Introduction

Why does charitable giving by individuals vary so widely among countries? Scholars have proposed many answers to this question, but most of these answers derive their theoretical logic from Western European culture and history, and test their theories on datasets made up primarily or exclusively of Western European countries. Whether these theories will explain variation in charitable giving in the majority of the world’s countries, most of which are less economically developed and do not have a Western European culture, has not yet been tested.

This paper describes existing theories of cross-national variation in charitable giving and derives new theories from the broader literature on cross-national variation in civil society, volunteering, and voluntary association membership. It tests these theories on a dataset with country-level data from 114 countries, comparing the results for Western nations, middle income non-Western nations, and low-income non-Western nations. It finds that the same factors that predict cross-national differences in giving in Western countries also predict giving in the wealthier non-Western countries, but not in the poorer ones.

This paper contributes to the literature in several ways. First, it brings together a number of existing theories about the predictors of cross-national variation in charitable giving and tests them on a single dataset. Second, it derives new theories about charitable giving from existing theories of about the size and strength of civil society and the non-profit sector. Third, it is one of the first papers ever to extend the analysis of cross-national variation in charitable giving to non-Western and less developed countries. Finally, it shows that economic development, not cultural or historic differences, are what separate the charitable giving patterns of Western countries from those of the developing world.

Review of the Literature

The non-profit literature contains a few theories that explain cross-national variation in charitable giving. It contains more theories that explain cross-national variation in volunteering, voluntary association participation, civil society, and the size of the non-profit sector, and this paper adapts these theories to predict variation in charitable giving. In doing so, it assumes that the factors that affect the size of the non-profit sector and the extent of volunteering would tend to affect charitable giving in the same way. Given that people make charitable donations to non-profit organizations, factors that increase the size of the non-profit sector would likely also increase participation in charitable giving. I assume that factors that increase volunteering would probably also increase charitable giving because studies have found that giving and volunteering are complements. Individuals who volunteer their time also tend to donate money, and the individual-level predictors of volunteering also predict charitable giving (Hill 2012). There appear to be no articles on whether the correlation between giving and volunteering occurs also at the country level of analysis, but the correlations at the individual level provide an initial reason to hypothesize that the variables that predict volunteering in countries may also predict charitable giving.

Theories of cross-national variation in giving, volunteering, and the size of the non-profit sector involve differences in countries’ economies, political systems, and culture, religion, and values. This section discusses each of these types of theories in turn and derives hypotheses from each subject area.

Economic Theories

As charitable giving is an economic activity, economic factors may play a large role in determining cross-national differences in donations. The overall strength of a country’s economy would likely correlate with levels of charitable giving, but the direction of that correlation is not obvious. Economic development encourages the growth of the middle class, who have the resources and skills to engage in civic activities (Bailer et al. 2012). Economic development also encourages occupational specialization and social status distinctions, which encourage the development of voluntary interest groups. As people become more civically active and join voluntary associations, they may donate money to these same associations. Finally, economic development may lead to charitable giving because wealthier citizens simply have more money to give away.

On the other hand, Salamon and Anheier (1998) hypothesized that wealthier countries would have a more developed welfare state and would therefore have smaller non-profit sectors, as the government would already be meeting the need for social services. By this logic, citizens of wealthy countries would be less likely to give money to charity, as they would perceive that the government’s welfare spending had made their own donations unnecessary.

A third possibility would be for charitable giving to have a curvilinear relationship with wealth. Charitable giving may increase as a country’s economy allows for the growth of a middle class who have enough money to give some away, but then may decrease as the welfare state grows so strong that most people’s basic needs are met.

From this theoretical background, it seems possible that economic may relate to charitable giving in a positive, negative, or curvilinear (positive to a certain threshold, then negative) fashion. However, the theoretical arguments seem strongest for a positive relationship, and any possible curvilinear relationship is merely speculative. Therefore, I hypothesize a positive correlation between economic development and charitable donations, but test for a possible negative or curvilinear relationship.

H1

Gross domestic product (GDP) per capita correlates positively with charitable donations.

