Determinants of Decision Making on Charitable Giving

In their focus group study Burgoyne et al. (Reference Burgoyne, Young and Walker2005) conclude that partners in general decide individually rather than as a couple over smaller incidental donations, like door-to-door collections, or street collections. The story is different when larger and more structural donations are considered. In these cases, charitable giving is considered a ‘normal’ household expenditure. Decisions on these expenditures follow the ‘normal’ rules for financial decision making in a household. Thus, how a couple decides on more structural donations depends for a large part on the general system of financial organization present in a household. In this study, we will focus on decision making in the case of these somewhat larger and more structural donations.

Pahl (Reference Pahl1983, Reference Pahl1995) and Vogler (Reference Vogler2005) have developed a widely used typology of general financial management in relationships, based upon whether couples operate as a single or as two separate economic units. According to Pahl and Vogler’s typology, when couples operate as a single economic unit (which we further refer to as ‘one economic actor’), the couple can allocate their money using three mechanisms: First, couples can arrange family finances according to the whole wage system, in which all money is managed by one partner, who distributes some spending money to the other partner. Second, couples can allocate money by means of the housekeeping allowance system. In this system one spouse—usually the husband—gives the other partner money to pay for daily household expenditures and controls the rest of the money. Third, couples can share their income using the joint pooling system. In this system, partners use a ‘common’ pot to pay for joint expenses, and take some individual spending money out of this pot.

When couples function as two separate economic units (which we further refer to as ‘two economic actors’), they can either choose to allocate their money by means of the partial pooling system or by means of the independent management system. In the partial pooling system a couple pools some of their income to pay for collective expenditures, but keeps most of the money separate. In the independent management system partners keep purses completely separate.

Except a study by Giesen and Kalmijn (Reference Giesen, Kalmijn, Kalmijn, Bernasco and Weesie1999), no research has been conducted on general financial management systems in the Netherlands, let alone on financial decision making in the case of charitable giving. The results by Giesen and Kalmijn indicate that financial management by Dutch couples can be deduced to the type of relationship. Married couples pool their income and decide on most finances jointly, and cohabiting couples have separate accounts and make individual financial decisions. Additional results show that Dutch couples more often have joint accounts when they are lower educated, have children, and are politically right oriented. In this study we will investigate whether these general results resemble financial decision making with respect to charitable giving.

In general research on financial management in households three theories prevail: Sunk cost theory, resource theory, and gender role theory (e.g., Treas Reference Treas1993; Pahl Reference Pahl1995; Heimdal and Houseknecht Reference Heimdal and Houseknecht2003). These three theories all identify different sets of determinants of financial management in households. We implement these theories in order to predict the financial management system that couples use for charitable giving. We start with sunk cost theory.

Sunk cost theory predicts that financial management in households is determined by the level of relation-specific investments partners have made (Treas Reference Treas1993). The more relation-specific investments a couple has made, the more likely it is that they will pool their money and function as one economic actor. As Treas puts it: ‘Marriage-specific investments call for collectivization’ (Treas Reference Treas1993, p. 725). The less relation-specific investments, the higher the probability partners decide separately on monetary issues. In addition, when the risk of separation is higher, people will be less inclined to make relationship-specific investments (Heimdal and Houseknecht Reference Heimdal and Houseknecht2003; Kenney Reference Kenney2006). Indicators for relation-specific investments are for example home ownership, and taking care of children.

Results of past research are in line with Treas’ argument about sunk costs. Couples in which one of the partners has experienced the disruption of a former relationship—either in the form of a divorce or by the decease of a partner—more often keep money separate, and thus function as two economic actors (Treas Reference Treas1993; Heimdal and Houseknecht Reference Heimdal and Houseknecht2003; Vogler et al. Reference Vogler, Brockmann and Wiggins2006). The same holds for cohabiting couples, who—in contrast to married couples—more often use separate purses (Giesen and Kalmijn Reference Giesen, Kalmijn, Kalmijn, Bernasco and Weesie1999; Treas and Widmer Reference Treas, Widmer, Weesie and Raub2000; Heimdal and Houseknecht Reference Heimdal and Houseknecht2003; Vogler Reference Vogler2005; Vogler et al. Reference Vogler, Lyonette and Wiggins2008).

