Category reporting in charitable giving: an experimental analysis.
To raise donations from potential donors, charities employ a variety of solicitation tools, including mail-campaign drives, matchingdonation schemes, road shows, charity raffles (lotteries), and donation reporting plans. One donation reporting plan often employed is category reporting. Under this plan, charities publicly announce individual donations according to prespecified categories ranked by the size of donations. Those who donate a sum lower than the lowest threshold amount specified for the lowest category would not have their donations publicly reported. In contrast, those who donate an amount that falls within one of the prespecified categories would have their donations publicly reported and placed in a donor honor roll.
There are many examples of the practical use of category reporting. For instance, Builders Without Borders give donors public recognition as Contributor, Donor, Sponsor, Patron, or Event Supporter if they donate an amount of, respectively, up to US$99, US$100-499, US$500-999, or over US$1,000. (1) Another example is the Rotary Foundation, a not-for-profit corporation running international humanitarian and educational programs which solely depends on voluntary contributions. Donors who give US$100 or more per year are recognized as Rotary Foundation Sustaining Members. (2) Other organizations that often employ category reporting include city libraries, art institutions, science centers, and schools.
The prevalence of category reporting implies that charities perceive this approach as an effective way to raise public funds. Harbaugh's (1998a) explanation of the effectiveness of category reporting rests on the premise that donors are prestige seekers. Category reporting allows donors to be publicly recognized, and being publicly recognized gives donors benefits of prestige. Indeed, using data on donations from a group of lawyers to their alma mater law school, Harbaugh (1998b) estimated a utility function that has as its arguments prestige and intrinsic benefits of giving. Using these parameter estimates, Harbaugh calculated the predicted amount of solicited donations under category reporting and compared them with those under both no reporting and exact reporting, that is, reporting exact amount. He showed that category reporting generated the highest donations, and attributed this result to donors' desire to pursue prestige benefits when making charitable contributions.
While this structural estimation (Harbaugh 1998b) is persuasive in showing the effectiveness of category reporting, it is, however, not immediately apparent whether the resulting incremental donations can actually be solely attributed to donors' prestige motive for giving. This is because, under category reporting, two elements are essentially in operation. The first element is the public reporting, which gives rise to the publicity effect. The prestige motive for giving exists only when the publicity channel is available. The second element is the category setting, which gives rise to the category effect. The category setting could affect a donor's decision in two possible ways. First, the category threshold might act as a reference donation, and second, the category itself could bring self-image benefits to donors even when there is no public acknowledgment. The dual presence of the public reporting and the category setting and their interactions are not captured in the existing studies on category reporting (Barbieri and Malueg 2013; Cartwright and Patel 2013; Harbaugh 1998a, 1998b). Without disentangling these two potentially confounding factors, the prestige motive in charitable giving, especially that in the public domain, could be overestimated. The separation of these two effects would not only shed light on the underlying mechanism behind the superiority of category reporting to other reporting plans, but would also enable us to gauge the extent to which each of these effects influences donors' charitable-giving behaviors.
For the purpose of untangling these two confounding factors, in this study we conducted a charitable donation experiment. Subjects in our experiment were given endowments to make a real donation to Doctors Without Borders (Medecins Sans Frontieres), a well-known international humanitarian charitable organization. Two treatment effects, the publicity effect and the category effect, were induced in three different reporting plans: category reporting (CR), category no reporting (CNR), and no reporting (NR). Under CR, a category scheme was utilized, and qualifying donors were publicly acknowledged. Under CNR, the same category scheme was utilized, but donors who qualified for the category were not publicly reported. For simplicity, we set only one category bracket for each CR and CNR session, but we varied the threshold at four different levels between sessions. Under NR. neither the category setting nor the public reporting were utilized. It was essentially an anonymous donation plan.
This experimental design enabled us to disentangle the category effect and the publicity effect. We evaluated the category effect on donors' charitable behavior by comparing the donations solicited by NR with those solicited by CNR. Having no public reporting, these two treatments differed only by the presence of category bracket. Furthermore, we evaluated the publicity effect on donors' charitable behaviors by comparing the donations solicited under CNR with those solicited under CR. The differentiating factor between these two plans was the presence of public reporting.
We found that category reporting significantly increased average donations relative to the anonymous donation plan, and that this superiority was due to the dual presence of the category setting and the public reporting. The category setting shifted the donation distribution toward the category threshold and it increased average donations when the threshold was modestly set. The category setting did not affect donors' behavior at the highest category level. The public reporting, on the other hand, took effect only at the highest category level. It further increased the donations beyond the amount induced by the category effect by providing extra prestige benefits to generous donors. In short, the category effect dominated the publicity effect at lower category levels, and the publicity effect dominated the category effect at the highest category level. These two effects complemented each other as the category threshold level was being varied from low to high.