In addition to overall economic activity, government welfare spending would likely correlate with individual charitable giving. The relationship could be negative, if government welfare spending covers peoples’ basic needs and makes individual charitable giving less necessary (Salamon and Anheier 1998). On the other hand, the relationship between welfare spending and charitable giving could be positive. Government funding for non-profits may provide a steady source of support that allows non-profits to devote resources to fundraising, and government funding may signal to donors that specific non-profits and the sector in general are trustworthy (Nguyen 2015).

Prior studies have found little support for the hypothesis that government funding “crowds out” private donations to non-profits or otherwise interferes with social capital or participation in civil society (Kääriäinen and Lehtonen 2006; Van Oorschot and Arts 2005; Van Oorschot and Finsveen 2010). Instead of cutting back on donations, citizens of countries with strong welfare states may transfer their giving from nonprofits that provide basic services to “expressive” non-profits that work in the arts, culture, recreation, the environment, advocacy, and international causes, resulting in still high levels of charitable giving (Einolf 2015; Nguyen 2015; Salamon and Anheier 1998; Sokolowski 2013). As most research supports a positive correlation between government welfare spending and charitable giving, this paper predicts the same:

H2

Government welfare spending correlates positively with charitable giving.

Political Theories

Several theories describe how political differences among countries may affect charitable giving. Democracy and political freedom should correlate with charitable giving, as governments in politically free countries will tend to place fewer restrictions on non-profits and their fundraising practices than governments in repressive countries (Bailer et al. 2012; Curtis et al. 2001; Hadenius and Uggla 1996; Parboteeah et al. 2004). Fewer restrictions lead to more and stronger non-profits that can solicit funds with fewer limitations or transaction costs placed by government regulation. Similarly, the effectiveness of government and the absence of corruption correlate positively with the development of civil society (Bailer et al. 2012; Hadenius and Uggla 1996), and should correlate positively with charitable giving as well. Where government is effective, it will be easier for non-profits to attain legal status; where government is non-corrupt, non-profits will be able to form and solicit donations without transaction costs in the form of government bribes and harassment.

Formerly, communist countries (Archimbault 2009; Howard 2003; Wiepking and Handy 2015) would be expected to have a smaller non-profit sector because communist governments outlawed or severely restricted the non-profit sector. After the fall of the Soviet Union, formerly communist countries had to create a non-profit sector from almost nothing, or reinvigorate institutions that had been dormant for most of the twentieth century.

Finally, the timing of state formation is important in the development of civil society and the non-profit sector (Ragin 1998), and therefore may correlate with charitable giving. In general, the more recently a state formed, the lower one would expect charitable giving to be, as a young state would not have the time to develop the legal and political structure that would support a strong non-profit sector.

H3

Political freedom, as measured in terms of the rights of freedom of association and expression, correlates positively with charitable giving.

H4

Strength of government, as measured by (a) effectiveness and (b) lack of corruption, correlates positively with charitable giving.

H5

Formerly communist countries have lower rates of charitable giving.

H6

Countries with earlier dates of state formation will have higher rates of charitable giving.

Cultural Theories

Cultural differences may also lead to different giving habits in different countries. Cultures can differ in values and ethnolinguistic diversity, but the key cultural difference related to charitable giving is that of religion. Charitable behavior is historically rooted in religious prescriptions to help others and helping activities have traditionally organized through religious institutions. Thus, levels of religiosity, religious diversity, and the character of the dominant religion may all affect charitable giving.

Religious individuals tend to give more money to charity (Bekkers and Wiepking 2011; Wiepking et al. 2014), so one would expect countries with highly religious populations to have higher rates of charitable giving (Grönlund and Pessi 2015). Most religions stress the value of helping others, and religious people internalize these values, making them more generous (Einolf 2011). Second, religious people belong to social networks in which people are more likely to ask them to give to charity, and the external norms associated with those networks make it difficult to refuse these requests (Bekkers and Wiepking 2011; Ruiter and De Graaf 2006). People living in highly religious countries tend to volunteer more as well. National religiosity influences individual volunteering independent of individual religiosity, suggesting that the national religious culture in devout countries affects the volunteering of all (Lim and MacGregor 2012). National religious culture may have a similar positive influence on charitable giving.

