23
TITLE: Charity and Gift Exchange in Pakistan Authors: Saima Naeem $ & Asad Zaman * $: State Bank of Pakistan (corresponding author: [email protected] ) *: Pakistan Institute of Development Economics ABSTRACT: Our goal was to replicate an elegant demonstration of the Gift Exchange Mechanism (GEM) by Falk (2007) in a door-to-door charity raising campaign. To our surprise, we were unable to replicate these results in Pakistan; instead of increasing with gift size, donations decreased. Also, we encountered two phenomena not previously reported in the literature: gift refusals and hospitality. Because we did not anticipate these effects, our experiment did not provide significant information regarding their causes. Indirect evidence suggests cultural and religious norms for charity may be responsible for the observed differences. Further experimentation is required to provide better evidence for our hypotheses in this regard. Key Words: Gift Exchange, Charitable Giving, Culture, Hospitality, Gift Refusals JEL-Classification: C93, D63

Charity and Gift Exchange in Pakistan

Embed Size (px)

Citation preview

TITLE: Charity and Gift Exchange in Pakistan

Authors: Saima Naeem$ & Asad Zaman*

$: State Bank of Pakistan (corresponding author: [email protected])

*: Pakistan Institute of Development Economics

ABSTRACT:

Our goal was to replicate an elegant demonstration of the Gift Exchange Mechanism (GEM) by Falk

(2007) in a door-to-door charity raising campaign. To our surprise, we were unable to replicate

these results in Pakistan; instead of increasing with gift size, donations decreased. Also, we

encountered two phenomena not previously reported in the literature: gift refusals and hospitality.

Because we did not anticipate these effects, our experiment did not provide significant information

regarding their causes. Indirect evidence suggests cultural and religious norms for charity may be

responsible for the observed differences. Further experimentation is required to provide better

evidence for our hypotheses in this regard.

Key Words: Gift Exchange, Charitable Giving, Culture, Hospitality, Gift Refusals

JEL-Classification: C93, D63

1. INTRODUCTION:

The history of economic literature is rich with contributions on gift exchange mechanism (GEM,

hereafter) and motives for giving charity. Pakistan is a developing country with strong social ties

and charity culture is rich. According to world giving index (2012)1, Pakistan is ranked 7th in the

world in helping the strangers and 65th in donating money. To study GEM in the context of

charitable behavior, we attempted to replicate an experiment of Falk (2007) designed around these

elements. In a postal campaign, Falk included no gift, a small gift and a large gift along with the

request for a donation to a charitable cause. As predicted by GEM, the donations received increased

with gift size.

Our experimental setup differs from that of Falk (2007) in that, instead of a postal campaign, we

conducted a door-to-door fund raising campaign due to low expected response in postal campaign.2

Contrary to our expectations, we were unable to replicate Falk’s results in Pakistan. Also, we

encountered two phenomena not previously reported in the literature: gift refusals and hospitality.

We speculate that cultural effects led to the different results. Since the results were not anticipated,

our experimental design could not differentiate between plausible alternative hypotheses which

could explain these results. Further work is required to establish the relevance and significance of

potential alternative cultural factors.

1.1 EXPERIMENTAL DESIGN

The original experimental design of Falk (2007) is simple and elegant. Letters soliciting charitable

contributions were sent to 10,000 potential donors. One third of the letters contained no gift; a

second third included a small gift of one postcard, while the final third contained a large gift of four

postcards. There was a clear and significant response: donations increase with gift size. Thus Falk

established the significance of reciprocity and gift exchange, in addition to “warm glow” as a motive

for charitable contributions.

Our experiment was designed to replicate the results of Falk. Some changes to the design were

essential in the context of Pakistan. Financial penetration is low, and sending checks by mail is rare,

so a postal campaign would have a huge “no response”.3 Instead, we decided to use a door-to-door

campaign. This is a more expensive method, and budget constraints forced a reduction to a planned

sample size of 800. We used essentially the same small and large gifts for a charity campaign for a

small group of doctors in a famous hospital of Islamabad, Pakistan. This organization is working for

poor and needy patients who are not able to pay for their medication and other major medical

expenses. A detailed description of experimental design is provided in an appendix B.

1 Available at: https://www.cafonline.org/PDF/WorldGivingIndex2012WEB.pdf 2 Cash is overwhelmingly used for transactions in Pakistan, and most people do not have checking accounts and sending checks by mail, whether for shopping or other purposes, is virtually unknown. 3Pakistan has one of the lowest financial penetration levels in the world with 56% of the adult population totally excluded, and another 32% informally served (source: http://www.sbp.org.pk/press/2012/MobileConf-15-Mar-2012.pdf )

Our experimental setup included a mid-experiment detailed meeting with the students hired as

solicitors for charitable contributions. In this meeting, our solicitor teams reported two very

interesting observations that we had not anticipated. These were: hospitality and gift refusals.

Solicitors were asked to properly record these two observations in the latter half of experiment.

1.2 GENDER AND INCOME

In addition to simple gift exchange, our experimental design also allowed us to assess the effects of

Gender and Income on donations. Here our findings are more or less consistent with those reported

in the literature with some minor variations.

Gender is among one of the consistent determinants of charitable behavior; relevant literature

shows that females in general are usually more generous in charitable giving than males (Leslie,

Snyder, and Glomb 2012; Einolf 2011; Mesch et al. 2011; Kamas, Preston, and Baum 2008; Piper

and Schnepf 2008; Mesch, et al 2006; Meier 2005). This finding is in conformity with our results. A

few of these studies also find that females are more generous in that a greater proportion

contributes, but when they donate, men donate more (Piper and Schnepf 2008; Andreoni and

Vesterlund 2001). This does not occur in our study, our studies are also in conflict with the findings

of a few studies where gender differences are either insignificant (Bolton and Katok 1995; Lo and

Tashiro 2012), or male donate larger amounts (Andreoni and Vesterlund 2001).4

The motives for charity giving also differ across gender; women are more likely to give to causes

related to education and health care (Einolf 2011; Mesch et al. 2011; Piper and Schnepf 2008),5

human services (e.g. child care, help for poor and homeless people) (Marx, 2000) and poverty

related issues (Regnerus, Smith, and Sikkink 1998). Male donations, on the contrary, are more likely

to be in sports, civil rights and adult recreations (Einolf 2011). Since we solicited charity for health

care, and females donated more, our results are in conformity with these findings.