The paper is organized as follows. Section II reviews the related literature. Section III describes the experimental design. Section IV discusses the main results and Section V concludes.
II. RELATED LITERATURE AND BEHAVIORAL HYPOTHESIS
A. The Publicity Effect
The publicity effect stems from donors' taste for prestige, or social image (Harbaugh 1998a). In behavioral economics, social image has been identified as one of the major motivations for pro-social behaviors (Andreoni and Bernheim 2009; Benabou and Tirole 2006; Harbaugh 1998a, 1998b). The main idea is that individuals prefer to be known to others as kind, generous, or responsible. As a result, they would act in a more generous manner when their behaviors could be observed by others, that is, when the publicity channel is available. Karlan and McConnell (2014) verified the positive effect of public recognition on charitable giving in a field study. By ruling out the desire to motivate others' contributions, they concluded that donors' concern for social image could be the major motivation behind the publicity effect.
Other evidence from the lab and the field are consistent: subjects act more pro-socially when their behaviors could be known to others compared to when their behaviors are private information. For instance, using a public goods contribution game. Rege and Telle (2004) showed that revealing subjects' identities and contributions led to significantly higher contributions. Andreoni and Petrie (2004) found a similar effect on contributions when subjects' identities were publicly revealed using photographs. Pan and Houser (2011) showed that displayable rewards led to larger contributions compared to non-displayable rewards, especially among male subjects. Samek and Sheremeta (2013) in their experiment showed that recognizing low donors had an equivalent effect to recognizing all donors in increasing funds, which suggests that the publicity effect is related to shame aversion. In a field study, Soetevent (2005) found that using open collection baskets raised more funds than closed collection bags in 30 Dutch churches. It implies that even a small-scale visibility could have a significant impact on individual's giving behavior.
To model donors' behaviors under category reporting, Cartwright and Patel (2013) considered a signaling model in which observers' inferences determined donors' social image benefits when entering a higher category. Together with Harbaugh's prestige model (Harbaugh 1998a), both studies suggest that the preference of having one's prosocial behavior seen by others is the key factor of category reporting. We argue, however, that in addition to this publicity effect, category reporting also shapes donors' giving behavior through the category effect that accompanies the public reporting. The category effect may well be present even when donors qualifying for the category are not publicly reported.
B. The Category Effect
Two factors might give rise to the category effect. The first one is the presence of the category threshold. The prespecified category threshold could serve as a reference donation point that facilitates donors to make more informed decisions. Shang and Croson (2009) found the effect of reference donation in a field experiment. They showed that providing information to telephone donors on the amount given by another donor changed the distribution of donations in an on-air fundraising campaign for a public radio station. In a field study, Edwards and List (2014) found that suggested donations not only increased the response rate among potential donors, but also attracted donors to give the suggested amount. (3) In the lab, Croson and Marks (2001) found that having suggested contributions increased the rate of public goods provision among subjects who are assigned heterogeneous valuations of the public goods. The analysis showed that the suggested amount led to a higher level of conformity when subjects value the public goods differently. Similarly, the mere presence of category threshold in category reporting could be taken as information provided by the charity that would cause conformity of donation decisions. The category setting would also restrict a donor's choice. Barbieri and Malueg (2013) in a private values subscription model showed that under certain conditions, restrictions on donors' choices would yield a greater expected contribution than a flexible scheme.
The second factor linked to the category effect is the potential self-image benefits that accrue from being enrolled in the honored category, even when the public reporting of donations is muted. The psychology literature documents self-enhancement as one of the motivations for pro-social behavior, which suggests that people would take pro-social actions to enhance or recover a desirable self-image even if there are no other observers (Brown and Smart 1991). Johansson-Stenman and Martinsson (2006) provided evidence that self-image concerns are influenced by an individual's own perception of their preferences. Most of the respondents in a self-report survey overestimated their own social responsibility concerns relative to others. In another study, Murnighan (2001) correctly predicted dictator's pro-social behavior with restrictions based on the self-impression management motivations.
C. Behavioral Hypotheses
As argued above, two effects are present in the mechanism of category reporting: the category effect and the publicity effect. The category itself provides reference donation point and private-image benefits that might affect a donor's donation decision. A properly set category would attract donors to give or give above the threshold amount, and thus increase total donations. This effect is independent of public reporting.
HYPOTHESIS 1 : The use of category would increase donations relative to no use of category due to reference point and private-image benefits.
The reference effect and private-image benefits of the category setting have limited power in motivating giving. As Imas (2014) found in a real-effort task, the intrinsic pro-social motive works only when the stake is small. Similarly, the private information of being a generous donor could motivate individuals only at moderate category levels. When category threshold is set high, private-image benefits are not sufficient enough to offset the monetary cost of charitable giving. In this case, public reporting would provide additional social-image benefits to qualifying donors by conferring certain status in their social surroundings. Compared with the mere use of category, public reporting would attract more donors to attain the category level, especially for a high category level where the reference effect and private-image benefits take limited effect.
