THE ALTRUISM BUDGET
Measuring and Encouraging Charitable Giving
Laura K. Gee and Jonathan Meer
IN RECENT YEARS, SOME CHARITIES HAVE seen huge increases in their receipts—for example, the viral Ice Bucket Challenge in 2014 coincided with an increase of over 200 percent in donations to the ALS Association (Steel 2014). Yet between 2012 and 2017, overall monetary donations to U.S. nonprofits held steady at around 2 percent of GDP (Giving USA 2018). Patterns like this inspire questions about the limits of the overall budget for the nonprofit field in general—what we will refer to as the altruism budget. When donations increase to one recipient, do they decrease for others? When they increase now, will they decrease later? And if people give more of one kind of donation—for example, money—do they give less of another—for example, time? In short, do donors have a fixed budget of altruistic acts? Or, conversely, is the overall altruism budget expandable? These questions are of fundamental importance to broader issues about philanthropy. The implications of fund-raising activities are very different for society if fund-raising is a zero-sum game, in which donations given to one cause fully offset donations to another. If resources spent on fund-raising merely cannibalize donations from elsewhere, then many of these efforts may be wasteful.
This chapter reviews the state of the research on the nature of the altruism budget. After a discussion of the definition of an altruistic act, we describe the different kinds of altruistic acts that we include in our concept of the altruism budget. The bulk of the paper reviews the state of the research on how changes in one type of giving affect others, both over time and contemporaneously. The findings of shifts over time are consistent, if somewhat counterintuitive: future giving does not seem to be reduced by current giving, suggesting that the altruism budget is expandable across time. In other respects, the nascent research in this challenging field of inquiry does not yet offer much beyond suggestive answers. We detail the difficulties with estimating causal effects of competition between recipients of altruism later in the chapter. In order to fully examine the nature of the altruism budget, researchers would need to observe all of the altruistic acts an individual engages in over their lifetime. This is infeasible. Precise answers can be found for more narrowly defined questions, like how increases in giving to one cause on a specific online platform affect giving to other causes on that platform at a certain time. But this does not truly answer the question of how the altruism budget shifts more broadly. We offer directions for further research into how increased giving to one recipient affects giving to others.
What Is Altruism?
Social scientists have yet to agree on a single definition of altruism. Indeed, it is quite difficult to establish a firm definition of the concept. The dictionary definition of altruism is “unselfish regard for or devotion to the welfare of others,” and the attendant example of the word used in context is “charitable acts motivated purely by altruism” (Merriam-Webster n.d.).
This example highlights the difficulty of offering a scientific definition of altruism: charitable acts are generally observable, but their motivations are less knowable. Donating money to a charity is generally accepted as an altruistic act, but is it always purely “unselfish”? Indeed, Rene Bekkers and Pamala Wiepking (2011) list altruism as only one of the eight mechanisms for making a charitable donation.
Often there are explicit benefits to a donor from making a donation. In the United States, donors can deduct monetary and in-kind gifts from their income taxes, thus lowering their tax bill.1 Even without this preferential tax treatment, acts that are generally considered altruistic offer many private—or one might go as far as to say selfish—benefits.