Our study shows that largest donations are from females to female teams of solicitors. This finding

is different from a few earlier findings supporting cross gender favors (Landry et al. 2006;

Andreoni and Petrie 2008). Also, Ben-Ner et al. (2003) find lesser likelihood of women being

generous while giving to other women in dictator game setting. Similarly, our findings conflict with

Kamas, Preston, and Baum (2008) who found that men tend to give more to women-only teams in

dictator game. Cultural norms governing cross-gender interactions differ sharply from western

norms, and are a likely source for these differences. However, our findings also conflict with

Razzaque (2009), who studies gender effects in the ultimatum game with Pakistani students as

subjects. He finds that Pakistani cultural norms of courtesy towards females lead males to make

hyper-fair offers to female in the ultimatum game. It seems likely that this is due to difference

between norms among fellow students versus those prevailing in the traditional culture at large,

but our study was not designed to assess this.

Another intriguing finding regarding charitable behavior shows that women donate more when

altruism is expensive, but when it is cheap, men are more altruistic (Andreoni and Vesterlund 2001;

4 See Wit and Bekkers (2012) for discussion on differences in results.(Wit & Bekkers, 2012) (Bekkers R. , Measuring Altruistic Behav ior in S urveys: The Al l-or- Not hing Dictator Game, 2007)(Bekkers R. , George Gives to Ge ology Ja ne: T he Name Letter Effect a nd Ot her

Similaritie s in Fundraising, 2010) 5 However, Bekkers (2007, 2010) finds that men are more likely to give to health causes.

Piper and Schnepf 2008). If we proxy altruism by hospitality, both men and women, as well as

donors and non-donors provide roughly equivalent hospitality. At the same time, women donate

more. This indirect evidence does not support the cited results, but it can only be taken as

suggestive, due to small sample sizes as well lack of experimental design to test this effect.

Another strand of literature show that female donations are motivated more by empathy

(Willer, Wimer and Owens 2012) and social pressure (Della Vigna et al 2013). Again, our

experiment was not designed to test for these effects, but provides some indirect evidence to

support these findings in the Pakistani cultural context.

Literature on income and charity has two major theories. Some studies find that charitable giving as

proportion of income declines with income (Auten and Rudney 1990; Bekkers 2004). In contrast, a

few studies have also reported a U-shape trend, stating high proportional donation level for lower

and higher income group, and lowest for middle income group (Hodgkinson & Weitzman 1996;

James & Sharpe 2007; Andreoni 2004). Analysis across income group in our data seems to support

the former view that shows declining charity as a proportion of income.

2. RESULTS

In 255 cases, there was insufficient contact between the team soliciting donations and the

households. Excluding these from the total sample of 800, our active sample consists of 545

households. About 20% of the households made some donation.

2.1 FAILURE OF GIFT EXCHANGE

From the summary statistics in table 1, the relative frequency of donation in no gift is 24 %, which

decreased to 20 % in small gift; there is further decrease to 13% in large gift group. The frequency

of donation trends exactly opposite to the prediction of GEM. Summary statistics in Table 1 indicate

that highest number of donations was collected in no gift treatment, while highest average

donations were raised in small gift treatment. The large gift group stands out to be least effective

group in term of frequency of donations that consequently decreased average donations.

Table1

Summary Statistics of Donation in each Treatment Group

No Gift Small Gift Large Gift

Number of household visited 187 235 123

Number of donations 44 48 16

Likelihood of Donation 0.24 0.2 0.13

Average Donations (in PKR.) 14.09 16.47 13.33

Max. of Donation Amount (in PKR.) 500 1000 1000

Average Contributions for Positive Donations (in PKR.) 59.9 80.6 102.5

While there is no evidence of GEM in term of frequency of donation, we do find that average

amounts of donation varied positively with the gift for those with positive donation. This fact

suggests that gifts might actually have a positive effect for those who decide to donate; they actually

give more after they got a larger gift, but this positive effect on donation amounts is completely

swamped by the decreasing relative frequency of donations. A more detailed discussion why GEM

might have failed is given in section 3.

Omitting significant covariates can occasionally reverse results based on simple ratios, as in

Simpson’s Paradox.6 A full fledged Probit and Logit analysis with all relevant observable regressors

on all observable relevant regressors is presented in Table 2 below. In the probit regression, the

dependent variable donation is defined as 1 for every positive donation, zero otherwise. Similarly,

we defined dummies for treatment groups and income level.7 Gender dummy is defined as “1” for

males, while all females were coded as “0”. We have also introduced dummies for gender of solicitor

teams i.e. both males (MM team), both females (FF team), and a male/female couple (MF team).

Dummy MM takes value 1 for Male-Male team and FF is 1 for female-female team. The second half

of the table 2 provides the results of tobit regression that investigate the effect of gift on the

donation amounts; here amount of the donation is regressed on the same set of variables. Both

regressions confirm the failure of GEM, with probit showing a significant negative effect of a large

gift on probability of donation. This is the only significant treatment coefficient in probit regression,

so gifts and gift size are either insignificant or have a negative effect, contrary to predictions of

GEM. Tobit regression also shows the insignificant effect of gift on the amount of donation collected.