HYPOTHESIS 2: The use of public reporting would further increase donations due to the additional social-image benefits, especially when category level is high.
This donation-enhancing ability of providing social-image benefits through public reporting can be rationalized by the desire of donors to signal their high type (more generous) in order to set themselves apart from the low type (less generous). Samek and Sheremeta (2015) looked at selective recognition in charitable giving and found that recognizing only highest contributors had a pronounced effect due to the high signaling value attached to the recognition. Similarly, as the category level increases, a larger monetary sacrifice is required for a donor to be acknowledged; meanwhile, the social-image benefits also increase. Social-image benefits become particularly valuable when only a small number of donors are in the roll. For donors who have a stronger desire for social image benefits, to be acknowledged as "very generous" would supplement the limited reference effect and private-image benefits and outweigh the loss of monetary benefits at a high category level. Thus, category reporting allows charity to capture these donors and solicit high donations from them.
III. EXPERIMENTAL DESIGN AND PROCEDURE
A. Experimental Design
The experiment was designed to investigate the role of the publicity effect and the category effect in motivating subjects to donate to a charitable cause beyond the amount that would be donated based on their intrinsic (baseline) motivations. To measure these two treatment effects, three donation reporting plans were employed in our experiment, namely NR. CR, and CNR.
Table 1 shows reporting plans of NR. CR. and CNR. As a baseline, NR resembles the usual anonymous donation plan. Driven by intrinsic motivations (Andreoni 1989, 1990) (4), subjects in the NR treatment made independent and anonymous donation decisions. Under CR, a category known as "star donor" was set by the experimenter. Subjects who gave the threshold amount or above would be placed in the honored roll and acknowledged in full view of all the other subjects in the same session. CNR is a variant of CR. It was designed to remove the publicity effect while to retain the category setting element. In CNR, the same category plan was used as in CR\ however, the list of donors who obtained the "star donor" status was not publicly disclosed. It is similar to a donation solicitation plan that utilizes a suggested donation amount.
Thus, by comparing NR and CNR we would be able to evaluate the role of the category effect in increasing donations while removing the publicity effect. The only distinguishing factor in these two treatments is the use of donation category. By comparing CNR and CR, we would be able to evaluate the role of publicity effect in increasing donations while removing the category effect. The only distinguishing factor in these two treatments is the presence of publicity.
Two endowments were assigned to subjects: 100 points (equivalent to 10 Singapore dollars or 7.69 US dollars) and 200 points (equivalent to 20 Singapore dollars, or 15.38 US dollars). Subjects in each session were to decide how many points out of each endowment they would like to donate to the charity, and the remaining points became their take-home earnings. Decisions under two endowments were made at the same time, but only one was randomly chosen for real donation and payment. Two levels of endowment were designed to test the income effect on donation decisions.
For simplicity, we set only one category threshold for each CR and CNR treatment and varied it between treatments to test the category effect and the publicity effect at four different levels. We maintain the proportion of each threshold level relative to the endowment. For the 100-point endowment, the four category levels were set at 20 points, 40 points, 60 points, and 80 points. For the 200-point endowment, the four category levels were set accordingly at 40 points, 85 points, 120 points, and 160 points. Altogether, we had nine experimental sessions: four CR sessions, four CNR sessions, and one NR baseline session.
B. Experimental Procedure
The experiment was conducted at Nanyang Technological University in Singapore. We recruited undergraduate students from various majors. Each subject was randomly assigned to one of the nine sessions, and each subject only participated once.
Upon entering the lab, subjects randomly picked a two-digit seat number, which was used as their ID throughout the session. A seating plan was projected on the screen placed at a prominent place inside the lab for all subjects to see. Panel A of Figure 1 illustrates the seating plan where subjects' seat numbers were clearly indicated. Thus, although subjects were seated a distance apart from each other, they still had a visually unobstructed view of one another. The visibility of subjects was crucial in enhancing the saliency of the public reporting, while the sparse seating arrangements were necessary to ensure the independence of subjects' decisions.
In the treatments where public reporting was absent, that is, the NR and CNR treatments, the seating plan remained on the screen throughout the session. Decisions were kept anonymous for all these treatments. In CNR, subjects would automatically know their status once they made their decisions given the prespecified category, while the experimenter could not tell whether a subject was on the honored list or not. Effort was made to ensure that both NR and CNR were conducted in the double-blind fashion. In CR, the seating plan on Panel A served as the platform for public acknowledgment. Subjects who qualified for the "star donor" status were invited to stand up while the experimenter would place a "star" on his or her position on the seating plan. The rest of the subjects remained anonymous. Panel B of Figure 1 illustrates what the seating plan might look like at the end of a CR session.