From sunk cost theory we predict that couples with more relation-specific investments (such as home ownership and the number of children) will more often function as one economic actor. Hence they will use one of the systems in which partners decide as one actor on financial expenditures in order to make charitable donations. In addition, we also predict that couples who have previously experienced a serious relationship disruption will be more careful to make relationship-specific investments, and prefer to act as two separate economic actors.

From resource theory we can deduce two different predictions about financial management in households, which depend on either absolute or relative socio-economic resources. With respect to absolute resources, researchers argue that higher levels of absolute resources enable partners to achieve goals without the help of their spouse and give them more bargaining power over financial decisions, such as charitable donations (Williamson Reference Williamson1975; Blumstein and Schwartz Reference Blumstein and Schwartz1983; Treas Reference Treas1993). Woolley (Reference Woolley2000) for example showed that when the male partner has a higher income, he is more likely to control financial management, a result also confirmed by Richie (Reference Richie1990) in Waseem Reference Waseem2004). Additionally, according to both Vogler (Reference Vogler1998) and Pahl (Reference Pahl1983), women who are employed are more likely to have control over financial management, irrespective of their income. Furthermore, Treas (Reference Treas1993) found that couples in which the wife works are more likely to have separate accounts. Therefore, from resource theory we first predict that couples with less financial and educational resources are more likely to pool income, and hence make charitable donations as one economic actor.

Relative resource theory predicts who in the relationship has most control over money, when couples decide as one actor. According to this theory, the level of control over financial management of a spouse increases with their level of socio-economic resources or knowledge about finances relative to that of their partner (Treas Reference Treas1993; Pahl Reference Pahl1995; Woolley Reference Woolley2000; Andreoni et al. Reference Andreoni, Brown and Rischall2003; Kenney Reference Kenney2006). When a man has a higher income, completed a higher educational level or has significantly more knowledge of financial matters than his female partner, his control over a household’s pooled money is larger (Waseem Reference Waseem2004; Kenney Reference Kenney2006). The same argument is valid for control over financial management by women. When their income, educational level, or knowledge of financial matters is higher than that of their male partners, they will have more to say about the household’s pooled money. When we implement relative resource theory in order to predict the financial management system couples use for charitable giving, we argue that in the case where couples make monetary decisions as one actor, the partner with highest level of financial and educational resources will be in control of the pooled money.

Finally, gender role theory states that traditional views on gender roles may impose limits to the financial management options of women (Thornton et al. Reference Thornton, Alwin and Camburn1983; Lundberg and Polak Reference Lundberg and Polak1996; Kenney Reference Kenney2006). In families with more traditional values, household finances are more likely to be characterized by the allowance system or by the male controlled pooling system, as handling money is traditionally regarded a male task (Tichenor Reference Tichenor1999). Ludwig-Mayerhofer (Reference Ludwig-Mayerhofer, Weesie and Raub2000) shows that in Germany, traditional gender values indeed lead to a higher likelihood of male controlled financial management. Treas and Widmer (Reference Treas, Widmer, Weesie and Raub2000) studied money management in 23 countries, and find that both the individual approval of women’s employment and the mean level of women employed in a country have a positive effect on keeping money separate. Kenney (Reference Kenney2006) found that more traditional views on gender roles increase the likelihood of the use of a male controlled pooling system in the United States. In contrast, Heimdal and Houseknecht (Reference Heimdal and Houseknecht2003) found no effect of gender values on a couple’s financial management system in both Sweden and the United States. They gave no explanation for this deviant result. From gender role theory we predict that couples with more traditional views on gender roles will more often donate money to charitable causes as one economic actor and have the male partner decide on these expenditures.

Financial Management as One Economic Actor

We start with arguments about the consequences of managing finances when couples decide upon expenditures as one economic actor. As discussed, according to Pahl (Reference Pahl1983, Reference Pahl1995) and Vogler (Reference Vogler2005) there are three mechanisms along which couples can arrange financial management as one economic actor: (a) the whole wage system, (b) the allowance system, and (c) the pooling system.