Table 2

Regression Results

Probit Tobit

Coefficient Std. Error Prob. Coefficient

Std. Error Prob.

C -0.7 0.18 0.000*** -173.3 38.54 0.000***

Small Gift 0.0 0.17 0.872 18.1 32.84 0.581

Large Gift -0.5 0.20 0.020** -50.4 39.60 0.203

Male -0.4 0.14 0.005*** -84.2 26.32 0.001***

6 See, for example, Freidman, Pisani & Purves (1998) p. 17-20(Freedman, Pisani, & Purves, 1998) 7 Small gift treatment dummy is “1” if observation lies in small gift group and “0” otherwise. Similarly, high income group dummy is “1” if respondent has income in high income range and “0” otherwise.

High Income -0.1 0.15 0.359 8.9 29.13 0.759

Low Income 0.4 0.17 0.013** 64.8 32.85 0.048**

FF 0.5 0.19 0.006*** 102.2 36.44 0.005***

MF 0.4 0.19 0.038** 46.5 37.41 0.214

R2 7% 5%

***: Significant at 1 %, **: Significant at 5 %, *: Significant at 10 % level of significance

In addition to confirming the failure of GEM, the regressions show certain significant gender and

income effects which we discuss below, in separate subsections.

2.2 GENDER EFFECTS

Pakistani cultural norms relating to gender, and to cross-gender interactions, are substantially

different from western norms. Therefore, we can expect to see differences from existing

experimental studies conducted in the west. Some of the effects of these cultural differences are

documented by Razzaque (2009) and Naeem & Zaman (forthcoming) in the context of the

ultimatum game. We also find some differences in charitable donations as documented below. (Naeem & Zaman, forthcom ing )

2.2.1 Women donate more

We here compare both the likelihood and level of donation that both males and females make. Most

of the previous studies have found that women are more likely to donate and donate higher

amounts than men (Leslie, Snyder, and Glomb 2012; Einolf 2011; Mesch et al. 2011); yet we do find

few studies where gender differences are either insignificant (Bolton and Katok 1995; Lo and

Tashiro 2012), or male donate larger amounts (Andreoni and Vesterlund 2001).

Our results also parallel previous literature, both in term of amounted donated and likelihood of

donation. In our experiment, we found that female were more likely to donate compared to male

respondents. Out of total, 28 % of the respondents were female and their relative frequency to

donate was 29 percent; that is significantly higher than their male counterparts, i.e., 16 percent.

Probit results in table 2 also support the significance of gender effect on frequency of donation. In

term of amount donated, in our sample significantly higher amounts were also donated by female

respondents. Average donations by women were rupees 29.8; that is around threefold higher than

average male donations i.e., rupees 9.2.

Table 3

Summary Statistics of donation by Gender of Respondent

Male Female

Sample Composition (%) 72.1 27.9

Likelihood of Donation 0.16 0.29

Average Donations (in PKR.) 9.21 29.77

Average Contributions for Positive Donations (in PKR.) 56.6 102.8

Fisher's exact test for Likelihood: P-Value =0.001*

***: Significant at 1 %, **: Significant at 5 %, *: Significant at 10 % level of significance

2.2.2 Women solicitors elicit more charity

Gender differences on the part of solicitors are prominent as well. As shown by the Probit- Tobit

regressions (see table 2), the FF solicitor team received significantly more donations, and also

greater frequency of donations.8 In fact, any team with at least one female on it has greater

likelihood of the getting donation with insignificant difference between FF and FM. This differs

from standard results where cross gender donations are significantly higher (Landry et al 2006;

Andreoni and Petrie 2008). We attribute higher donation probabilities to female solicitors to

cultural norms of showing chivalry and courtesy towards females in Pakistan (Razzaque 2009;

Naeem and Zaman forthcoming).

Table 4

Summary Statistics of donation for Solicitor Teams

Likelihood of Donation

Average Donation (in PKR)

Positive Average Donation (in PKR)

M-M Team 16.2 12.5 77.1

F-F Team 29.2 27.7 95.0

M-F Team 25.6 11.0 43.0

Teams with at least one female 17.8 12.2 68.6

Interestingly, even though the frequency of donation to MF team is close to FF team, the average

size of donations are significantly smaller than FF team and statistically equal to MM teams. It

appears that courtesy to females in FM team is token: it increases the likelihood of donation, but not

the amount donated. Our results contrast with those of Kamas, Preston, and Baum (2008), who

found greater effect of social image in the male-female mixed environment and concluded that

both males and females become more altruistic in the presence of opposite gender.9 Again, it

appears likely that these differences are due to cultural norms governing cross-gender interactions.

2.3 INCOME EFFECTS

8 To test the hypothesis of female donations are largely driven by same gender favors, we run a probit regression by excluding female encounters to FF teams only (see table A in appendix A). The likelihood of female donations are still higher than male respondents showing greater generosity by female in general. 9 In our experiment the higher frequency of female donation is higher for all the team combination i.e., MM, FF and MF; however, we do not discuss the cross gender differences due to small sample size.

Charity to the poor is among the five founding pillars of Islam. Donating 2.5% of wealth as “Zakat” is compulsory for every adult Muslim every year. Additional charity beyond this minimal limit is also strongly encouraged. The Non-Zakat charity remains 54 percent of total monetary charity during 1998, while over half of the individuals are reported to give charity outside their families (Bonbright and Azfar 2000). In their “National Survey of Individual Giving” for Pakistan, they find household income (rather than personal income) to be a key determinant of charity. In terms of household income, 32 percent of monetary giving was by high-income households. The lowest income group had the next highest share (27 percent), though with higher sample share.

2.3.1 The Likelihood of Donation decreases with income

The results of our study show that likelihood of donating decreases with income.

The greatest likelihood comes from lower income group with 27% donation percentage.