The charity chosen as the beneficiary of this donation experiment was Doctors Without Borders (Medecins Sans Frontieres), a secular, nongovernmental, and nonprofit international charitable organization that provides urgent medical care and humanitarian aid to war-tom regions and to developing countries struck by natural disasters. A brief video clip describing the organization and its work was shown to subjects at the beginning of the experiment. After everyone had made their decisions, we tallied the total donations raised in the session given the binding endowment and donated it to the charity through its online donation portal. The whole donation process was projected on the screen for all participants to witness. We took great care to ensure that subjects were fully aware of this "live" online donation process at the beginning of the experiment. Once the online donation was completed, we would receive an email acknowledgment from the charity and this email was shown to all subjects. The remaining endowment points that were not donated were placed in envelopes and paid to subjects privately in cash at the end of the session.
IV. EXPERIMENTAL RESULTS
Table 2 presents the summary of statistics for the 200-point endowment. We use the donation decisions from the 200-point endowment for the main results, and will briefly discuss the results from the 100-point endowment at the end of the section. In general, these two endowments yield qualitatively similarly results. We see that CR outperformed NR in generating donations, except for the lowest category threshold. The average donation under NR was 68.28 points, which captured subjects' intrinsic motive for giving. Compared to the standard dictator game, the "real donation" game yields a higher average contribution. Camerer (2003) documented 10% average contribution rate in the former case. In the latter case, Eckel and Grossman (1996) reported 31% average contribution rate to the Red Cross, and Li et al. (2011) reported 27% average contribution rate to private charities. The average contribution rate under NR in our experiment is 34%, which is slightly higher but still within the ball-park figure.
For the CR treatments, the average donations were respectively 88.88 points (CR1), 103.74 points (CR2), 106.63 points (CR3), and 128.18 points (CRA). The Mann-Whitney test for the equality of means shows that the average donations under CR1 were not significantly different from that under NR, while the rest of the CR treatments (i.e., CR2, CR3, and CRA) yielded significantly higher average donations than the NR treatment (p values were . 1543, .0043, .0030, and .0001, respectively). Our result showed that CR was superior to NR in terms of incentivizing participants to make higher donations. As mentioned earlier, both the category effect and the publicity effect have the ability to increase donations. The question is how the two effects interact and which of the two effects dominates.
Table 2 also depicts the donation patterns. On average, subjects donated around 99.51 points out of their initial 200-point endowment. About 3% of subjects did not donate to the charity at all, and about 12% of subjects donated all the endowment to the charity. Meanwhile, approximately 58% of subjects donated more than half of their endowment to the charity. (5)
A. The Category Effect
We compared donations from the CNR treatments and the baseline NR treatment to measure the category effect. The difference between these two reporting plans is the presence of the star donor category. Subjects who qualified for the category were acknowledged as star donors but did not obtain any public award of status. As mentioned above, two possible mechanisms, the reference donation and the self-image benefits, should affect subjects' donation decisions in treatments where such category is present even when the public recognition of the qualifying donors is absent.
Table 2 shows that the average donations obtained under the NR treatment were 68.28 points. Meanwhile, the average donations obtained under the CNR treatments were 104.61 points (CNR1), 97.26 points (CNR2), 113.11 points (CNR3), and 86.45 points (CNRA), respectively. With the exception of the highest category level (CNR4), the CNR treatments resulted in significantly higher average donations than those accrued under the NR treatment. The p values from the Mann-Whitney test between the CNR treatments and the NR treatment are .0243, .0495, .0006, and .2164 at the respective category levels. This result shows that the mere presence of the category setting significantly increased donations when the category threshold was set at modest levels. However, when the category threshold was set at an excessively high level, that is, 80% of the endowment, the category setting alone was not sufficient to induce donors to give a greater amount than the amount they would give in the baseline treatment.
The effectiveness of the category setting at the low and medium category levels is evident not only from its impact on average contributions, but also from its impact on the distribution of donations. Figure 2 depicts the distribution of donations in the NR treatment and Figure 3 depicts that in the CNR treatments. The vertical arrows on the horizontal axis in Figure 3 denote the respective category threshold levels across treatments.
In the NR treatment, where category setting and public reporting were absent, donations were driven by donors' intrinsic pro-social preferences. We observe that over 40% of subjects gave at least 100 points or more, and about 63% of subjects gave at least 50 points or more. It is also noteworthy that five focal contribution levels emerged in the NR treatment, namely 10 points, 50 points, 100 points, 150 points, and 200 points. Most of them are multiples of 50 points. Without external influences, multiples of 50 points became the natural focal donation points for donors. The order of salience among these focal reference donations were, respectively, 100 points, 50 points, and 10 points.