When couples arrange their finances according to the whole wage system or the allowance system, one partner controls all financial resources, and hence couples have no discussion about expenditures. The results by Andreoni et al. (Reference Andreoni, Brown and Rischall2003) show that it does matter specifically for the level of charitable giving whether the male or the female partner controls the whole wage or allowance system. With gender role theory we can account for this difference. Couples with more traditional gender role values are more likely to have the male partner in control of financial management. At the same time, couples with more traditional gender values are also likely to be members of more conservative denominations (Jansen Reference Jansen and Jansen2002). And in the Netherlands, members of conservative denominations are also known to be large charitable donors (Bekkers and Schuyt Reference Bekkers and Schuyt2008). Hence, from gender role theory we predict that couples arranging financial management as one economic actor by means of the male controlled whole wage or allowance system are large charitable donors because of their more conservative religious denomination and the more traditional gender role values connected to this religion.

When couples arrange financial management with the pooling system, partners take spending money out of a common pot, and use the rest of the money to pay for joint expenses. Using the pooling system, couples can either jointly decide about charitable giving, or use their personal spending money to make charitable donations. In the latter case the couple acts as two economical actors, a situation that is discussed later.

According to signalling theory, partners give each other incentives to donate higher amounts to charitable organizations. Signalling theory emphasizes the private benefits for spouses in negotiations on charitable giving. It is well known that people have a preference to select partners with pro-social attitudes and personality characteristics on the marriage market (Buss and Barnes Reference Buss and Barnes1986; Botwin et al. Reference Botwin, Buss and Shackelford1997), and it is likely that these traits contribute to maintenance of the marital relationship. In order to maintain a good marital relationship, spouses have an incentive to appear to each other as cooperative, trusting, empathic, and having high moral standards (Milinski et al. Reference Milinski, Semmann and Krambeck2002; Wedekind and Braithwaite Reference Wedekind and Braithwaite2002). We argue that spouses use charitable giving strategically in order to signal their pro-sociality to each other. Donating higher amounts to charitable organizations can signal a partner’s pro-sociality, but only under the condition that partners agree—at least to some extent—with each other on donations. It is hard to signal pro-sociality while donating to organizations that the partner disagrees of. From signalling theory we predict that having similar preferences for charitable giving will increase the amount donated to charitable organizations.

Transaction cost theory predicts that most couples will try to make joint financial decisions in a way that causes them least time and trouble. Partners try to minimize their transaction costs, which have to do with negotiating and coordination of financial management (Williamson Reference Williamson1981; Treas Reference Treas1993; Andreoni et al. Reference Andreoni, Brown and Rischall2003). According to transaction cost theory, when partners jointly decide it matters whether or not they agree with each other on charitable giving. When partners have opposing preferences on charitable contributions, the amount of time required to reach decisions on charitable contributions will be higher. In addition to that, when partners have opposing tastes, the charitable donation of one partner creates ‘negative externalities’ experienced by the other partner, and this will reduce the donations of a jointly deciding couple drastically (Andreoni et al. Reference Andreoni, Brown and Rischall2003, p. 113). On the other hand, when partners have similar preferences for charitable giving, it takes little time to make decisions about the charitable expenditures and this is likely to benefit the amounts donated to charitable causes. From transaction cost theory we predict that when jointly deciding couples have similar tastes for giving, this will increase the amount donated. But when jointly deciding couples have opposing tastes, this will decrease the amount donated.

Financial Management as Two Economic Actors

When partners function as two separate economic actors, they will make charitable donations out of their own financial accounts, no matter whether they allocate their money by means of the partial pooling system or by means of the independent management system. According to transaction cost theory separate financial decision making results in higher levels of giving than financial decision making as a single actor (Andreoni et al. Reference Andreoni, Brown and Rischall2003). This argument is based on the assumptions that each partner receives utility for making charitable donations, that the utility of each donated dollar is equal for both partners, and that both jointly and separately deciding actors strive for the same level of utility.

Assume that $1 equals 1 ‘utility’-unit. When a couple donates €20 jointly (as a single economic actor), both partners receive 20 utility-units. In a situation where both partners separately donate €10, the total amount received by the charity is also €20, but both partners receive only 10 utility-units. In this example, in order to achieve the same level of utility, separately deciding partners need to donate twice the amount of couples deciding as one economic actor in order to achieve the same level of utility. From transaction cost theory we predict that couples deciding as two economic actors will donate higher amounts than couples deciding as one economic actor.

