of trust in one’s neighbors is positively associated with interest in participating. This apparent mismatch in findings appears to stem from a mistaken assumption that shared identity would be a serviceable proxy for perceptions of shared interests and interpersonal trust. Upon closer examination, our measures of shared identity and trust are virtually unrelated (correlation = −0.044). Taken together, these findings suggest that fostering improved interpersonal trust – rather than emphasizing material needs – may be critical to increasing participation in PB. Moreover, they cast doubt on the notion that more racially or ethnically homogeneous neighborhoods are inherently fertile grounds for collective action. In short, and somewhat paradoxically, high levels of social capital and trust may be important preconditions for robust participation in a process advocates posit has the potential to build social capital.

We also considered respondents’ answers to an open-ended question that was designed to offer insight into the types of projects respondents might be expected to prioritize in the PB process. Although our survey data suggest that social inequalities influence who is inclined to participate in PB, responses to this question suggest that those who are most interested in participating have broadly similar priorities when it comes to spending public funds to those who report lower levels of interest.

As with all studies, our evidence has limitations. Our survey data measured interest in PB. Although there is reason to expect interest to predict actual participation, this connection is indirect. Expressed interest in participating may not translate into actual participation. Conversely, those who reported low levels of interest in our survey may be mobilized by additional information or recruitment efforts. In Table 6 we report the correlates of interest in participating in PB separately for the 30 percent of respondents who reported having heard of PB prior to the survey and those who said they had not. The patterns that emerge here should be viewed as particularly tentative. In addition to the possibility that social desirability considerations led some respondents to claim that they had heard of PB (and that susceptibility to those pressures is not randomly assigned), column (1) demonstrates that those who reported having heard of PB were demographically different from those who said they had not.

Table 6. Interest in participating in PB by whether respondent reported having heard of PB prior to survey

Note. Cell entries are unstandardized OLS coefficients; robust standard errors in parentheses. The outcome in column (1) is an indicator for whether the respondent reported having heard of participatory budgeting. Columns (2)–(3) report models analogous to the model reported in column (1) of Table 3 separately for those said they had not (2) and had (3) heard of participatory budgeting. One respondent did not answer the question about whether they had heard of participatory budgeting. We estimated a model interacting the “Heard of PB” indicator with each covariate. Italicized coefficients in columns (2) and (3) indicate that the interaction associated with that covariate is statistically significant (p < 0.05).

^∗ p < 0.05.

^∗∗ p < 0.01.

Encouragingly for advocates, the relationship between educational attainment and interest in participating in PB is significantly weaker among those who had heard of PB. Less encouragingly, the income relationship strengthens among this latter group. Among those who had not heard of PB, Black respondents reported more interest than their White counterparts but Latine respondents did not. This pattern is inverted among those who said they had heard of PB. The age relationship is attenuated among those who had heard of PB (p = 0.052 for coefficient on Heard of PB × Age interaction [model available upon request]). Finally, among those who had heard of PB, Chicago residents expressed more interest in participating than their suburban counterparts – a pattern that does not emerge among those who said they had not heard of PB. We emphasize that these patterns should not be viewed as conclusive. Rather, they offer suggestive support for the possibility that exposure to information about PB (or, perhaps, experiences with the process) can alter patterns regarding who is inclined to engage in the process.¹⁹

The open-ended responses we considered also have important limitations. We were only able to capture which broad areas respondents were inclined to prioritize. The fact that those who report high and low levels of interest in participating cite sidewalk repair as a priority at similar rates does not demonstrate that they, for example, agree on which sidewalk should be repaired. Beyond this, the PB process is meant to be deliberative. The priorities people cite in response to a survey may only loosely reflect the priorities they would land on after being asked to explain those priorities or being exposed to their neighbors’ priorities.

These limitations aside, this article reports novel evidence that sheds light on a basic question about PB: is this process, by its nature, likely to attract a more representative pool of participants than other forms of participation? Our evidence suggests that the answer to this question is no. Like other forms of participation, prospective PB participants must weigh the costs and benefits of participating and prevailing social conditions may foster or impede participation.

Our findings have practical implications that can help practitioners and policy-makers design better PB processes and suggest that substantial outreach efforts will be needed to achieve broader and more equitable participation. Identifying avenues for broadening participation is particularly relevant considering expansions in adoption of PB processes in the United States and around the world. The Participatory Budgeting World Atlas reports that the United States had 145 active PB processes in 2022, and notes continued growth in the number of PB processes in the United States, as well as in the size of the budgets these processes allocate (Dias, Enrquez, and Jlio 2022). The evidence we report here can assist practitioners as they refine existing programs and design new ones. In order for PB to achieve supporters’ goals, organizers will need to pursue targeted strategies for mobilizing underrepresented populations and overcoming persistent inequalities in the individual- and neighborhood-level resources that reinforce unequal patterns of participation.