The following observations are obtained from the comparison of the distributions of donations in the CNR treatments with that in the NR treatment. First, as the category threshold level increased, donations became clustered at or near the category threshold level. The only exception to this was CNR4, where the category threshold level was set the highest. In CNR 1, where the category threshold was set at 40 points, the focal donation point was also 40 points. Furthermore, 90% of the subjects gave 40 points or more, which is a sharp contrast to the distribution of donations in the NR treatment. Similarly, 85 points became a focal donation point in CNR2, and 120 points became a focal donation point in CNR3. Neither of these two donation points were the natural focal donation points in the NR treatment. This change in behavior is attributed to the presence of the category brackets. Furthermore, when we examine the interval of 20 points from either side of each threshold level, we find that more donors gave the threshold amount or above than those who gave less than it. Donors were not only attracted to the category brackets, but also willing to give more than the threshold level even when one's giving behavior was not made public.
Second, in CNR4, where the category threshold level was set excessively high, that is, 160 points (80% of the endowment), many donors ignored the category bracket. Only about 20% of the donors (6 out of 31 subjects) gave the threshold amount or above, a proportion much smaller compared to that found in the other threshold levels. Instead, the donors donated in a way similar to what they would do in the NR treatment. For instance, less than 50% of donors gave 100 points or above. In addition, the focal donation point in this treatment was 100 points, the same as that in NR. The p value from the Kolmogorov-Smirnov test for the equality of the distributions between CNRA and NR is .624, indicating that the two distributions are not statistically different. All in all, we show that the category effect is present only when the category threshold is set at modest levels, but not when it is set at the highest level.
B. The Publicity Effect
To identify the publicity effect, we pair-wisely compare donations solicited from the CR treatments and the CNR treatments at each threshold level. The only difference between these two reporting plans is the public disclosure of the qualifying donors. Table 2 shows the average donations. At the lowest threshold level (20% of the endowment), the average donations were 88.88 points under CR 1 and 104.61 points under CNR 1 (p=.3353). At the second lowest threshold level (42.5% of the endowment), the average donations were 103.74 points under CR2 and 97.26 points under CNR2 (p=.6496). At the third lowest threshold level (60% of the endowment), the average donations were 106.63 points under CR3 and 113.11 points under CNR3 (p=.7580). Also at the highest threshold level (80% of the endowment), the average donations were 128.18 points under CRA and 86.45 points under CNRA (p=.0061). Thus, it is only at the highest threshold level that the average donations in the two reporting plans differ significantly. This is a contrasting result from our previous result on the category effect, where we showed that the category effect was not effective when the threshold was set at the highest level. Here, in contrast, the publicity effect is only effective when the category threshold was set at the highest level.
Using the data obtained from the CR and CNR treatments, we ran the following OLS dummy variable regression:
(1) [[gamma].sub.i] = [[beta].sub.0] + [[beta].sub.1][D.sup.publicity.sub.i] + [[beta].sub.2][D.sup.gender/sib/i] + [[beta].sub.3][D.sup.nationality.sub.i] + [[epsilon].sub.i]
where [y.sub.i] denotes donor Vs donation points, [D.sup.publicity.sub.i] is a dummy variable for public reporting which takes the value of 1 for CR, and 0 for CNR. We also include a dummy variable for gender. [D.sup.gender.sub.i], which takes the value of 1 for female and 0 for male, a dummy variability for the donor's nationality, [D.sup.nationality.sub.i], which takes the value of 1 for Singaporean and 0 for non-Singaporean. Table 3 presents the regression results. It can be seen that with and without demographics, publicity was significant only at the highest threshold level.
The publicity effect at the highest threshold level not only affects the average donations, but also the distribution of donations. Figure 3 illustrates the distribution of donations for all CNR treatments, while Figure 4 illustrates those for all CR treatments. A comparison between the two figures shows that the distributions of donations are similar between CR and CNR for the first three threshold levels, while they are statistically different for the highest threshold level. The p values from the Kolmogorov-Smirnov test for the equality of distributions were .783, .815, .685, and .019, respectively.
At the highest threshold level, many more donors chose to qualify for the star donor category in CR4 than those in CNR4. Table 2 summarizes the behavioral patterns from these treatments. It can be seen that 15 donors (45% of the subjects) gave no less than the threshold level in CR4, while there were only six donors (19% of the subjects) who did so in CNR4. Also, there were more donors who gave no less than 100 points (half of their endowment) to the charity in CR4 than in CNR4: 27 donors (82% of the subjects) in CR4 and 15 donors (48% of the subjects) in CNR4. For lower category threshold levels, these numbers are fairly comparable. Relative to CNR4, public reporting in CR4 shifted its distribution of donations to the right tail. The above results suggest that public reporting would increase donations above the category effect only when the category threshold was set relatively high. The additional social-image benefits were not salient at lower category levels.