The level of religious diversity in a country may also correlate with charitable giving. Wiesbrod’s (1977) government failure theory states that the non-profit sector arises in order to provide services that the government fails to provide. In religiously diverse countries, religious diversity causes the population to have different preferences that the government has difficulty providing for, encouraging a stronger non-profit sector in order to meet those needs (Salamon and Anheier 1998). A demand-side theory (James 1987) states that non-profits arise due to the actions of humanitarian entrepreneurs. Where religious diversity creates much religious competition, these entrepreneurs are more likely to step forward. Finally, in states with one dominant religion, the government and the religious authorities tend to ally, producing a corporatist regime with a small number of big charities getting government support (Salamon and Anheier 1998). The security of government funding and the lack of competition may lead charities to put less effort into soliciting individual gifts, reducing charitable giving.

A long tradition in sociology dating back to De Tocqueville (2004 [1840]) holds that Protestant countries have higher rates of civic participation, and this implies that Protestant countries may also have higher rates of charitable giving. While some European countries have a state-sponsored Protestant church, Protestant denominations tend to be less closely allied with the state than the Catholic Church and many have had to depend on individual contributions, not state sponsorship, for survival. Protestant churches are also less hierarchical than Catholic churches and encourage more participation in worship and governance, which may also encourage more participation through charitable giving (Curtis et al. 2001).

Of course, the encouragement of charity is not limited to the Christian religion. Hinduism (Bhat et al. 2010; Juergensmeyer and McMahon 1998), Buddhism (Denoon 2010; Guruge and Bond 1998; Kawamura 1998), Judaism (Kabalo 2010; Penslar 1998), and Islam (Hasan 2010; Kochuyt 2009; Koslowski 1998), and all consider helping others to be of central moral importance, and these religions have institutionalized giving to others within the structure of their societies.

In Hinduism, actions of generosity were valued both as goods in themselves and as ways to accumulate merit for a better rebirth. In ancient times gift-giving, or “dana,” meant giving ritual gifts for sacrifice and gifts to the priestly class, and service, or “seva,” meant giving service to the temple. In later times, however, these terms expanded to include giving and volunteering to help the poor. Mohandas Gandhi reinterpreted these terms to apply to political activism and social change, and his ideas continue to influence philanthropy and volunteering in contemporary India (Bhat et al. 2010; Juergensmeyer and McMahon 1998).

Buddhism also values generosity as a good in itself, a means to a higher rebirth, and a foundation for the moral values and mental clarity needed to attain nirvana. Buddhism places “dana,” or giving, as the foremost of the ten virtues that any individual must perfect to attain enlightenment. Buddhist monks and nuns depend on charitable donations for their survival, and Buddhist laity feel a duty to contribute money and food to support monasteries and temples (Guruge and Bond 1998; Kawamura 1998). While Buddhists have historically valued compassion highly, Buddhist societies have only recently encouraged the development of charitable organizations. Denoon (2010, p. 1159) explains this as due to “the Buddhist emphasis on the purity of thought over action and the stress on generalized ‘compassion towards others’ in all aspects of life, rather than on charitable giving per se.”

Islam holds “zakat,” or giving alms to the poor, to be one of its five pillars, equal in importance to daily prayers and the pilgrimage to Mecca. Islam also encourages voluntary, spontaneous charity, or “sadaqa.” Islamic societies institutionalized the “waqf,” or charitable and religious foundation, in medieval times, and some of these institutions continue to operate today. Historically, waqfs were operated by the government, and the rise of modern secular states has made the relationship between government and waqfs difficult. Private foundations have taken over some of the functions of waqfs but have not become as prevalent in Muslim societies as they are in Christian ones (Hasan 2010; Kochuyt 2009; Koslowski 1998).

