Social Capital Project

About the Project

Today, Americans face a wide variety of challenges in our era of tumultuous transition. We are materially better off in many ways than in the past. But despite this real progress, there is a sense that our social fabric has seen better days. Leading thinkers have issued warnings that we are increasingly “bowling alone,” “coming apart,” and inhabiting a “fractured republic.” At the heart of those warnings is a view that what happens in the middle layers of our society is vital to sustaining a free, prosperous, democratic, and pluralistic country. That space is held together by extended networks of cooperation and social support, norms of reciprocity and mutual obligation, trust, and social cohesion. In short, it is sustained by what we do together.

The Social Capital Project is a multi-year research effort that will investigate the evolving nature, quality, and importance of our associational life. “Associational life” is our shorthand for the web of social relationships through which we pursue joint endeavors—namely, our families, our communities, our workplaces, and our religious congregations. These institutions are critical to forming our character and capacities, providing us with meaning and purpose, and for addressing the many challenges we face.

The goal of the project is to better understand why the health of our associational life feels so compromised, what consequences have followed from changes in the middle social layers of our society, why some communities have more robust civil society than others, and what can be done—or can stop being done—to improve the health of our social capital. Through a series of reports and hearings, it will study the state of the relationships that weave together the social fabric enabling our country—our laws, our institutions, our markets, and our democracy—to function so well in the first place.

Few countries are as generous as the United States when it comes to volunteering.1 One quarter of Americans donated time to an organization in 2015. One need look no further than the outpouring of assistance in response to recent natural disasters for powerful illustrations of American civic-mindedness. In 2006 and 2007, over a million volunteers joined the recovery effort in the wake of Hurricane Katrina. One year after Superstorm Sandy, the Federal Emergency Management Agency reported that, “some 173,544 volunteers had invested more than 1 million volunteer hours in the Sandy recovery effort.”2 And already, Americans have responded to the devastation of Hurricanes Harvey and Irma with characteristic generosity.

One of the findings in the Social Capital Project’s first report, “What We Do Together,” was that volunteering rates in the US are no lower than in the 1970s, in contrast to many indicators of social capital that have worsened over time. In fact, we found that the number of hours spent volunteering per person has increased.3

This post goes further, arguing that volunteering rates may have actually risen over the long run. The post also explores the demographics of volunteerism. Volunteerism is more common among women and among socioeconomically advantaged groups. The volunteering rates of Americans of different ages have converged over time. Over the long run, volunteerism may have declined among adults ages 25 to 44, but it increased among other age groups, in particular the elderly. Volunteerism may have declined since 2002, though different data sources come to different conclusions. Any recent decline has been concentrated among non-Hispanic whites, who continue to have relatively high volunteer rates. Volunteering is more common in the North than in the South, with Utah leading the way overall.

Trends in Volunteerism

The Current Population Survey (CPS) is the key source of volunteerism data from the federal government. Since 2002, the CPS has included a set of questions on volunteer work that are highly comparable over time. The lower part of Figure 1 displays the trend in volunteerism from these surveys and from two earlier ones conducted in 1974 and 1989.4 These earlier estimates may not be comparable to the ones since 2002 because their questions were worded differently. However, we can compare the CPS trend to the trend in volunteerism from a series of Gallup surveys (shown in the upper part of Figure 1).5

Unfortunately, the two data sources produce different results, though neither indicates that volunteerism has declined. First, the Gallup volunteering estimates are higher in every year than those from the CPS, possibly reflecting differences between the two data sources.6 According to the CPS, in 2015 only one in four Americans aged 16 and over volunteered at an organization, compared to the two in three adults Gallup finds volunteered time to a religious organization or other charitable cause. Second, Gallup shows a clear rise in volunteerism over time, while the trend is flat in the CPS data.

In particular, the estimates from the two data sources have diverged since 2004. According to the CPS, volunteering rates may have fallen slightly, while the Gallup data show a continued increase in volunteering. It is unclear what accounts for this difference.