C. Donors' Giving Behavior across Different Endowments
Two levels of endowment, 100 points and 200 points, were designed to test the income effect, that is. if the donation rate would change as income rises. In this section, we show that the qualitative aspects of our results remain the same between two endowments. Figure 5 depicts the average donation rates, that is, the donation amount as a proportion of the endowment, across treatments. It can be seen that donors exhibited similar donation behavior between these two endowment levels. The Wilcoxon matched-pair test results confirm that they were not statistically significantly different from each other.
The pairwise correlations, conditional on the treatment, between the donation rates accrued from different endowment levels by subject ranged from 0.83 to 0.96. Donors tended to maintain the same proportion of donation across endowment levels. Thus, the variation in income level does not affect the donation rates in our experiment.
V. CONCLUDING REMARKS
Category reporting has been widely used by charities for their donation soliciting campaigns. In this article, we evaluate the performance of category reporting in increasing donations relative to the standard anonymous donation plan. We argue that in addition to the publicity effect that caters for donor's taste for prestige, the category threshold which is essential to category reporting also affects donors' giving decisions.
We tested these two effects in a lab experiment where subjects were given money to donate to a real charitable cause. In order to single out the publicity effect, we introduced a reporting plan known as category no reporting, in which we retained the same category setting as in category reporting, but kept private the information about the status donors had obtained. A comparison between category reporting and category no reporting allows us to infer the importance of the publicity effect, while a comparison between category no reporting and no reporting allows us to infer the importance of the category effect.
The result shows that both the publicity effect and the category effect contribute to the donation soliciting capability of category reporting. When we set the category threshold at a low or medium level, the mere presence of the category setting significantly increased donations, while acknowledging donors publicly did not add on to that effect. However, when we set the category threshold at a high level, the category setting alone failed to boost donations, while acknowledging donors publicly complemented the category setting and increased the donations significantly. On the basis of our results, we conclude that the category effect is more effective for low-category threshold levels, while the publicity effect is more effective for sufficiently high-category threshold levels. Without disentangling these two confounding effects, one might overestimate the effect of public acknowledgment associated with low-category thresholds. (6)
In practice, when setting the appropriate category threshold, charities employ various approaches. Many of them set the threshold based on past average donation amounts. Still many others may decide to give individual donors specific suggested amounts based on previous donations for these individuals.
Finally, one caveat is in order. In our experiment, we considered only one category threshold level. In reality, charities employing a category reporting plan often use multiple category threshold levels. The higher the category, the more prestigious the status awarded to donors who qualify for the category. It would therefore be necessary to extend our experimental analysis to the case of multiple category threshold levels for a more general insight into the behavior of contributors and the mechanism of the category reporting strategy.
ABBREVIATIONS CNR: Category No Reporting CR: Category Reporting NR: No Reporting
Online Early publication May 2, 2016
Andreoni, J. "Giving with Impure Altruism: Application to Charity and Ricardian Equivalence." Journal of Political Economy, 97(6), 1989. 1447-58.
--. "Impure Altruism and Donations to Public Goods: A Theory of Warm-Glow Giving." The Economic Journal, 100(401), 1990, 464-77.
Andreoni, J., and B. Bernheim. "Social Image and the 50-50 Norm: A Theoretical and Experimental Analysis of Audience Effects." Econometrica, 77(5), 2009. 1607-36.
Andreoni. J., and R. Petrie. "Public Goods Experiments without Confidentiality: A Glimpse into Fund-Raising." Journal of Public Economics, 88(7-8), 2004, 1605-23.
Barbieri, S., and D. A. Malueg. "Increasing Fundraising Success by Decreasing Donor Choice." Journal of Public Economic Theory, 16(3), 2013, 372-400.
Benabou, R., and J. Tirole. "Incentives and Prosocial Behavior." American Economic Review, 96(5), 2006, 1652-78.
Brown. J. D., and S. A. Smart. "The Self and Social Conduct: Linking Self-Representations to Prosocial Behavior." Journal of Personality and Social Psychology, 60(3), 1991, 368-75.
Camerer, C. Behavioral Game Theory: Experiments in Strategic Interaction. Princeton: Princeton University Press, 2003.
Cartwright. E., and A. Patel. "How Category Reporting Can Improve Fundraising?" Journal of Economic Behavior & Organization, 87, 2013, 73-90.
Croson, R., and M. Marks. "The Effect of Recommended Contributions in the Voluntary Provision of Public Goods." Economic Inquiry, 39(2), 2001, 238-49.
Crumpler, H., and P. J. Grossman. "An Experimental Test of Warm Glow Giving." Journal of Public Economics, 92(5), 2008, 1011-21.
Eckel, C. C., and P. J. Grossman. "Altruism in Anonymous Dictator Games." Games and Economic Behavior, 16(2), 1996, 181-91.
Edwards, J. T., and J. A. List. "Toward an Understanding of Why Suggestions Work in Charitable Fundraising: Theory and Evidence from a Natural Field Experiment." Journal of Public Economics, 114, 2014, 1-13.