Judaism has valued philanthropy since ancient times, dating back to biblical injunctions to farmers to leave food unharvested in the fields for the benefit of the indigent. In the modern period, Jews were excluded from much of the philanthropy and government assistance in the predominantly Christian and Muslim societies in which they lived, so Jews developed their own systems of philanthropy and mutual support to take care of the poor, sick, and elderly in their communities. The close urban life of many Jewish communities also made it difficult for wealthy Jews to ignore the needs of the poor. Beginning in the nineteenth century and continuing to this day, support of migration to Palestine and later support of the state of Israel encouraged even further philanthropy. Thus, the Jewish religion contains injunctions to be generous and give to the poor, but the social structure of Jewish life throughout the modern era encouraged Jews to be particularly generous (Kabalo 2010; Penslar 1998).

Little research is available on the effect of Hinduism, Buddhism, Islam, or Judaism on charitable giving at the country level, but the emphasis each of these religions places on helping others leads me to expect that the proportion of a country’s population that identifies with each of these religions will have a positive relationship with charitable giving.

H7

Religiosity correlates positively with charitable giving.

H8

Religious diversity correlates positively with charitable giving.

H9

The percentage of Protestants in a country’s population will correlate positively with charitable giving, and this correlation will be stronger than that of percent Catholic.

H10

The percentage of Hindus, Muslims, Buddhists, and Jews in a country’s population will correlate positively with charitable giving.

Ethnic and linguistic diversity may also affect charitable giving. In their study of “citizenship behaviors,” which include political participation, voluntary association participation, and trust, Anderson and Paskeviciute (2006) acknowledge that most scholars assume ethnic and linguistic diversity are harmful for positive citizenship. Citizens living in ethnically and linguistically diverse societies are more likely to come into conflict over resources, and diversity can discourage the feeling of unity that leads to cooperative behavior. However, ethnic and linguistic diversity can lead to more political and voluntary participation as diverse groups mobilize their members in the competition for resources and political power. Diversity may also lead to higher charitable giving because governments may be unsuccessful in meeting the needs of a diverse population, and may therefore relegate service provision to non-profits (Salamon and Anheier 1998). In a study of forty-four countries, Anderson and Paskeviciute (2006) found that linguistic diversity correlated with more voluntary association membership, particularly within less democratic countries. The authors did not find positive effects of ethnic diversity in their sample, but there may be such effects in the current sample. As membership in voluntary associations tends to correlate with charitable giving, the relationship between diversity and giving is predicted to be positive.

H11

Ethnically diverse countries have a higher rate of charitable giving.

H12

Linguistically diverse countries have a higher rate of charitable giving.

In addition to these twelve hypotheses, this paper tries to answer a broader research question of whether the theories that explain charitable giving in Western countries also explain giving in non-Western ones. Accordingly, each hypothesis will be tested four times: first on the full dataset, then on the subsample of Western and non-Western countries, and then on the non-Western countries divided further into a set of middle income non-Western countries and a set of low-income non-Western countries.

If the theories about cross-national variation in charitable giving apply universally, the hypotheses should be supported equally in all four datasets. If the theories do not apply across cultures, the hypotheses should receive stronger support in the Western sub-sample than the non-Western one. If economic development is all that separates non-Western from Western countries, then the theories should predict as well in middle income non-Western countries as they do in Western countries. However, if religion, culture, and politics are what separates Western and non-Western countries, then the theories should not work to explain giving in the middle income non-Western countries.

H13

The first twelve hypotheses will be better supported in Western than non-Western countries.

H14

The first twelve hypotheses will be better supported in middle income non-Western countries than in low-income non-Western countries.

Method

Data

The hypotheses are tested using charitable giving data from the 2007 and 2008 waves of the Gallup World Poll (GWP), and country-level variables taken from a number of sources. The Gallup organization gathered individual-level data via phone or face-to-face interviews in the main language of each country, based on nationally representative probability samples. The data capture self-reported demographics, attitudes, and behaviors of respondents, including formal and informal volunteering (helping strangers). The GWP is a useful data source as it not only contains an unusually large sample of 174,590 respondents from over 100 countries, but it also contains a standardized set of questions across all countries. For this paper, only country-level average variables are used, as the Gallup organization charges a high fee for access to the individual-level data. There are valid giving data for 114 countries (21 Western and 93 non-Western), but missing data for some countries in the independent variables places the valid N for correlations and regression analyses between 104 and 112.