The Giving in the Netherlands Panel Study

We use data from the third wave of the Giving in the Netherlands Panel Study (GINPS05 2005). The GINPS is a bi-annual longitudinal study on charitable giving and volunteering in the Netherlands, which started in 2001. In May 2006, 1,868 persons were questioned about their donating behavior in 2005. In total, 1,474 respondents (79%) completed the questionnaire, using Computer-Assisted Self-Administered Interview procedures (CASI).

We are interested in financial decision making in couple households. Therefore, we selected only households in which respondents are involved in a relationship.Footnote ¹ This selection reduced our sample to 1,103 respondents, of which 892 (81%) are married, 151 (14%) are unmarried cohabiting, and 60 (5%) are ‘Living Apart Together’ (LAT; couples in which partners live separately from one another). This is representative for national figures of types of partnerships in the Netherlands (Dijkstra and Komter Reference Dijkstra and Komter2004; Eurostat 2005). In addition, we excluded two cases with a monthly after tax household income over €11,000 from the analyses, because these cases influenced the results significantly.Footnote ² This reduced the sample used in the analyses to 1,101 cases.

Measures of Dependent Variables

The first dependent variable measures whether a couple in general makes decisions on charitable donations over €10 as one economic actor.Footnote ³ In order to obtain this information, we asked respondents whether (a) the respondent and partner make decisions on charitable giving separately; (b) whether the respondent decides; (c) whether the respondent’s partner decides or (d) whether the respondent and partner decide jointly. Couples in which partners decide separately are regarded as deciding as two economic actors (16%), all others are considered to be deciding as one economic actor (84%). We included the minimum amount of €10 in the question to specifically investigate decision making on somewhat larger and more structural donations, and exclude decision making on incidental small donations (such as door-to-door or street collections).

The second dependent variable is a measure of financial decision making authority in the case where couples decide as one economic actor. By combining the reports on couples’ decision making with the respondents gender, we create dichotomous variables for the female partner decides (18%), the male partner decides (10%), or decisions are made jointly (72%).

To investigate the effects of financial management on charitable donations, we construct the amount of money a household donated to various charitable causes in 2005. We are first and foremost interested in the different effects of decision making on structural (larger) donations. However, we also want to compare charitable decision making on larger donations with decision making on total donations. Therefore, we construct two variables for the amount households donated. The first variable measures total amount donated in 2005 for all donations. The second variable is limited to the amount donated for households donating more than €50 in 2005, which likely excludes most households that only make small spontaneous donations. 702 couple households (64%) have donated over €50 in total in 2005.

The amount of money a household donated in 2005 is measured with questions on donations to eleven different categories of charitable causes. The questionnaire uses a survey module that minimizes the risk that people underreport donations.Footnote ⁴ First, respondents were questioned about which method they used to make a donation, for example, a donation canister or a credit slip. After that, for each category of charitable causes, respondents were asked whether or not their household had made a donation. Third, the respondents reported the amount donated in 2005 to each of the categories they indicated having donated to. Out of the sample of 1,101 households, 258 (23%) did indicate to have made a donation to one or more specific sub-sector(s), but failed to specify the exact amount. In these cases, the missing amounts were imputed independently for each charitable sub-sector using the complete original dataset (N = 1,474). One problem with using multiple imputation in giving data is that the missing values are not Missing at Random (MAR), an assumption that needs to be satisfied when imputing missing data (Allison Reference Allison2002). This is problematic, but multiple imputation seems the best solution for dealing with missing values in giving data. See Brooks (Reference Brooks2004) for more information and a discussion on using multiple imputation in charitable giving data. Multiple imputation uses linear predictions, which can result in negative donations. This is a minor problem, as after imputation only 1% of the donations are negative. In line with Schafer (Reference Schafer1997) we substitute all donations below 1 with a donation of €1. This enables us to calculate and use the natural log of the total amount donated and the natural log of total amount donated for households donating more than €50 as third and fourth dependent variable.