Examples abound: when parents volunteer at a child’s school, they believe they have more control over their child’s class placement (Gee 2011). People donate more to their alma maters as their children approach college age, and they expect those donations to buy their children favorable treatment in the admissions process (Meer and Rosen 2009). Someone may make a person-to-person gift of time, such as babysitting for a neighbor in need, with the expectation that they will be owed a favor. The frequent use of “donor premiums,” or gifts from the charity to donors (either before making a donation, as an inducement to give, or afterward as a token of appreciation) suggests that a more explicit exchange of consumption goods for donations plays a role in giving (Alpizar, Carlsson, and Johansson-Stenman 2008; Eckel, Herberich, and Meer 2018; Falk 2007; Meer and Rosen 2012). The Internal Revenue Service (IRS) explicitly recognizes that donations are not fully altruistic: it offers charities guidance on how to make “good faith estimates of the fair market value” of the goods and services (e.g., an invitation to a gala) they provide to donors and uses these estimates to reduce the amount that itemizing taxpayers can deduct on their income tax returns.2
In addition to explicit benefits that result from donations, how someone appears when they donate is of utmost importance. The increase in donations when someone is able to appear more generous from donating is one of the most robust findings in the literature (Andreoni and Bernheim 2009). A related desire to acquire “status” can also motivate giving (Blumkin and Sadka 2007; Glazer and Konrad 1996; Harbaugh 1998a, 1998b; Kumru and Vesterlund 2010). The benefits to charities and donors of observable donations can multiply when the donation sets an example for others or provides information about the most effective or worthy charities, particularly if the individual making the donation is well known (Andreoni 2006a; Karlan and List 2012; List and Rondeau 2003; Potters, Sefton, and Vesterlund 2005, 2007; Vesterlund 2003; Kessler 2017). Evidence suggests that when gifts are made publicly, prospective donors are influenced not only by others’ giving patterns but also by knowing that they are being observed themselves (Castillo, Petrie, and Wardell 2014; Gee and Schreck 2018; Shang and Croson 2009; Smith, Windmeijer, and Wright 2015). A complementary mechanism is that people may donate because they want to avoid shame or explicit punishments for not contributing.3
Making donation decisions more observable is not always helpful, however. For instance, if donors are promised some explicit benefit when they donate—such as some highly valued gift—they may be concerned that donating will make them appear selfish instead of generous (Ariely, Bracha, and Meier 2009). Such a backlash to explicit benefits because of image concerns may be particularly likely among individuals with less established or known reputations (Exley 2017). Other, related image concerns may also arise, such as when donations signal not only one’s generosity but also one’s income (Bracha and Vesterlund 2017).
Most of the private benefits described here are available only if the donor’s identity is known. It may seem, then, that anonymous donations are purely altruistic. But even anonymous donors may experience giving’s implicit benefits. For example, donating can be an act of self-signaling that allows donors to think of themselves as good and generous people (Bénabou and Tirole 2006). Donors may also receive a “warm glow”—a personal feeling of well-being—from saying yes to requests for donations. And the amount of warm glow may vary by how the charity plans to use the donation.4
There is a well-established correlation between being asked for a donation and actually making a gift of time or money. But it can be difficult to ascribe causality to correlations between being asked and being a donor, since the former may also influence the latter. Using a variety of approaches, including both field and natural experiments, researchers have found substantial evidence in favor of “the power of the ask” (Andreoni and Rao 2011; Meer and Rosen 2011).5 Nonetheless, there are several ways to make asks more effective, and these strategies may work because they make the donor enjoy the act of giving more, or the act of declining to give less. First, asks should be in person and trigger empathy (Jason et al. 1984; Small and Loewenstein 2003). Second, asks are more effective when the solicitors are more physically attractive or have a personal connection with the prospective donor (Meer 2011; Price 2008; Raihani and Smith 2015).6 Third, asks are more effective if there will be future interactions between a donor and the recipient (Rand and Nowak 2013). Fourth, asks are more effective when they cannot be avoided (Andreoni, Rao, and Trachtman 2017; DellaVigna, List, and Malmendier 2012). Fifth, asks are more effective when they come as a surprise (Exley and Petrie 2018).
As the use of premiums makes clear, nonprofits themselves don’t care as much as Merriam-Webster does about whether donations are purely selfless. For the nonprofit, understanding the motivations behind donating is helpful only to the extent that it helps increase donations. Broader society too may care less about the psychology of altruism than about using resources for positive actions. We adopt this more use-based definition: the altruism budget, in our view, includes but is not limited to all donations, no matter what their motivation. This use-based definition is the most expansive view—an upper bound—on the items in the altruism budget. This approach allows us to remain agnostic on donors’ motivations and avoid making often arbitrary decisions about the underlying motivation for various actions.
What Is in the Altruism Budget?