I measure the independent variables using measures taken from the World Database of Happiness (Veenhoven 2015), which are adapted from other sources. These sources include the United Nations reports, human rights reports generated by Freedom House, corruption and governance data taken from the World Governance Indicators project (Kaufmann et al. 2015), and data on religious and ethnic diversity (Alesina et al. 2003). These variables were measured in the years 2001–2007, meaning that some are not measured simultaneously with the GWP data. However, all are within 6 years, and all measure aspects of a country’s population which change slowly, such as economy, religion, religion, and ethnicity. Thus, the fact that the independent variables were measured in different years should not significantly bias the results.

Variables

The dependent variable in this study is a single question on the GWP, “During the last month, did you give any money to charity?” The country-level measure is the percentage of respondents who answered yes to the question, which varies from 5 to 83 % with a mean value of 31 % and a standard deviation of 19 % (Table 1).

Table 1 Descriptive statistics
Full size table

There are twelve independent variables in this study. I test the role of economic development (H1) using a measure of real gross domestic product (GDP) per capita. Government welfare spending (H2) is measured through government spending on public health as a percentage of GDP. Both economic measures are for the year 2005 and are taken from the United Nations Human Development Report.

I measure freedom of association and expression (H3) through a summed index of the 2004 Freedom House measures of political rights and civil liberties. Political rights include free, fair, and contested elections, the rights of minorities, and the decentralization of political power. Civil liberties include freedom of speech, assembly, the press, religion, and the freedom from political terror. Freedom House uses expert evaluators to generate a score between 2 and 14, and these measures were recoded for this paper so that higher numbers represent a higher level of freedom.

Government Effectiveness

(H4a) is measured through subjective assessments of the quality of the bureaucracy and the provision of public services, the independence and competence of civil servants, and the credibility of government’s commitment to its policies. Corruption (H4b) is measured through subjective assessments of the role of direct bribery as well as the indirect use of money to influence political decisions. Both are measured in the year 2007 and are taken from the World Governance Indicators project, an academic research project funded by the World Bank (Kaufmann et al. 2015). I coded a dummy variable for formerly communist countries (H5) using data from the Encyclopedia Britannica and CIA World Factbook. For years since state formation, I use the number of years the nation has been in existence as an independent state. If a nation came into existence more than once, the paper uses the most recent year.

Religiosity

(H7) is measured using a question that asks whether religion is important in the respondent’s daily life. The variable measures the percentage of respondents in the country who answered yes. Religious diversity (H8) is measured through a variable that measures the probability of any two randomly selected individuals being members of different religious denominations, measured in 2001 (Alesina et al. 2003). The religious composition (H9) of a country is measured by five variables that contain the percentage of the population belonging to each of five religions (Protestant, Catholic, Muslim, Buddhist, and Jewish) taken from different sources and measured in 1998–2006. As only two countries, Pakistan and India, had a significant Hindu population, I did not include this variable in the dataset. For both ethnic diversity (H10) and linguistic diversity (H11), I use a variable that uses 2001 data and calculates the probability that two randomly selected individuals will belong to different ethnic or linguistic groups (Alesina et al. 2003).

Analytical Method

This paper first uses bivariate correlations (Pearson’s R) to test each hypothesis individually, then multiple regression to examine whether each variable has a significant relationship with giving when controlling for the others. To test H13, it tests the first twelve hypotheses on sub-samples of Western and non-Western countries. Western countries include those of Western Europe and the former British colonies of the United States, Canada, New Zealand, and Australia, and all other countries are considered non-Western.

To test the hypothesis (H14) that economic development instead of politics or culture divides the Western and non-Western countries, I divided the non-Western countries into two groups, a middle income group with per capita income above $10,000 (N = 27) and a low-income group with per capita income below $10,000 (N = 66). The $10,000 cutoff point is close to the mean per capita income of non-Western countries ($8166) and allows for a large enough sample size in each group to make statistically significant comparisons possible. The low-income group includes most African countries, Brazil and some other Latin American countries, China, India, Indonesia, and some Asian countries. The middle income group includes most Eastern European countries, some Latin American countries such as Argentina, Chile, and Mexico, and some Asian countries including South Korea, Malaysia, and Iran.