High income group on the other hand is significantly less likely to donate than low and

middle income groups. Regression results in table 2 also show significantly greater

frequency by the low income group. A similar negative coefficient for income and

frequency of donation was observed by List (2004).

2.3.2 Proportion of income donated Decrease with income.

Income is also an important determinant of charity giving behavior. The exact

relation between income and donation as share of income remains controversial in

literature. Some studies found a U-shaped curve, such that giving as a share of income was

highest among the poor and the very rich (Hodgkinson and Weitzman 1996; James and

Sharpe 2007; Schervish and Havens 1995; Andreoni 2004), while other studies show that

there is a linear downward trend for proportion of income donated (Breeze 2004; Hoge

and Yang 1994; McClelland and Brooks 2004; Wiepking 2004; Wilhelm et al. 2007; Kamas,

Preston, and Baum 2008).

Table 5

Summary statistics of Donations by Income Group of Respondents

Low Income

Group

Middle

Income

Group

High Income

Group

Sample Composition (%) 23.1 38.7 38.2

Likelihood of Donation 0.27 0.20 0.15

Max. of Donation Amount 100 350 1000

Average absolute Donations (in PKR) 11.87 10.07 21.75

Average Donation as Proportion of Income 0.00518 0.0004 0.00043

Positive Average Donations (in PKR) 44.0 49.4 146.0

Fisher's exact test for Likelihood (P-Value )

Low Income Group vs. Middle Income Group 0.18

Low and Middle income vs. High Income Group 0.03**

***: Significant at 1 %, **: Significant at 5 %, *: Significant at 10 % level of significance

The highest absolute average donations were collected from the high income group if

analyzed in absolute terms. The absolute average donation follows somewhat U shaped trend;

middle income group donated lesser amounts in absolute terms. The difference is not significant,

and our findings are consistent with the hypothesis that the low and middle income groups

contribute the same average amount, while the high income group contributes more on the

average. This increase in average donation (in absolute terms) with income is parallel to the finding

of (Bonbright and Azfar 2000), Auten and Rudney (1990), Bekkers (2004) and Rooney, Steinberg,

and Schervish (2001). However, if we adjust the donation amount with the income group,10 our

data supports a negative relation of donation as proportion of income. Donation amounts in the

low income group constitute a larger share as percentage of their income compare to middle and

high income groups.

3. WHY DID GEM FAIL

The most significant finding of our paper is the failure of GEM. Here we discuss some possible

explanations of this.

Literature on incentives has typically attached two possible responses to the incentives in GEM

context; first and obvious is the increase in desired behavior with incentives. The alternate

“crowding out” hypothesis assumes that sometimes, especially in pro-social situations, economic

incentives can suppress intrinsic motivation, and lead to reductions in output (see Titmuss (1970)

for blood donations).11 However, our analysis suggests that the failure of GEM is due to cultural

factors beyond the range of these theories. It seems more likely Pakistani culture, which is based on

Islamic injunctions regarding charity, leads to the differences we observed. Since we did not expect

10 The proxy for average income in these income groups is taken as the middle value of the income range. 11 Nobel Laureates Arrow and Solow both went astray in their analysis of an anomaly pointed out by Titmuss (1970). Titmuss argued that monetary incentives would actually reduce blood donations, since it would undermine the sense of civic duty which leads donors to donate. Both Arrow (1972) and Solow (1971) thought otherwise; they argued that the two incentives would supplement each other. Frey and Oberholzer-Gee (1997) show that crowding-out holds in donation: monetary incentives interfere with the sense of civic duty, as argued by Titmuss (1970).

GEM to fail, our experiment does not provide direct evidence on the factors responsible for the

failure.12

Table 6 presents summary statistics of post mid-survey meeting, after which we collected data on

hospitality and gift refusals. This second half does not differ significantly from the first half of

experiment: i.e. negative gift exchange relationship, especially for large gift. Analysis of the data

shows that taking these phenomena into account does not affect our basic finding of failure of GEM.

Table 6

Summary Statistics of Latter Half of Experiment

No Gift Small Gift Large Gift

Number of household visited 96 129 51

Donations 24 25 7

Relative frequency of Donations 25.0 19.4 13.7

Hospitality 6 9 6

Relative frequency of Hospitality 6.3 7.0 11.8

Gift Refusals 0 9 18

Relative frequency of Gift Refusal 0.0 7.0 35.3

Despite scripted communication between the fundraisers and the respondents, people were

offering tea, soft drinks, and cold water to fundraisers instead of donation, although they could

easily donate an amount costing less or equal to cost of these offers. Hospitality is very prominent

norm in the Pakistani culture, and serving something to people at your doorstep is considered as

act of kindness. The finding of world giving index report (2012) also support the hospitable

behavior where Pakistan is ranked 7th in the world in helping strangers. A plausible hypothesis to

save GEM would be that hospitality replaces donations. However, analysis does not support this

conjecture. Even taking hospitality into account as a donation, we do not get a positive relation

between gift size and donations (see table B in Appendix A). Another evidence that hospitality is not

a substitute for donations is the roughly equal frequency of hospitality among donors and non-

donors (table C Appendix A).

The cultural and social norm of reciprocity is embedded in gift exchange; thus giving a gift

creates an implicit social obligation of returning the gift. One implicit assumption in experiments on

GEM is that gift will be accepted. However, in our experiment people refused to accept gifts. There

are three plausible motives behind these gift refusals;

12

Human interactions have been influenced by complex social structures operated through systems of social norms. For many, religion is an important factor that influences social norms Durkheim, 1976). Despite its important role both in shaping the societies in which we live as well as in affecting our behavior, religion has only recently caught the attention of economists (see overviews in e.g., Hoffman 2011; Iannaccone 1998; Kumar 2008).