The most common measure of total charitable giving is simply the total amount of money given to nonprofit organizations—although from 1929 to 1959, the National Bureau of Economic Research (NBER) included in its measures of total charitable giving both government spending on social welfare programs and person-to-person gifts, such as remittances sent overseas (Soskis 2017).7 These kinds of monetary gifts are relatively easy to measure, but focusing solely on them may lead us to miss the bigger picture.8
Recent years have seen a burgeoning interest in measuring other forms of charitable giving. The most frequent focus is on gifts of time—that is, volunteering.9 Hybrid gifts, such as donating blood or an organ, which take up time but also involve literally giving up part of one’s self, could form yet another category. Or consider a highly educated lawyer who turns down a well-paid law firm job to work at the nonprofit Innocence Project, or even at a public defender’s office, for a much lower salary. Is this lawyer making a charitable donation of some kind? Are their actions altruistic? To what extent are they personally beneficial? It can be difficult to categorize many of these actions.10
Beyond time and money, many other gifts could be included in the overall altruism budget, but the value of these gifts is often difficult to quantify. A donation of $500 to the local YMCA is most likely valued at $500, but what about a donation of a couch worth $500 to the donor but only $300 to the charity?11 In a similar vein, when actor Ethan Hawke appears in a video campaign for the YMCA, he gives not only his time but also his reputation (Look to the Stars 2018). How can the latter be measured? How does the value of Ethan Hawke’s appearance compare with celebrity chef Marcus Samuelsson’s in the same video?12
Paying a premium for charity-linked goods—for example, outdoor gear with a charity’s logo, from which the charity receives a portion of the profits—may be another form of charitable donation (Elfenbein and McManus 2010). One could also add donations to politicians whose views one believes would benefit society. Intrafamily altruism can play a role as well, though an act may be thought of as less selfless if the beneficiary is one’s own kin. From a purely evolutionary perspective, a parent sacrificing his life for his child still gains the private benefit of passing along his genes (Samuelson 1993). None of these actions falls under the traditional notion of “charitable giving,” yet they may be motivated by altruism nonetheless.
Arguably, the altruism budget should include but not be limited to gifts of money, time, material goods, and reputation. And since the altruism budget includes so many different kinds of gifts, it is impossible to measure with any accuracy for any individual or group of people. It may, then, not truly be possible to answer questions about the altruism budget’s flexibility with complete accuracy. What we can do, though, is look at the relationships between some of its components for hints about any trade-offs between them.
In the next section, we discuss selected findings on whether changes in one type of charitable giving lead to changes in others. We will not summarize the rather large literature on whether donations from one party crowd out donations from another party, as in the case of government spending crowding out individual giving (see Vesterlund 2006; Andreoni 2006b; Ottoni-Wilhelm, Vesterlund, and Xie 2017).
To grasp the difficulty of measuring the altruism budget’s flexibility, it’s helpful to imagine the ideal experiment that would fully measure it. The ideal experiment would allow for the collection of granular data on every aspect of the altruism budget, including formal giving to a charity, informal giving (say, to panhandlers), volunteering, purchases of charity-linked goods, and intrafamily altruism. But even a snapshot of every type of altruistic act would be insufficient. Donations may crowd out others intertemporally: an inducement to make a donation today may reduce giving next year, or even in one’s bequest.13 If all of this data were available, a researcher could randomly assign perturbations to one type of giving or another (through, say, providing incentives to give to a specific charity or in a specific manner). By observing these incentives’ ripple effects on donations both within and across time periods and comparing the behavior of incentivized donors with those who did not receive those incentives, we could definitively answer the question of whether gifts at different times, to different sources, and of different forms are substitutes or complements. For obvious reasons, this kind of experiment is impossible. This level of detail and intrusiveness in data collection over a span of decades is simply unachievable. But by examining subsets of this question, researchers can begin to sketch the outlines of these relationships.
When Donors Increase Their Gifts to One Recipient, Do They Decrease Their Gifts to Others?
Perhaps the most natural place to begin is with different recipients. Does giving to one recipient result in fewer gifts to another recipient? Do efforts to raise money for one recipient cannibalize donations to others, or do they increase total giving?