Results

At the bivariate level (Table 2), strong support was shown for the positive role of economic development (H1a) when all countries were considered (r = .557). I generated an exponential term for GDP per capita to test the hypothesis that there was a curvilinear relationship between economic development and charitable giving (H1c), but found no support for this hypothesis. There was a moderate correlation between charitable giving and government spending on health care (r = .398, p ≤ .001), supporting the hypothesis that government welfare spending is positively associated with charitable giving (H2).

Table 2 Correlations between independent variables and charitable giving
Full size table

Freedom of association (H3, r = .409), government effectiveness (H4a, r = .526), and corruption (H4b, r = .539) all correlated significantly (p ≤ .001) with charitable giving in the full sample, and formerly communist countries (H5) had lower rates of charitable giving (r = −.263, p = .004). Timing of state formation (H6) did not correlate significantly with giving.

Neither religiosity (H7) nor religious diversity (H8) correlated significantly with charitable giving, but the percentage of Catholics (r = .213) and Protestants (r = .246) in a country correlated significantly (p ≤ .05) with charitable giving. Ethnic diversity (H9) significantly correlated with charitable giving in the full sample but the direction was negative, the opposite of what was hypothesized (r = −.253, p = .007). Linguistic diversity (H10) did not correlate significantly with charitable giving.

These results describe the relationships found in the entire sample of countries. When one divides the sample into Western, poor non-Western, and middle and upper non-Western countries, a different picture emerges. As predicted (H13), most of the theories that describe variation in charitable giving within Western countries do not describe charitable giving in non-Western countries. Nine variables were statistically significant predictors in the full sample: all of the economic hypotheses and all of the political hypotheses except for country age were supported in the full sample, and significant positive effects were also found for percent Protestant and percent Catholic, as well as a negative effect of ethnic diversity.

In the sample of Western countries only, seven variables had significant effects. As in the full sample, there were significant (p ≤ .10) positive effects of GDP per capita (r = .296), civil liberties (.492), government effectiveness (.468), control of corruption (.450), and percent Protestant (.342). Country age (.492) and religious diversity (.336) were significant in the Western sample but not the full sample, and Percent Catholic and ethnic diversity were significant in the full sample but not the Western sample. As there were no formerly Communist Western countries, this variable could not be tested in the Western sample.

In the sample of non-Western countries only, five variables had significant (p ≤ .10) effects: GDP per capita (r = .328), government effectiveness (.180), formerly communist (−.215), percent Jewish (.253), and percent Buddhist (.377). This finding supports Hypothesis 13 that the variables derived from existing theories of variation in charitable giving more effectively describe giving in Western countries than non-Western ones.

Dividing the sample into middle income and low-income non-Western countries showed that the theories did predict variation better in middle income non-Western countries, as expected. Five variables predicted variation in charitable giving among the wealthier countries: GDP per capita (r = .580), government effectiveness (.405), control of corruption (.361), formerly communist (−.390), and religiosity (.441). Among poorer non-Western countries, only three variables, country age (r = .253), percent Jewish (−.237), and percent Buddhist (.341) had a significant relationship with charitable giving.

After testing bivariate relationships, I used multiple regression to build a predictive model. I constructed regression models for each sample using only those variables that had substantive (r > .3) or statistically significant (p ≤ .05) bivariate correlations with the independent variable. A model for giving in the full sample, using only those variables that significantly or substantively correlated at the bivariate level, predicted giving at R 2 = .441 (Table 3). Similar models incorporating only variables that were substantively or statistically significant at the bivariate level predicted giving at R 2 = .480 for Western countries, .154 for non-Western countries, .501 for middle income non-Western countries, and .191 for poor non-Western countries.