Table 7

Hypotheses for Gift Refusals

Motive Explanation

Crowding Out Introduction of gift changes the framing of request for charity to a

market exchange context, and interferes with intrinsic motivations for

charity.

Social pressure Respondents feel social pressure for saying no after accepting gift, so they refuse the gift instead.

Religious Motives One important element of Islamic injunctions for charity is the need to

keep the intention pure; seeking fame, or other worldly benefits or

compensations for charity, ruins the good deed. Accepting a gift can be

viewed as a violation of this condition; in accordance with this, people

start refusing the gifts.

Table 7 presents explanations to three possible hypotheses behind gift refusals. While crowding out

and social pressure are plausible explanations for gift refusal in other cultural contexts, at least

three cases of donation despite the gift refusals do not favor these hypotheses. Della Vigna et al

(2012) show that social pressure as important determinant of door-to-door fund raising. A later

study by Della Vigna et al (2013) shows that females contribute more to charities because they are

more sensitive to social pressure. In our sample gift refusals in female respondent (1.3%) was

significantly lower than male responders (6.4%) providing some support to the idea that females

are more sensitive to social pressure; however the sample size is too small to confirm this. Refusing

a gift is contrary to social norms, and so men are more likely and capable of doing this then women,

because women are more subject to social pressure and hence more likely to accept gift even when

they don’t want to. However, the religious motive provides a better match to many observations.

Firstly, it explains donation with refusals, since Islamic injunctions are to seek neither fame nor any

other worldly return for charity. Also in accordance with this, we find that people seek to remain

anonymous by not taking the receipts. Many of them explicitly mentioned religious motives. If

religious motives are dominant in charity, than the observed higher gift refusal rate by males

requires explanation, because Seguino and Lovinsky (2009) find that males and females are equally

religious in Pakistan. A likely possibility is that women refused less frequently due to social

pressure, as women experience more social pressure and they also react more to social pressure

they face (Croson and Gneezy 2009). 13

A potential explanation for the highest number of donations in low income group also comes from

religiously rich cultural background. 14 Similarly, non-experimental observation also suggests that

13 While most of the literature says that women are more religious, the world value survey shows almost equal score scores of importance of religion in Pakistan for both males and females(Seguino & Lovinsky, 2009). 14 In US where more than 82 percent of population attends church (Wiepking, 2007), lower income group is more affiliated to religion so they donate more to religious donation than high income group (Schervish and Havens 1995). In a study for Canada, Berger (2006) found Protestants are more generous, largely because of stronger social norms and greater church attendance. Among the Arabs there was clear tendency of more donations for exceedingly religious donors than the rest (Shai et al. 1999). Wiepking (2007) in her study for

religiosity is a major driver of behaviors in Pakistan, and must be explored more deeply in

subsequent work along these lines. The negative relationship with income and charity has also been

associated to the religious motivations (James and Sharpe, 2007; Brooks, 2005; Hoge and Yang,

1994). While there are studies showing religion as an important determinant for giving donation,

role of religious beliefs is rarely explored. Davidson and Pyle (1994) find that more orthodox and

stronger religious beliefs are positively linked to religious contributions. There is no such study for

Pakistan; however, Najam (2007) documents the strong role Islam plays in generating large

charitable donations from the Pakistani community in the USA. Also 98 percent of the donors in the

national survey of individual giving cited religious motivation for making donations (Bonbright &

Azfar, 2000).(Brooks, 2005)

4. CONCLUDING REMARKS

Peripheral to our main investigation of the GEM, we found that gender and income have a

significant effect on donations. Females were extraordinarily generous relative to male

counterparts in general and to female solicitors in particular. We also found that increases in

income lead to decrease in frequency of donations and also a decrease in proportion of income

donated; this is also consistent with other research on charity in Pakistan (Bonbright and Azfar

2000).

Our main finding is that the simple GEM does not function as expected in our field

experiment. The frequency of the donation decreases with the gifts. However, once respondent

decides to reciprocate through donation the amount of donation varies positively with the gift

exchange hypothesis. Two additional behavioral observations of hospitality and gift refusals

complicate our results. Our experimental design does not allow us to provide a clear explanation for

these behaviors. Evidence from other studies of charity suggests that the social and religiously rich

cultural norms may be responsible for the failure of GEM.

Gift refusals can also be explained within a religious context; data show some support for

religious concerns of avoiding wordly returns in exchange for charity. Gift refusals may also stem

from an unwillingness to incur a social obligation; i.e. the social pressure of saying no after

accepting gift. A religious motive for gift refusals would have substantially different implications,

and future studies may attempt to differentiate the two. However, a host of un-anticipated

complications created by use of door-to-door campaign can be reduced by employing more

impersonal and anonymous methods for soliciting donations. Greater anonymity might lead to

closer replications of conventional results. At the same time, this suggests that conventional

findings favoring simple gift exchange in the lab or by postal methods may not generalize to

situations involving less anonymity and more social interactions, even in other cultural contexts.

Dutch economy which is more secular (only 20 percent of church attendance) failed to find any effect of income on probability of giving; he also argued that religious affiliation of low income group can explain the giving behavior.

REFERENCES

Andreoni, J. (2004). Economics of philanthropy. In L. Gerard-Varet, S. Kolm, & J. M. Ythier, Handbook

of giving, reciprocity and altruism (pp. 11369-11376). North-Holland: Elsevier.

Andreoni, J., & Petrie, R. (2008). Beauty, gender and stereotypes: Evidence from laboratory

experiments. Journal of Economic Psychology , 29 (1), 73-93.

Andreoni, J., & Vesterlund, L. (2001). Which is the Fair Sex? Gender Differences in Altruism.

Quarterly Journal of Economics , 116 (1), 293-312.

Arrow, K. J. (1972). Gifts and Exchanges. Philosophy & Public Affairs , 1 (4), 343-362.