Laboratory experiments offer the advantage of a controlled environment and examine how much substitution results from increasing incentives to donate to one recipient in a set of choices and from expanding the set of choices itself. Some laboratory studies find that increased giving to one recipient decreases giving to others (Corazzini, Cotton, and Valbonesi 2015; Harwell et al. 2015; Deck and Murphy 2018). Other laboratory studies, however, find that increased giving to one recipient increases total giving (Krieg and Samek 2017; Filiz-Ozbay and Uler 2018). Although concerns about the validity of applying findings from laboratory experiments to real-life situations are not unfounded, the unparalleled control they offer allows researchers to examine these questions with minimal interference from other, unobserved influences. Therefore, the mixed evidence in the lab settings suggests that this is a particularly difficult and perhaps context-specific question to answer. As an example of how this answer can vary depending on the type of competition, an increase in the number of charities available increases aggregate giving (though giving to each individual charity falls somewhat), while the availability of matches for some charities shifts donations toward them while leaving overall individual giving unchanged (Schmitz 2018).
Outside the laboratory, natural disasters are commonly used as an unexpected shock to giving to one type of charity. Somewhat surprisingly, most of these studies find that increased giving in response to a natural disaster is positively associated with other giving both at the time of disaster and following it (Brown, Harris, and Taylor 2012; Deryugina and Marx 2015; Scharf, Smith, and Wilhelm 2017). Similarly, when a matching grant induces giving to one project on the Donorschoose.org platform, giving does not decline to similar projects (Meer 2017). However, David Reinstein (2011) finds that when controlling for an individual’s time-invariant attributes (like innate altruism), giving to one type of charity is correlated with less giving to other types of charities.
Research on directed giving, in which donors can target their donations toward specific functions within a charity, also sheds some light on this question. Judd Kessler, Katherine Milkman, and C. Yiwei Zhang (2017) and Catherine Eckel, David Herberich, and Jonathan Meer (2017) find that allowing donors to express their preferences results in higher donations among those who give, but Mackenzie Alston and her colleagues (2018) find no strong effects. Altogether, the evidence on this particular question is mixed, and additional research would be fruitful.
When Donors Give More of One Kind of Gift, Do They Give Less of Another Kind?
Most of the previous work looking solely at gifts of time and gifts of money finds that an increase in gifts of time is correlated with an increase in gifts of money (Andreoni, Gale, and Scholz 1996), so at first glance one might take this as evidence that giving money causes an increase in gifts of time, and vice versa.14 However, generous people are likely to give in multiple ways, so finding a correlation does not imply such a causal relationship. Merely observing that someone who makes significant monetary donations also volunteers a relatively large amount of time does not mean that making the monetary donations caused the volunteerism (or vice versa). Understanding the causal, rather than correlational, relationship is necessary to answer the question of whether the altruism budget is fixed across gifts in different forms.
Early studies using survey data found evidence in favor of the idea that increasing one form of giving would increase other forms (Brown and Lankford 1992), but later studies find that the relationship may be more complicated (Andreoni et al. 1996; Feldman 2010; Yeomans and Al-Ubaydli 2018). For example, a person who is induced to give more money because of a tax break may learn about worthy causes when making those monetary donations and then start to volunteer more. Here one would observe a positive correlation between gifts of money and time, but the gift of money did not directly cause the increase in gifts of time. This distinction matters because the early evidence implied that inducing donors (or volunteers) to give (or volunteer) more would lead them to engage in the other activity more, whereas the later evidence implies that the altruism budget may be more fixed.
Beyond studies of observational data, a number of laboratory experiments have investigated the relationship between gifts of time and gifts of money. The results from these studies imply that gifts of time and money tend to offset each other, suggesting a relatively fixed altruism budget (Andreoni et al. 1996; Lilley and Slonim 2014; Brown, Meer, and Williams 2018).
As discussed earlier, the altruism budget includes many kinds of donations beyond time and money. Unfortunately, there is little research on choices involving two or more of these other types of gifts, most likely because relatively little data has been collected about these other forms of giving. Daniel W. Elfenbein, Raymond Fisman, and Brian McManus (2012) examined how charity-linked goods affect purchasing behavior but did not gather data about the interplay between the purchase of these goods and other donative behavior. More broadly, however, some research indicates that people may substitute one moral act for another; for example, people are more likely to donate money after they have lied (Blanken, van de Ven, and Zeelenberg 2015; Gneezy, Imas, and Madarász 2014). This finding suggests that the altruistic act of giving money is sometimes a substitute for other “moral” behavior.