Table 3 Regression analyses
Full size table

Note that most slope coefficients were not statistically significant, a result that has to do with multicollinearity and the small sample size. The sample size also varies greatly among groups, and as R 2 tends to go up as sample size goes down, the small sample size of the Western (n = 19) and middle income non-Western (n = 21) countries tends to exaggerate the predictive power of the models for these two groups. With these warnings in mind, the results provide at least some support for the hypothesis (H13) that the theories work better in Western (n = 19, r = .480) countries than non-Western countries (n = 84, r = .154), and the hypothesis (H14) that the theories work better in middle income non-Western countries (n = 21, r = .501) than poor non-Western countries (n = 53, r = .191).

Discussion

This paper presented a number of theories that predict international variation in the strength of civil society generally and charitable giving in particular, and then tested them on a dataset that included giving data for over 100 countries. Much support was found for economic theories (GDP and government welfare spending) and political theories (civil liberties, government effectiveness, corruption, communism, and country age). Some support was found for cultural theories (religiosity, religious diversity, denomination, and ethnic diversity), but this support was not consistent across samples.

Why are variables that measure economic and political factors more powerful predictors of charitable giving than variables that measure religion and culture? Several possible explanations exist. First, all major world religions emphasize the importance of helping others and giving money to charity, so culture is something of a constant: philanthropy is universally valued. Indeed, extensive cross-cultural study has found that benevolence, the value of helping socially close others, and universalism, the value of helping socially distant others, are not only common across cultures but also at the top of most cultures’ hierarchy of values (Schwartz 2012). What differs from country to country is not the value that people place on helping, but the ways in which they can express this value. Economic and political systems can facilitate the expression of this helping impulse into formal charitable giving, but they can also discourage giving or divert the helping impulse into other activities.

Political factors matter because charitable giving is an activity that takes place through government-recognized charities, and government policies, tax systems, and financial regulation affect how easily non-profits can organize themselves and solicit donations. Thus, countries that are high in civil liberties such as freedom of speech and association would also be countries in which non-profits could easily organize and solicit donations. Countries in which the government is effective and not corrupt provide the stability and rule of law that non-profits need to operate effectively. Countries that suffered from a lack of civil liberties and an outlawed nonprofit sector under communist rule still have lower rates of charitable giving decades after the fall of the Soviet bloc.

The high importance of economic factors is no surprise because charitable giving is itself an economic activity. People must have money in order to give it away, and this study finds that the more money people have, the more they can afford to give. Furthermore, economic factors may strongly influence cultural ones. As countries develop economically, they move from values of tradition and survival to values of secularism, rationality, and self-expression (Inglehart and Welzel 2005). As people move from traditional to secular and rational values, they may also turn from traditional types of giving through friend and family networks to secular and rational forms of giving through organized institutions.

It would be instructive to test whether informal giving to people one knows personally varies as much from country to country as charitable giving, and whether it varies for similar reasons. Unfortunately, there are no good comparative data on direct giving. Studies of internal remittances, or money that migrants from rural areas to cities send home to family members, find this practice to be very common in the developing world, and this type of interpersonal giving may substitute for formal charitable giving as a form of poverty relief in many developing countries (Housen et al. 2013). There are a few single-country studies of informal giving. A nationally representative Mexican survey (Butcher and Sordo 2016) found that 51.2 % of their sample gave money to charity in the last year and the exact same proportion (51.2 %) gave money to people they knew personally. A nationally representative South African survey found that 54 % of respondents had given money to charity in the last month, and 45 % had given money to “a beggar, street child, or someone asking for help” (Everatt et al. 2005). Studies of volunteering have found that people in wealthy and Western countries are more likely to engage in formal volunteer volunteering than people in poor and non-Western countries, but that there is little difference between Western and non-Western countries in informal, direct volunteering. It is possible that the gap between interpersonal giving is smaller, or even that people in poorer countries engage in more interpersonal giving than people in wealthier ones.