Auten, G. E., & Rudney, G. (1990). The Variability of Individual Charitable Giving in the US. Voluntas

(1), 80-97.

Bekkers, R. (2010). George Gives to Geology Jane: The Name Letter Effect and Other Similarities in

Fundraising. International Journal of Nonprofit and Voluntary Sector Marketing , 15 (2), 172–180.

Bekkers, R. (2004). Giving and Volunteering in the Netherlands: Sociological and Psychological

Perspectives. Ph.D. dissertation, Department of Sociology, Utrecht University, Utrecht, the

Netherlands.

Bekkers, R. (2007). Measuring Altruistic Behavior in Surveys: The All-or-Nothing Dictator Game.

Survey Research Methods 1 .

Ben-Ner, A., Kong, F., & Putterman, L. (2003). Share and share alike? Genderpairing, personality,

and cognitive ability as determinants of giving. Journal of Economic Psychology , 25, 581-589.

Berger, I. E. (2006). The Influence of Religion on Philanthropy in Canada. Voluntas , 17, 115–32.

Bolton, G. E., & Katok, E. (1995). An experimental test for gender differences in beneficent behavior.

Economic Letters , 48, 287-292.

Bonbright, D., & Azfar, A. (2000). Philanthropy in Pakistan:A Report of The Initiative on Indigenous

Philanthropy. The Aga Khan Foundation.

Breeze, B. (2004). Widow's Mite or Widow's Might? The Relative Giving of Rich and Poor in the UK.

The 33rd annual conference of the Association for Research on Nonprofit Associations and Voluntary

Action. . Los Angeles, USA.

Brooks, A. (2005). Does Social Capital Make You Generous? Social Science Quarterly , 86, 1-15.

Croson, R., & Gneezy, U. (2009). Gender differences in preferences. Journal of Economic Literature ,

47 (2), 1-27.

Davidson, J., & Pyle, R. (1994). Passing the Plate in Affluent Churches: Why Some Members Give

More Than Others. Review of Religious Research , 36, 181-96.

DellaVigna, S., List, J., & Malmendier, U. (2012). Testing for Altruism and Social Pressure in

Charitable Giving. Quarterly Journal of Economics , 127, 1-56.

DellaVigna, S., List, J., Malmendier, U., & Rao, G. (2013). The Importance of Being Marginal: Gender

Differences in Generosity. American Economic Review , 103 (3), 586-90.

Durkheim, É. (1912 ; 1976). The Elementary Forms of Religious Life. (J. W. Swaim, Trans.) London:

Allen & Unwin.

Einolf, C. J. (2011). Gender Differences in the Correlates of Volunteering and Charitable Giving.

Nonprofit and Voluntary Sector Quarterly , 40, 1092-1112.

Falk, A. (2007). Gift Exchange in the Field. Econometrica , 75 (5), 1501-1511.

Freedman, D., Pisani, R., & Purves, R. (1998). Statistics (3rd ed.). New York: W. W. Norton and Co.

Frey, B., & Oberholzer-Gee, F. (1997). The Cost of Price Incentives: An Empirical Examination of

Motivation Crowding-Out. American Economic Review , 87 (4), 746-755.

Hodgkinson, V. A., & Weitzman, M. (1996). Giving and Volunteering in the United States.

Washington: Independent Sector.

Hoffman, R. (2011). The experimental economics of religion. University of Nottingham, mimeo.

Hoge, D. R., & Yang, F. (1994). Determinants of Religious Giving in American Denominations: Data

from Two Nationwide Surveys. Review of Religious Research , 36, 123-48.

Iannaccone, L. R. (1998). Introduction to the Economics of Religion. Journal of Economic Literature ,

36, 1465-1496.

James, R. N., & Sharpe, D. L. (2007). The Nature and Causes of the U-Shaped Charitable Giving

Profile. Nonprofit and Voluntary Sector Quarterly , 36, 218-38.

Kamas, L., Preston, A., & Baum, S. (2008). Altruism in Individual and Joint-Giving Decisions: What’s

Gender Got to Do With It? Feminist Economics , 14, 23-50.

Kumar, V. (2008). A Critical Review of Economic Analyses of Religion. IGIDR Working Paper Series

WP-2008-023 .

Landry, C., Lange, A., List, J., Price, M., & Rupp, N. (2006). Toward an Understanding of the

Economics of Charity: Evidence from a Field Experiment. Quarterly Journal of Economics , 121, 747-

782.

Leslie, L. M., Snyder, M., & Glomb, T. M. (Forthcoming). Who Gives? Multilevel Effects of Gender and

Ethnicity on Workplace Charitable Giving. Journal of Applied Psychology.

List, J. A. (2004). Young, selfish, and male: Field evidence of social preferences. Economic Journal ,

114, 121-49.

Lo, C. P., & Tashiro, S. (2012). Are women more generous than men? Evidence from the US

Consumer Expenditure Survey. Journal of Gender Studies , 1-15.

Marx, J. D. (2000). Women and Human Services Giving. Social Work , 45, 27-38.

McClelland, R., & Brooks, A. C. (2004). What is the Real Relationship between Income and Charitable

Giving? Public Finance Review , 32 (5), 483-97.

Meier, S. (2005). Conditions under Which Women Behave Less/More Pro-Socially than Men

Evidence from Two Field Experiments. Public Finance Review , 35 (2), 215-232.

Mesch, D. J., Brown, M. S., Moore, Z. I., & Hayat, A. D. (2011). Gender Differences in Charitable Giving.

International Journal of Nonprofit and Voluntary Sector Marketing , 16, 342-355.

Mesch, D. J., Rooney, P. M., Steinberg, K. S., & Denton, B. (2006). The Effects of Race, Gender, and

Marital Status on Giving and Volunteering in Indiana. Nonprofit and Voluntary Sector Quarterly , 35,

565-587.