Harbaugh, W. "The Prestige Motive for Making Charitable Transfers." American Economic Review, 88(2), 1998a, 277-82.
--. "What Do Donations Buy? A Model of Philanthropy Based on Prestige and Warm Glow." Journal of Public Economics, 67, 1998b, 269-84.
Imas, A. "Working for the "Warm Glow": On the Benefits and Limits of Prosocial Incentives." Journal of Public Economics, 114, 2014, 14-18.
Johansson-Stenman, O., and P. Martinsson. "Honestly, Why Are You Driving a BMW?" Journal of Economic Behavior & Organization, 60, 2006, 129-46.
Karlan, D., and M. A. McConnell. "Hey Look at Me: The Effect of Giving Circles on Giving." Journal of Economic Behavior & Organization, 106, 2014, 402-12.
Li, S. X., C. C. Eckel, P. J. Grossman, and T. L. Brown. "Giving to Government: Voluntary Taxation in the Lab." Journal of Public Economics, 95(9), 2011, 1190-201.
Murnighan, J. K. "Player Types and Self-Impression Management in Dictatorship Games: Two Experiments." Games and Economic Behavior, 37(2), 2001, 388-414.
Pan, X., and D. Houser. "Competition for Trophies Triggers Male Generosity." PLoS One, 6(4), 2011, el8050.
Rege, M., and K. Telle. "The Impact of Social Approval and Framing on Cooperation in Public Good Situations." Journal of Public Economics, 88, 2004, 1625-44.
Reiley, D., and A. S. Samek. "How Do Suggested Donations Affect Charitable Gifts? Evidence from a Field Experiment in Public Broadcasting." CESR-Schaeffer Working Paper No. 2015-031.SSRN, 2015. Accessed February 17, 2016. http://ssrn.com/abstract=2702174.
Samek, A. S., and R. M. Sheremeta. "Recognizing Contributors: An Experiment on Public Goods." Experimental Economics, 17(4), 2013, 673-90.
--. "Selective Recognition: How to Recognize Donors to Increase Charitable Giving." Working Paper 2695236, SSRN, 2015. Accessed February 17, 2016. http://ssrn.com/abstract=2695236
Shang, J., and R. Croson. "A Field Experiment in Charitable Contribution: The Impact of Social Information on the Voluntary Provision of Public Goods." Economic Journal 119(October), 2009, 1422-39.
Soetevent, A. R. "Anonymity in Giving in a Natural Context--A Field Experiment in 30 Churches." Journal of Public Economics, 89, 2005, 2301-23.
Tonin, M., and M. Vlassopoulos. "Disentangling the Sources of Pro-Socially Motivated Effort: A Field Experiment." Journal of Public Economics, 94, 2010, 1086-92.
Zizzo, D. J. "Experimenter Demand Effects in Economic Experiments." Experimental Economics, 13(1), 2010, 75-98.
JINGPING LI and YOHANES E. RIYANTO *
* We thank Jonathan Meer, Martijn van den Assem, Parimal Bag. Timothy Cason, Haoming Liu. Lise Vesterlund, Wong Wei Kang, and seminar participants at the European Regional Meeting of the Economic Science Association (Lyon, France), the Conference on the Economics of Charitable Giving (Mannheim, Germany), the Asia Pacific Conference of the Economic Science Association (Kuala Lumpur, Malaysia), the World Meeting of the International Economic Association (Beijing, China), the International Economic Science Association Conference (Chicago, USA), the Asian Meeting of the Econometrics Society (Seoul, Korea), and the Annual Conference of the Society for the Advancement of Behavioral Economics (Granada, Spain) for their useful comments. The usual disclaimer applies.
Li: Assistant Professor, Center for Economic Research, Shandong University, Jinan 250100. P.R. China. Phone 88363816, Fax 88364981, E-mail firstname.lastname@example.org
Riyanto: Associate Professor, Division of Economics, School of Humanities and Social Sciences, Nanyang Technological University, Singapore 637332, Singapore. Phone 67906271, Fax 67936303, E-mail email@example.com
(1.) Builders Without Borders is an international network for sustainable building solutions to homelessness. See http:// www.builderswithoutborders.org for more information.
(2.) See http://www.rotary.org for more information.
(3.) The empirical facts about the effect of suggested amount is inconclusive, though. Reiley and Samek (2015) found in a recent field study on direct mail fundraising that a 20% increase in the suggested amount led to a significant decrease in response rate and total revenue.
(4.) Altruism and warm-glow are intricately interwoven as individual intrinsic motivations of giving. In this article, we do not distinguish these two motives from one another. See Crumpler and Grossman (200S) for a study that tests the warm-glow giving, and Tonin and Vlassopoulos (2010) for a study that separates warm-glow from altruism.