Within the area of religion, it is striking that the percentage of the population that is Protestant, Catholic, Buddhist, or Jewish correlates positively with levels of charitable giving, but the percentage of the population that is Hindu or Muslim does not. All of these religions value charitable giving highly in principle, so why do Hinduism and Islam not correlate with charitable giving in practice? The correlation between Jewish population and charitable giving may be explained by the unique history of Jewish philanthropy, as Jewish people, excluded from many government programs, had to found their own charities to provide for basic needs. Most Buddhist societies lack strong philanthropic institutions, but the common practice of making small donations to temples and monasteries may explain the high rate of charitable giving among Buddhists. By contrast, Islamic societies traditionally linked charitable giving through zakat and waqf to government institutions, and the rise of modern secular nation states in the Muslim world has disrupted the traditionally close alliance between government and religious philanthropy. Most countries with a large Hindu population are poor (India, Nepal, Bangladesh, Mauritius, Surinam, and Guyana, among others), so the negative relationship between Hinduism and charitable giving may reflect the poverty of most Hindu countries rather than any characteristic of Hinduism itself.

Beyond the tests of individual hypotheses, the major finding of the paper lies in how well theories derived to explain giving in Western countries explain giving in non-Western ones. At first glance, the variables derived from these theories seem to correlate only weakly with giving in non-Western countries. However, the inapplicability of the theories does not seem to be due to differences between the West and the rest of the world in politics, history, or culture, but due to differences in the economy. When one divides non-Western countries into low-income and middle income countries, the variables effectively predict variation in giving among middle income countries but poorly predict variation in low-income countries. This finding again supports the conclusion that economic factors, not religious or cultural ones, are the main driver of cross-national differences in charitable giving.

While the strength of this paper lies in the number of countries included in the sample, the paper has significant limitations. The question on charitable giving is a simple yes/no question, and the paper only looks at country-level aggregate data, not individual responses. The measures of independent variables are taken from the years 2001 to 2008, so not are all from the same years (2007–2008) that the GWP data were collected.

It was also not possible to test all possible theories of cross-national variation in charitable giving. Wiepking and Handy (2015) list eight facilitating factors for philanthropy: a culture of philanthropy, public trust, regulatory and legislative frameworks, fiscal incentives, the state of the non-profit sector, political and economic stability and growth, population changes, and international giving. Testing all eight theories would be outside of the scope of this paper, but form a promising foundation for future work. This paper concentrated on theories that were testable across a wide range of countries using publicly available data, which fall primarily into Wiepking and Handy’s categories of “culture of philanthropy” and “political and economic stability and growth.” Formulation of variables through detailed country by country research would be necessary to make quantitative tests of many of the other factors. Testing all eight theories is therefore outside of the scope of this paper but the theories form a promising foundation for future work.

Despite these limitations, the findings of this paper have implications for both researchers and practitioners. With R-squared values above .400, the variables used in the regression analysis explain a good deal of the cross-national variation in giving. While some of the hypotheses were not supported, the combined theories of charitable giving do a good job overall of explaining giving in developed Western countries and middle income non-Western countries. Economic and political theories appear to be particularly effective, while theories related to culture, religion, and ethnic and linguistic diversity have less explanatory power (Table 3).

The results indicate that existing theories do a good job of predicting giving in middle income countries but are of little use in predicting giving in low-income countries. There are two possible conclusions from this. First, it may be that the factors that predict variation in giving in low-income countries are simply different from those that predict variation in giving in middle income countries, and that further research only on low-income countries will discover new theories about causes. However, it could also be that the variation in the small amounts of charitable giving that occur in these countries is more or less random or relates to factors specific to each country, and that giving patterns do not become predictable until economic development reaches a certain level.

Of the two variables that predict giving in low-income countries, it is not surprising that country age predicts and percent Buddhist predicts giving, but it is surprising that other religious affiliations do not predict giving. In regards to country age, countries need time to develop non-profit institutions that can accept donations and the legal framework to make donations possible. Countries that had only recently become independent would have less time to develop these institutions. The fact that the percentage of Buddhists in the population predicts giving is not surprising, but it is curious that the presence of Muslims and people of other religions does not have a similar positive effect, and the percentage of Hindus actually correlates negatively with charitable giving. The question of what is distinctive about the Buddhist religion in encouraging charitable donations is a question for future scholarship. Also, future research using more complete data about amounts of charitable giving and giving to different types of organizations can replicate and expand upon the findings of this paper.