Naeem, S., & Zaman, A. (forthcoming). Gender and Ultimatum in Pakistan: Revisited. Pakistan

Development Review .

Najam, A. (2007). Portrait of A Giving Community: Philanthropy by the Pakistani-American Diaspora .

Harvard: Harvard University Press.

Piper, G., & Schnepf, S. V. (2008). Gender Differences in Charitable Giving in Great Britain.

International Journal of Voluntary and Nonprofit Organizations , 19 (2), 103-124.

Razzaque, S. (2009). The ultimatum game and gender effect: experimental evidence from Pakistan.

Pakistan Development Review , 48 (1), 23-46.

Regnerus, M. D., Smith, C., & Sikkink, D. (1998). Who Gives to the Poor? The Role of Religious

Tradition and Political Location on the Personal Generosity of Americans Toward the Poor. Journal

for the Scientific Study of Religion , 37, 481–93.

Rooney, M. P., Steinberg, K., & Schervish, P. G. (2001). A Methodological Comparison of Giving

Surveys: Indiana As a Test Case. Nonprofit and Voluntary Sector Quarterly , 30, 551-68.

Schervish, P. G., & Havens, J. J. (1995). Explaining the curve in the U-shaped curve. Voluntas , 6, 202-

25.

Seguino, S., & Lovinsky, J. (2009). The Impact of Religiosity on Gender Attitudes and Outcomes.

available at: http://www.uvm.edu/~sseguino/pdf/Religiosity.pdf .

Shai, S., Lazer, A., Ducin, R., & Gidron, B. (1999). Philanthropy in Israel – Pattern of Giving and

Volunteering of the Israeli Public (Heb.). Beer-Sheva: Ben-Gurion University of the Negev, Israeli

Center for Third-Sector Research . (A. 2. Rudich, Compiler)

Simpson, E. H. (1951). The Interpretation of Interaction in Contingency Tables. Journal of the Royal

Statistical Society, Series B , 13, 238–241.

Solow, R. M. (1971). The State of Economics: The Behavioral and Social Sciences Survey: Discussion.

American Economic Review , 61 (2), 63-65.

Titmuss, R. M. (1970). The Gift Relationship. Allen and Unwin.

Wiepking, P. (2004). Do the Poor Donate More? The Effect of Income on Philanthropic Donations.

The 33rd annual Arnova conference. Los Angeles, CA, 2004.

Wiepking, P. (2007). The Philanthropic Poor: In Search of Explanations for the Relative Generosity

of Lower Income Households. Voluntas , 18, 339–358.

Wilhelm, M. O., Rooney, P. M., & Tempel, E. R. (2007). Changes in Religious Giving Reflect Changes in

Involvement: Age and Cohort Effects in Religious Giving, Secular Giving, and Attendance. Journal for

the Scientific Study of Religion , 46, 217-32.

Willer, R., Wimer, C., & Owens, L. (2012). Explaining the Gender Gap in Charitable Giving: Lower

Empathy Leads Men to Give Less to Poverty Relief. The Stanford Center on Poverty and Inequality

working paper: available at: http://www.stanford.edu/group/scspi/_media/working_papers/x-

Gender_Gap_Poverty_Relief.pdf .

Wit, A. d., & Bekkers, R. (2012). Explaining Gender Differences in Charitable Giving: the Dutch Case.

available at: https://renebekkers.files.wordpress.com/2011/10/dewit_bekkers_arnova2012_rb2.pdf .

APPENDIX A: SUPPLEMENTARY TABLES

Table A

Probit Regression after excluding female encounters to FF Team

Variable Coefficient Std. Error Prob.

Constant 0.30 0.05 0.000***

Small Gift 0.01 0.04 0.792

Large Gift -0.08 0.05 0.086*

Male -0.07 0.04 0.065*

High Income -0.01 0.04 0.868

Low Income 0.13 0.04 0.005***

MM -0.10 0.04 0.010**

R-Square 4.10%

***: Significant at 1 %, **: Significant at 5 %, *: Significant at 10 % level of significance

Table B

Probit regression after treating Hospitality as GEM

Variable Coefficient Std. Error Prob.

Constant 0.35 0.11 0.002***

Small Gift 0.04 0.07 0.560

Large Gift -0.01 0.08 0.872

Male -0.02 0.06 0.731 High Income -0.13 0.06 0.034** Low Income 0.04 0.07 0.588

MM -0.11 0.09 0.216

FF 0.13 0.11 0.229

R-Square 6.70%

***: Significant at 1 %, **: Significant at 5 %, *: Significant at 10 % level of significance

Table C

Relative Frequencies of Hospitality and Gift Refusals

Hospitality Gift Refusals

Male 8.2 12.1

Female 5.8 2.9

Low Income 5.0 8.3

Middle Income 11.2 12.9

High Income 5.0 7.0

Donated 8.9 5.4

Did not donate 7.3 10.9

MM 7.3 11.0

FF 9.1 5.5

MF 6.7 10.0

APPENDIX B: EXPERIMENTAL DESIGN

This experiment used door to door fund raising campaign for an organization founded by a

small group of doctors in a famous hospital of Islamabad, Pakistan. This organization is working for

poor and needy patients who are not able to pay for their medication and other major medical

expenses in given hospital. Apart from control group, we used gifts, used in Falk (2007), i.e., one

post card as small gift and a set of four post cards as large gift (Sample card in Attached below).

However unlike Falk (2007), our experiment was performed under different treatment setup i.e.,

door to door fundraising instead of a postal campaign.