Figure 1. Volunteer Rates, 1974-2015

Note: Clicking legend entries removes or adds individual series. Source: See notes 4 and 5.

In the CPS, volunteerism is down at nearly all organization types since 2002 (not shown), the exceptions being social and community service groups, environmental or animal care organizations, and “some other type of organization.” About one third of volunteers (34 percent in 2015, not shown) give their time primarily to religious organizations, according to the CPS. Child educational and recreational groups are second-most common (accounting for the most time among 19 percent of volunteers), followed by social and community service groups (15 percent). The remaining third of volunteers work primarily in health care, education, civic, or other organizations. These percentages have been fairly stable since 2002.

Who Volunteers?

A report summarizing the 1974 CPS data reads, “The most typical American volunteer in 1974 was a married, white woman between ages 25 and 44 who held a college degree and was in the upper income bracket.”7 This was still true in 2015. In the rest of this post, we focus on the CPS estimates.

Women and Men

Women have higher volunteer rates than men (26 percent versus 21 percent in 2015). The gap has been strikingly stable over time, as shown in Figure 2; in the CPS, volunteering rates are about the same today as in 1974 among both men and women. While one might speculate that women’s higher rates of volunteerism are due to their lower employment rate, women who were employed full-time, part-time, and who were not employed in 2015 all volunteered more than their male counterparts with the same employment status. In fact, women who work full-time have higher volunteerism rates than men who do no work, as shown in Figure 3. The fact that the gender gap in volunteerism is no lower today than in 1974, when fewer women worked (especially full-time), also suggests that the gap is unrelated to hours spent on the job.8

Figure 2. Volunteer Rates by Gender, 1974-2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 1974, 1989, and 2002-15 Current Population Survey microdata.

Figure 3. Volunteer Rates by Gender and Employment Status, 2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 2015 Current Population Survey microdata.

Marital Status

Volunteering is much more common for married individuals than others (Figure 4). In 2015, married Americans were fifty percent more likely to have volunteered than those who had never married (three in ten volunteering, versus two in ten). This may simply reflect that couples have more time than single adults, who cannot share other household responsibilities with a spouse (though they might be cohabiting with a partner). Alternatively, it could be that the kind of people who get and stay married would be relatively more likely to volunteer even if they were single.

As mentioned above, volunteering is frequently connected to educational and youth service organizations. Parents are naturally inclined to support their children’s activities. However, married but childless Americans between ages 25 to 54 volunteer more than never-married adults whether or not the latter have children, and married parents volunteer only a bit more than childless married adults (not shown).

Figure 4 shows that the gap in volunteerism between married and never-married adults widened between the mid-1970s and the late 1980s. According to the CPS, volunteerism increased among married adults and among adults who are separated, divorced, or widowed. Among never-married adults volunteerism was no lower in 2015 than in 1974. Taken together, these facts suggest that had marriage not become rarer over time, volunteerism would have risen, even in the CPS.

Figure 4. Volunteer Rates by Marital Status, 1974-2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 1974, 1989, and 2002-15 Current Population Survey microdata.


Age differences in volunteering have narrowed sharply over the past 45 years (Figure 5). The CPS indicates that volunteer rates rose among adults older than 44, especially among those older than 64 years old. Volunteering also rose among Americans under the age of 25, particularly after 1989. However, adults ages 25 to 44 volunteer less than they used to. They are a somewhat larger share of the population than in 1974 or 1989, which also puts downward pressure on the aggregate rate of volunteerism. Volunteering among those 65 and older has been relatively stable since 2005, but in the CPS it has declined among other groups.

Figure 5. Volunteer Rates by Age, 1974-2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 1974, 1989, and 2002-15 Current Population Survey microdata.


Figure 6 shows that volunteerism increases with educational attainment.9 In 2015, those with a college degree were nearly three times as likely to have volunteered in the past year as those with less than a high school degree (37 percent rate versus 13 percent). Volunteering fell between the 1974 and 2015 CPS surveys among those with at least a high school diploma, but it was stable among high school dropouts. Since 2005, however, the CPS data indicate that volunteering fell among adults in all three educational categories. Educational attainment rose between 2002 and 2015.10 If getting more education has a causal influence on volunteering—a questionable conclusion, to be sure—the recent decline in volunteerism in the CPS would have been larger absent rising levels of educational attainment, and the stable long-term trend would have pointed downward.