Taken together, these studies indicate that although donations of time and money tend to increase together, that doesn’t necessarily mean the altruism budget is expandable, and the few controlled laboratory studies imply that the altruism budget is fixed. Yet researchers have at most focused on two forms (usually time and money), though the altruism budget is far more diversified. A promising approach for future research would be to expand the number of forms across which prospective donors can substitute.
When Donors Give More Now, Do They Give Less Later?
Finding ways to increase giving today is less meaningful if it simply reduces giving in the future. Generous people are likely to give throughout their lives, so merely observing that certain people give generously at multiple points in time doesn’t tell us whether current giving affects later giving in general. By using random assignment in field experiments or natural experiments that create shocks to giving, researchers can better ascertain whether giving today crowds out giving tomorrow.
A few articles imply that gifts today may lead to fewer gifts tomorrow (Meier 2007; Van Diepen, Donkers, and Franses 2009). Donors may become “fatigued” by multiple or frequent solicitations and simply tune out or even actively avoid these requests (Damgaard and Gravert 2018). However, the preponderance of the evidence finds that gifts today do not cannibalize gifts tomorrow. Jen Shang and Rachel Croson (2009), Maja Adena and Steffen Huck (2019), and Marco Castillo, Ragan Petrie, and Anya Samek (2017) find that donors induced to give more by a certain message do not give less in subsequent donation drives. Jason Cairns and Robert Slonim (2011) find that total donations at Catholic masses increased with an additional collection, with only about a fifth of the amount of the second round coming from reductions in the first round. Meer (2017) finds that matching-driven increases in giving to certain types of charities do not reduce future giving to related charities. Craig E. Landry and his colleagues (2010) similarly find that donors induced to give by a lottery give more in future solicitations that don’t include the lottery incentive. Bekkers (2015) finds that those offered a match do not give less in response to a natural disaster months later. Kimberley A. Scharf and her colleagues (2017) use high-frequency data on donations following natural disasters and similarly find that giving is not offset by lower donations later.
All of these articles focus on donations of money, but Nicola Lacetera and his colleagues (2012) find similar patterns in blood donations: providing material incentives at a blood drive increases the number of people donating blood and has no negative effect on future blood drives at the same location without incentives. Taken together, these results dovetail with the evidence on habit formation in charitable giving, which suggests that creating a habit of giving when young can lead to greater generosity later in life (Meer 2013; Rosen and Sims 2011). Well-specified research on short-term substitution (over, say, a matter of months) can add to this body of knowledge. But investigating substitution over the longer run and even into bequest giving would provide a fuller picture.
Conclusion
Increasing donations to a single recipient may be a worthy endeavor. But society at large is more concerned about overall charitable giving. Therefore, we must go beyond a focus on increasing a particular type of altruistic act, at a specific time, to a particular recipient. Instead, we must strive to understand the potential for increases in one kind of donation to one recipient at a particular time to be offset by decreases in other kinds of donations to other recipients later on. Obstacles to this understanding include the difficulty of defining altruistic acts and measuring the many kinds of altruistic acts.
Overall, the evidence is decidedly mixed on whether the altruism budget is fixed or flexible. Perhaps surprisingly, gifts at one point in time do not seem to be neutralized by lower giving later. But the impact on contemporaneous gifts to other recipients or through other forms of giving is more difficult to summarize. This is still a fairly new area of research that suffers from a relative lack of data. A truly comprehensive data set would provide information about the total amount of all possible types of gifts a person gives to all possible recipients at all the possible times.
A complementary question to this discussion considers what might lead individuals to make a conscious decision not to give. For instance, preventing the ask from being avoided or making surprise asks may be effective because individuals have difficulty finding excuses not to give.15 Indeed, one provocative finding from the literature is that it can be profit maximizing for a charity to allow prospective donors to opt out of being solicited (Kamdar et al. 2015). Even when the ask is not directly avoided, individuals often search for excuses—such as some chance that their donation will not have an impact or charity performance metrics—as a reason not to give (Exley 2015, 2018). But understanding why people don’t give is even more difficult than understanding why people do give. Lack of response to a solicitation may arise from inattention, lack of interest, procrastination, or active distaste. As a counterpart to questions about what underlies altruistic behavior, understanding the complexity of motives for not giving is an important avenue for future work.