In the city of Islamabad, housing tends to be homogenous by incomes, which make it easy to

stratify the area according to income group. We have stratified area being surveyed into three

income groups i.e. low, middle and high income groups15. The houses were then picked randomly

from the selected areas within each stratum16. Treatments were then allocated randomly in equal

proportion within each income group. The respondents in control group were given a brochure and

a call for donation, a post card was added into the set for small gift treatment group. Respondents in

large gift treatment got a set of four cards along with a brochure and donation call. After giving the

brochures /gift, fundraisers made donation call, they also politely asked about the income group

out of three given income ranges and noted the gender of the respondents. Solicitors were also

asked to note down any unusual observations which they felt were important (Data Recording

Sheet attached below).

A group of 20 students was hired from different universities as fundraisers and were given

a short training session for fundraising. During training they were provided with a small

introductory speech about organization and cause of the fundraising (Appendix C), for further

inquiry they were asked to guide them to call the helpline numbers mentioned on the brochure.

They were asked to stay away from any informal discussion to the maximum possible. Fundraisers

were divided into pairs randomly, and were asked to wear normal clothes consisting of dress pants

and shirt for males and traditional dress i.e. shalwar, kameez and duppatta for female fundraisers;

any fancy clothing was strictly prohibited. They were displaying their identity cards issued by

charity organization for authenticity; only one out of all visited household called hospital helpline to

15 We provided three income brackets to choose from, instead of asking their exact income. Further due to expected homogeneity within strata, we replaced few cases of missing income group of respondent by income group of three surrounding households. 16 Repeated every 20th house after selecting one random start.

confirm the authenticity of fundraising campaign. Fundraisers were offered flat wages (PKR 150

per hour; which is slightly above the market wage for entrants to public sector) for a given numbers

of respondents per day in randomly assigned treatment group. Each solicitor pair was randomly

assigned new treatment selected randomly at the start of each day. Further, we did not inform the

fundraisers about the experimental aspect of fundraising campaign.

Fundraisers visited around 800 households but were able to collect the data from 545

households. The rest were either not at home or refused to give time without listening to

fundraiser. As these respondents were not exposed to any treatment before they refused to

respond, we cannot attribute this non-response to treatment so these observations were discarded

from the study. To increase the chances of equal encounter with both sexes, solicitation was

planned after office hours.

SAMPLE POST CARD

DATA RECORDING SHEET

APPENDIX C: TRAINING FOR FIELD EXPERIMENT

The FS a nonprofit private organization FS (Name of organization here) initiated by a group

of God fearing and committed doctors and volunteers from ABC Hospital. They are committed to

help the poor and needy patients to assist them in the cost of treatment, essential diagnostic tests,

consultation, nursing care, and medicine either free or at subsidized rates. We not only

provide free consultation to needy patients, but also raise money for them by holding various

social events like annual fund raising dinner, Eid cards sales, food festivals, art exhibitions and

door to door fund raising schemes.

As a part of our effort, you are hired you to help us in raising funds for our noble cause. You have to

work in door to door fund raising campaign for three hours in afternoon 5 to 8 pm. The donations

you collect is signal of trust from people that the amount that they are paying will go to this noble

cause, to entrust the responsibility that has given to us, please take care to enter all the relevant

information on data recording sheets without any mistake and issue the receipts to the donating

respondents.

Task Details

All of you are provided with an introduction and donation call, we would appreciate if you stick to

these words. Each of you will be paired with different team member every day. At the start of

each day, you will receive a folder containing data recording sheets, fund raising location for that

day, a house number and fundraising protocol material. We have decided to use three different

protocols for raising fund.

Protocol 1: A simple donation call according to given instructions

Protocol 2: A post card will be added to donation appeal. Please present post card before

making the donation call.

Protocol 3: A set of four post cards will be added to donation appeal. Please present

post card before making the donation call.

Please present post card before making the donation appeal in protocol 2 and 3.

You will be given a house number to start with; you have to select every 10th house for

donation campaign. You need to enter the house number, gender of the respondent, his or her

monthly income from given income brackets. Please don’t ask about their income directly, rather

ask to choose the income bracket that apply very politely.

You must enter the receipt number in case of any positive donation. If someone refuses to donate

without listening to the donation call please note it in the remarks section. If you feel to

share any interesting information, please record it in remarks field.

At the end of three hours, project coordinator will call you to collect the folders with properly

recorded data sheets and other material at the assigned place nearby.

Please remember

You are requested to stay away from any informal discussion.

Properly display the identity cards issued to you. You are advised to follow dress code (shalwar,

Kameez and dupatta for females and dress pant and shirt for boys) during campaign; any

fancy clothing is strictly prohibited.

If anyone needs to authenticate the validity of fundraising campaign, you should provide

them the official number.

If you have any questions or if any unexpected situation arises please immediately call the

project coordinator for help.

We will conduct a meeting after three working days to share your experiences and

comments.

Payment

You will be paid after delivering your data along with donation amounts with receipts at the end of

the day. You will receive a fixed wage of Rs. 150 per hour for three hours without considering the

donation you raise. The project time must not exceed three hours.

Introductory Speech (original speech was in Urdu)

My Name is Saima (Name of speaker), We are from a nonprofit private organization FS

(Name of organization) which consists of a group of God fearing and committed doctors and

volunteers. These are dedicated people who help poor patients at ABC Hospital. This money is used

to assist patients with the cost of treatment, essential diagnostic tests, consultation, nursing

care, and medicine either free or at subsidized rates. We raise money by holding various

social events like annual fund raising dinner, Eid cards sales, food festivals, art exhibitions and

similar door to door fund raising schemes.

FS is dedicated to make a difference in the lives of deserving and needy patients at ABC Hospital.

This is small gift from FS for you so that you can give it to your loved ones or enjoy it yourself .17 We

ask you to join us and help us in our noble cause. Your donations, in any amount either small or

large, are valuable to our organization.

17 For control group the underlined sentence was not included in the speech.