Figure 6. Volunteer Rates by Education, 1974-2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 1974, 1989, and 2002-15 Current Population Survey microdata.


In 2015, the volunteer rate for individuals from households making $100,000 or more (35 percent, not shown) was 21 points higher than the volunteer rate for individuals from households making less than $20,000 (14 percent).11 In general, as income falls, so does volunteering. Below around $10,000, this relationship breaks down, primarily due to income mismeasurement. Figure 7 plots income (in 2015 dollars) against volunteerism rates, showing lines for 1974, 1989, 2002, and 2015. Except toward the bottom of the income distribution (where measurement issues intrude), volunteer rates were higher in 1974 than in the other four years. Over the long run, it is possible that rising incomes kept the volunteerism rate in the CPS from falling.

Figure 7. Volunteer Rates by Family Income, 1974-2015, Selected Years

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 1974, 1989, 2002, and 2015 Current Population Survey microdata.


As indicated in Figure 8, volunteering is most common among non-Hispanic whites, with 27 percent volunteering in 2015. At the other end, 14 percent of Hispanics volunteered.12 In between, African Americans (18 percent) and others (18 percent) have similar rates of volunteering. The gap between the volunteer rates of whites and nonwhites has narrowed because as the volunteer rate among whites has fallen since 2003, while rates for blacks and Hispanics have remained relatively stable. In fact, Figure 8 suggests that essentially the entire decline in volunteerism since 2003 occurred among non-Hispanic white Americans (and adults in the much smaller “other” category, including Asian Americans and multiracial adults, among others). In addition, though, the rising share of the population that is nonwhite has also put downward pressure on volunteering, since nonwhites tend to volunteer less than whites.

Figure 8. Volunteer Rates by Race, 1974-2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 1974, 1989, and 2002-2015 Current Population Survey microdata.

Volunteering by State

In 2015, the highest state volunteer rate was in Utah with an estimated 38 percent of the population having volunteered in the past year; Mississippi’s volunteer rate was the lowest at 16 percent. In general, volunteering is more common in northern states than in southern ones (see Figure 9).

Figure 9. Volunteer Rate by State, 2015

Note: Clicking legend entries removes or adds individual series. Source: Social Capital Project calculations using 2015 Current Population Survey microdata.


Volunteerism is the rare indicator of social capital that has not worsened (or even rarer, that has improved) over the last forty years. However, there is clear divergence across states in rates of volunteering as well as divisions along demographic and socioeconomic lines. In future work, we hope to dive deeper into these differences.


1. Charities Aid Foundation, “World Giving Index 2016”, accessed June 5, 2017. The U.S. ranked fifth among 140 countries in the percentage who had volunteered with an organization, falling behind four quite poor nations (Turkmenistan, Myanmar, Indonesia, and Sri Lanka).

2. Federal Emergency Management Agency, “In Disaster Recovery, Volunteer Efforts are Priceless”, press release, release date, November 7, 2013,

3. U.S. Congress, Joint Economic Committee, Social Capital Project, What We Do Together: The State of Associational Life in America, report prepared by the Vice Chairman’s staff, 115th Cong., 1st Sess., May 2017,

4. All of the results in this paper are from our own analyses of CPS microdata, downloaded from the National Bureau of Economic Research website ( for 1974 and 1989 and from the Census Bureau's Data Ferrett utility for 2002-2015. We include all persons at least 16 years old in all analyses. For comparability with the earlier estimates, we ignore the follow-up question in the 2002-2015 surveys that double-checks with respondents about whether they volunteered, following Robert Grimm, Jr. et al., Volunteer Growth in America: A Review of Trends Since 1974, (Washington: Corporation for National and Community Service, 2006),

The CPS is administered by the Census Bureau and sponsored by the Bureau of Labor Statistics. The federal agency ACTION co-sponsored a supplement to the April 1974 CPS, and another supplement was fielded in May 1989 (apparently without a co-sponsoring agency). Volunteer data since 2002 come from the September CPS, co-sponsored by USA Freedom Corps (2002-2007) and the Corporation for National and Community Service (2008-2015), both of them federal agencies involved in promoting volunteerism.