(5.) Notice that there are only few participants who gave 0 to the charity when the threshold is high. This result may not extend to the field. It is likely that many actual donors in the field would give 0 if the category threshold employed is very high. This is because in the field there is less "pressure" from being in the lab to give. It would be interesting to explore the differences in donating behaviors exhibited in the lab and in the field in further study.
(6.) We gratefully thank an anonymous referee for bringing up this excellent point for discussion. In experiments studying charitable giving and category reporting, the experimenter demand effect is always a possibility (Zizzo 2010). Indeed, in treatments where the category threshold is utilized, it is hard to tell whether the reason for a subject to choose to give above the threshold is a response to social/private-image benefits attached to the category or it is driven by the experimenter demand effect. We would, however, argue that the experimenter demand effect in our treatments could actually enhance the external validity of the experiments because the way we introduced the game mirrored a charity fundraising campaign. In such a setting the presence of the experimenter demand effect is not really problematic (Zizzo 2010).
TABLE 1 The Reporting Plans Reporting Plans Category Public Setting Reporting No Reporting (NR) No No Category Reporting (CR) Yes Yes Category No Reporting (CNR) Yes No TABLE 2 Summary of Statistics Treatments Average Donations Giving Profiles (Endowment = 200 points) (Threshold) (SD) N 0 pts 200 pts l NR 68.28 (52.06) 32 1 2 2 CR1 (40 pts) 88.88 (60.12) 32 0 4 3 CR2 (85 pts) 103.74 (48.21) 31 1 2 4 CR3 (120 pts) 106.63 (57.25) 30 0 5 5 CR4 (160 pts) 128.18 (61.00) 33 0 7 6 CNR1 (40 pts) 104.61 (61.23) 28 0 5 7 CNR2 (85 pts) 97.26 (58.66) 31 3 2 8 CNR3 (120 pts) 113.11 (51.30) 28 0 3 9 CNR4 (160 pts) 86.45 (56.13) 31 2 2 Total 99.51 (58.01) 276 7 32 Percentage 2.54% 11.59% Treatments Giving Profiles (Endowment = 200 points) (Threshold) [greater than =Threshold or equal to] 50% l NR 13 -- 2 CR1 (40 pts) 14 5 3 CR2 (85 pts) 20 0 4 CR3 (120 pts) 18 7 5 CR4 (160 pts) 27 6 6 CNR1 (40 pts) 15 5 7 CNR2 (85 pts) 17 3 8 CNR3 (120 pts) 21 9 9 CNR4 (160 pts) 15 2 Total 160 37 Percentage 57.97% 15.16% Treatments Giving Profiles (Endowment = 200 points) (Threshold) [greater than or equal to] Threshold l NR -- 2 CR1 (40 pts) 29 3 CR2 (85 pts) 22 4 CR3 (120 pts) 15 5 CR4 (160 pts) 15 6 CNR1 (40 pts) 27 7 CNR2 (85 pts) 21 8 CNR3 (120 pts) 16 9 CNR4 (160 pts) 6 Total 151 Percentage 61.89% TABLE 3 The Publicity Effect Category 1 (1) (2) Publicity -6.239 -7.866 1 for CR, 0 for CNR (8.107) (7.854) Gender 0.182 1 for Female, 0 for Male (8.055) Nationality 9.909 1 for Local, 0 for Others (7.606) Constant 46.51 *** 52.30 *** (9.297) (5.779) Observations 60 60 [R.sup.2] .039 .017 Category 2 (3) (4) Publicity 2.715 3.242 1 for CR, 0 for CNR (6.236) (6.819) Gender 22.85 *** 1 for Female, 0 for Male (7.458) Nationality 6.527 1 for Local, 0 for Others (6.078) Constant 28.83 *** 48.63*** (6.600) (5.268) Observations 62 62 [R.sup.2] .202 .004 Category 3 (5) (6) Publicity -2.674 -3.237 1 for CR, 0 for CNR (8.091) (7.128) Gender 5.483 1 for Female, 0 for Male (8.384) Nationality 0.269 1 for Local, 0 for Others (8.434) Constant 52.11 *** 56.55 *** (8.982) (4.844) Observations 58 58 [R.sup.2] .012 .004 Category 4 (7) (8) Publicity 22.22 *** 20.87 *** 1 for CR, 0 for CNR (7.392) (7.321) Gender -5.78 1 for Female, 0 for Male (7.889) Nationality 9.005 1 for Local, 0 for Others (7.637) Constant 42.41 *** 43.23 *** (7.624) (5.038) Observations 64 64 [R.sup.2] .137 .115 Note: Dependent variable: donation points in CR and CNR.
|Printer friendly Cite/link Email Feedback|
|Author:||Li, Jingping; Riyanto, Yohanes E.|
|Date:||Jan 1, 2017|
|Previous Article:||Testing commitment cost theory in choice experiments.|
|Next Article:||Fundraising intermediaries inhibit quality-driven charitable donations.|