Measures of Predictor Variables

Difference in educational level between husband and wife is composed by subtracting the wife’s education from that of the husband’s. Mean educational level consists of the sum of the educational level of both partners, divided by two. Educational level of both partners is measured on a seven-point scale, with (1) only primary education; (2) primary education and some vocational school; (3) lower secondary education; (4) middle secondary education; (5) higher secondary education; (6) higher vocational education; (7) higher tertiary education.Footnote ⁵

Difference in monthly after-tax income (in €1,000) between husband and wife is composed by subtracting the wife’s income from that of the husband’s.Footnote ⁶Mean income in €1,000 consists of the sum of monthly after-tax income of both partners, divided by two. Monthly after-tax income is measured for both the respondent and partner by asking for the exact amount each partner earned net per month in 2005, for eight different sources of income (including income from a paid job, and income from various benefits). When respondents would not give their exact net monthly income from these sources, for each source they were given eight categories to indicate the level of their or their partner’s income. We summed all sources of income for each respondent and their partners in order to acquire a variable for net monthly household income. This method resulted in less than 1% missing values on the income variables, which were replaced with the mean income on the specific source of income.

Respondents were prompted about their traditional views on gender roles with three–five-point Likert scale items: (1) ‘A working mother can be just as good a mother as a mother who stays at home for her children’ (reversed); (2) ‘It is not right if a man stays at home and a woman has a paid job’; (3) ‘It is the task of a man to earn money and the task of a woman to care for children’. Cronbach’s alpha for these three items is 0.704.

In addition we include religious affiliation in the analyses, as we know that this is a strong determinant of traditional norms on gender roles (Jansen Reference Jansen and Jansen2002). Households belonging to strict Protestant denominations are more likely to have traditional views of gender roles. We questioned respondents about their own religious denomination and that of their partner. With this information we computed variables indicating whether the partners form a strict Protestant couple (13%), with non-religious couples and couples belonging to other—less strict—denominations (Roman Catholic, Dutch Reformed Protestant) as the reference category. In the Netherlands there are several strict Protestant Christian denominations, which either are part of, or split off of the Rereformed Protestant Church. These congregations are mainly located in small towns in the upper-half of the country, in the province of Zeeland and in the Veluwe, also known as the ‘Biblebelt’. Members of these congregations live according to strict religious rules, for instance, they are not allowed to use any means of transportation on Sundays and they visit church at least twice on that day. Among these people traditional gender roles are strictly enforced. We control for church attendance in times a year.

We measured the level of consensus on philanthropic donations in the household with the question whether respondents usually agree with their partner on deciding on charitable giving. Response categories include: (1) I do not interfere in donating at all (8%); (2) we mostly disagree (2%); (3) sometimes we agree and sometimes we disagree (11%); (4) we mostly agree (41%); (5) we always agree (38%). We merged response categories 4 and 5, so that four new categories were created: respectively indicating being indifferent (response category 1), having opposing tastes (response category 2), having fluctuating tastes (response category 3), and having (more or less) similar tastes with respect to philanthropic donations (response categories 4 and 5; reference category).

As indicators for sunk costs, we include information on home ownership (69%), number of children in the household (μ = 1.8, σ = 1.4), and whether the respondent has ever experienced a divorce (8%).

Control variables include respondents’ age (under 35 and over 65 years, with between 35 and 65 as reference category) and the dichotomous variable female (53%) to indicate whether the respondent is female. Table 1 gives an overview of all variables used in the analyses.

Table 1

Descriptive statistics for study variables (GINPS05)

Variable	Obs.	Mean	S.E.	Min.	Max.
Decision making as one economic actor	1,101	0.84	–	0	1
Financial decision making authoritya
Male decides	928	0.10	–	0	1
Female decides	928	0.18	–	0	1
Couple decides jointly	928	0.72	–	0	1
Total household donationb	1,101	4.31	1.88	0	8.52
Total household donation over €50c	702	5.43	1.08	3.93	8.52
Marital status
Married	1,101	0.81	–	0	1
Cohabiting	1,101	0.14	–	0	1
Living apart together	1,101	0.05	–	0	1
Divorced	1,101	0.09	–	0	1
Church attendanced	1,101	6.70	21.33	0	104
Strict Protestant couple	1,101	0.13	–	0	1
Traditional gender role values	1,101	2.14	0.80	1	5
Tastes in giving
Indifferent tastes	1,101	0.08	–	0	1
Opposing tastes	1,101	0.02	–	0	1
Fluctuating tastes	1,101	0.11	–	0	1
Similar tastes	1,101	0.79	–	0	1
Mean educational level couple	1,101	3.94	1.40	1	7
Educational differencee	1,101	0.24	1.54	−5	5
Mean income couplef	1,101	1.10	0.54	0	5
Income differenceg	1,101	0.79	1.02	−4.05	5.50
Female	1,101	0.52	–	0	1
Age under 35	1,101	0.23	–	0	1
Aged between 35 and 65	1,101	0.58	–	0	1
Age over 65	1,101	0.17	–	0	1
Children	1,101	1.84	1.36	0	10
Home ownership	1,101	0.69	–	0	1