There was another CPS supplement on voluntarism in November 1965, but the way that the questions were asked make it especially incomparable to the subsequent surveys. See Howard V. Hayge, “Volunteers in the U.S.: who donates the time?,” Monthly Labor Review 114:2 (February 1991), 22.

5. The Gallup estimates come from three different survey sets. Gallup has conducted surveys on volunteerism for the Independent Sector for a number of years. We took 1981 and 1985 estimates (for respondents at least 14 years old) from U.S. Bureau of the Census, Statistical Abstract of the United States: 1989 (109th Edition), Washington, DC, 1989, Table 614. Estimates for 1988-1999 are from our analyses using the Survey Documentation and Analysis (SDA) online utility at They represent respondents at least 18 years old.

Gallup also partnered with Princeton Survey Research Associates on surveys about volunteerism over the years. The estimates cited in Figure 1 are taken from Everett Carll Ladd, The Ladd Report (New York: Free Press, 1999), 64. Estimates for 1981 and 1987 are two-month averages. It is unclear what ages are represented in these estimates.

The 2002-2013 Gallup series is from the Gallup editors, “Most Americans Practice Charitable Giving, Volunteerism,” December 13, 2013, The estimates are for adults age 18 or older.

6. The Gallup surveys collect no information for household members other than the respondent, in contrast to the CPS, though this appears to be a relatively small factor, based on our analyses of CPS data. Volunteering for any “charitable cause” is counted in recent Gallup surveys, but only volunteering for organizations is counted in the CPS. Finally, the Gallup surveys are primarily about civic engagement and voluntarism, while the CPS is primarily focused on labor force questions. That may prime Gallup respondents to think about their volunteering activities.

7. The Agency for Volunteer Service (ACTION), Americans Volunteer - 1974, (Washington: ACTION, 1975), 3.

8. The 1965 CPS data also indicate a gender gap just as wide as today.

9. In the 1974 and 1989 data, “college degree” means completed 16 or more years of schooling, “high school diploma” means completed 12-15 years, and “less than high school” means fewer than 12 years completed. In the 2002-2015 data, “college degree” means obtaining a bachelor’s, master’s, professional, or doctoral degree; “high school diploma” means graduating from high school and excludes getting a Graduate Equivalency Degree (GED); “less than high school” means not getting a high school diploma (including getting a GED). Note that teenagers and young adults are included in these analyses, so some future high school and college graduates are categorized as “less than high school” or “high school diploma” despite the fact they will eventually have higher educational attainment.

10. National Center for Education Statistics, Digest of Education Statistics, Table 104.10,

11. The income variable is the family income of the householder’s family, and it is derived from a survey question that asks respondents to place themselves in one of several categories. The number of categories (and the income ranges used to define them) is not the same from year to year. We coded everyone at the midpoint of their nominal income category (at 1.5 times the lower bound of the highest category in each year), then we adjusted these midpoints for inflation using the Bureau of Economic Analysis Personal Consumption Expenditures (PCE) deflator, putting them in constant 2015 dollars. Note that members of a household that are not in the householder’s family—such as roommates, cohabiting romantic partners, and others not linked to the householder by blood or marriage—are assigned the householder’s family income.

12. The 1974 Spanish origin variable appears to be problematically coded, so the categories for that year are “white,” “black,” and “other.” Beginning in 2003, respondents could indicate multiple racial categories. From this year forward, the categories are “non-Hispanic white alone” (no other racial categories indicated), “non-Hispanic black alone” (no other racial categories indicated), “Hispanic,” and “other” (including multiracial respondents).