^aDecision making as one economic actor, separate deciding households (N = 173) are excluded; ^b ln(total household donation); ^c ln(total household donation if over €50); ^d church attendance in times a year; ^e educational level husband − educational level wife; ^f mean income couple/1,000; ^g (income husband − income wife)/1,000; descriptive statistics based on first imputed dataset

Consequences of Financial Management for Total Donations

Table 3 shows the results of OLS regression analyses on the natural log of the total amount all couple households donated to charitable organizations in 2005. Model 3a in Table 3 includes only decision making authority. Separately deciding households donate significantly less to charitable organizations than households with male decision makers. Using additional Wald tests, we find no difference between female deciding and joint deciding couples (Wald test χ² = 0.80, p = 0.37). We do find significant differences in the amounts donated between female and separate deciding couples (Wald test χ² = 4.93, p = 0.03) and between joint and separate deciding couples (Wald test χ² = 3.64, p = 0.06). Separately deciding couples donate significantly lower amounts than couples with other decision making strategies. This is not what would be expected from the transaction cost hypothesis as formulated by Andreoni et al. (Reference Andreoni, Brown and Rischall2003). According to this transaction costs hypothesis separate financial decision making (as two economic actors) results in higher levels of giving than financial decision making as a single actor.

Table 3

Results of OLS regression analyses on the natural log of the total amount donated to charitable causes by all Dutch couple households in 2005 (GINPS05; based on nine imputed datasets)

	All couple households (n = 1,101)
	Model 3a		Model 3b		Model 3c		Model 3d
	B	S.E.	B	S.E.	B	S.E.	B	S.E.
Male decides (ref.)	–	–	–	–	–	–	–	–
Female decides	−0.035	0.259	0.170	0.244	−0.399	0.257	0.168	0.235
Couple decides jointly	−0.186	0.217	−0.118	0.201	−0.290	0.231	−0.103	0.206
Separate decisions	−0.503*	0.253	−0.274	0.235	−0.596*	0.254	−0.307	0.224
Church attendancea			0.022**	0.003			0.020**	0.003
Strict Protestant couple			1.275**	0.189			1.265**	0.178
Traditional gender role values			−0.201**	0.069			−0.114(+)	0.069
Similar tastes (ref.)					–	–	–	–
Indifferent tastes					−0.896**	0.258	−0.619**	0.226
Opposing tastes					−0.187	0.517	0.304	0.452
Fluctuating tastes					−0.620	0.295	0.241	0.259
Jointly decide with similar tastes (ref.)					–	–	–	–
Jointly decide with opposing tastes					−0.846	0.984	−1.521(+)	0.851
Jointly decide with fluctuating tastes					−0.274	0.389	−0.549	0.344
Jointly decide with indifferent tastes					−0.374	0.484	−0.367	0.418
Divorced							−0.517**	0.177
Mean educational level couple							0.235**	0.042
Educational differenceb							−0.019	0.036
Mean income couplec							0.407**	0.115
Income differenced							0.052	0.055
Female							0.026	0.106
Aged under 35							−0.438**	0.140
Aged over 65							0.640**	0.153
Children							0.086*	0.043
Home ownership							0.384**	0.112
Constant	4.511**	0.204	4.519**	0.249	4.704**	0.215	2.603**	0.336
R square	0.01		0.17		0.01		0.31

⁽⁺⁾p ≤ 0.10; * p ≤ 0.05; ** p ≤ 0.01

^aChurch attendance in times a year; ^b educational level husband − educational level wife; ^c mean income couple/1,000; ^d (income husband − income wife)/1,000

The gender role hypothesis predicts that male deciding couples donate higher amounts because of their more conservative religious denomination and the more traditional gender role values connected to this religion. In model 3b in Table 3 we include belonging to a strict Protestant denomination, having more traditional gender role values and church attendance as indicators for traditional gender role characteristics. These indicators suppress the significant difference in giving between separate deciding and male deciding households, and between joint deciding and separate deciding households. However, the inclusion of more traditional gender role characteristics result in a new significant difference in level of giving between female and joint deciding couples (Wald test χ² = 3.31, p = 0.07). The significant difference between female and separately deciding couples remains significant in model 3b (Wald test χ² = 5.12, p = 0.02). Once we control for traditional gender role characteristics, female deciding households are more generous than separate and joint deciding households.

Next, we consider the effect of tastes for giving in model 3c in Table 3. From signalling theory we derived the hypotheses that similarity in tastes for giving lead to higher charitable donations, because partners use giving in order to signal their pro-sociality. The results show that similar tastes for giving lead to higher donations. However, couples with similar tastes for giving only donate significantly more money than couples with indifferent tastes for giving. From transaction cost theory we predicted that similar tastes in giving have a positive effect on giving when couples decide jointly. Also, from transaction cost theory we derived the hypothesis that when jointly deciding couples have opposing tastes this lowers the amount donated. The interaction effects of jointly deciding and opposing, fluctuating, and indifferent tastes for giving do not differ significantly from the interaction between jointly deciding and similar tastes. At the same time, after the inclusion of tastes for giving, differences between decision making authorities remain. Specifically, the separately deciding couples donate significantly less compared to all other couples. Separately deciding households give less than male deciding households (see model 3c in Table 3), female deciding households (Wald test χ² = 6.85, p = 0.01), and jointly deciding households (Wald test χ² = 2.97, p = 0.09). Hence, the results in model 3c in Table 3 show no support for the transaction costs hypothesis.

After the inclusion of all predictor and control variables in model 3d in Table 3, still two differences in effects of decision making authority on charitable giving remain. Female deciding households donate higher amounts than jointly deciding households (Wald test χ² = 2.83, p = 0.09) and separately deciding households (Wald test χ² = 6.19, p = 0.01). Model 3b in Table 3 shows that this effect is caused by the inclusion of traditional gender role characteristics. Another interesting result in the complete model is the significant interaction effect of jointly deciding and opposing tastes. Couples deciding jointly with opposing tastes for giving donate lower amounts than jointly deciding couples with similar tastes. This result provides support for the transaction cost hypothesis. This hypothesis predicted that jointly deciding couples with opposing tastes donate lower amounts.

Furthermore, in the complete model in Table 3 there is a positive effect of church attendance, belonging to a strict Protestant religious denomination, mean educational level, mean income, age, the number of children, and home ownership on total amount donated to charitable organizations. Traditional gender role values, having indifferent tastes for giving, and having experienced a divorce negatively effect total charitable giving.

Consequences of Financial Management for Donations over €50

Next, we discuss the effect of decision making authority on the level of giving, for donations over €50. The results in model 4a in Table 4 show that when only larger donations are considered, female deciding households donate significantly lower amounts to charitable organizations than male deciding households. Using additional Wald tests, we find no other significant differences between decision making authorities in amount donated over €50.

Table 4

Results of OLS regression analyses on the natural log of the amount donated over €50 to charitable causes by Dutch couple households in 2005 (GINPS05; based on nine imputed datasets)

	Couple households with a total donation over €50 (n = 702)
	Model 4a		Model 4b		Model 4c		Model 4d
	B	S.E.	B	S.E.	B	S.E.	B	S.E.
Male decides (ref.)	–	–	–	–	–	–	–	–
Female decides	−0.349*	0.175	−0.159	0.161	−0.337(+)	0.175	−0.142	0.156
Couple decides jointly	−0.235	0.151	−0.176	0.137	−0.250	0.155	−0.129	0.134
Separate decisions	−0.273	0.180	−0.099	0.162	−0.283	0.181	−0.115	0.156
Church attendancea			0.012**	0.002			0.011**	0.002
Strict protestant couple			0.523**	0.110			0.568**	0.106
Traditional gender role values			0.043	0.051			0.097(+)	0.053
Similar tastes (ref.)					–	–	–	–
Indifferent tastes					−0.188	0.205	−0.135	0.180
Opposing tastes					−0.234	0.397	−0.053	0.353
Fluctuating tastes					0.017	0.225	0.198	0.194
Jointly decide with similar tastes (ref.)					–	–	–	–
Jointly decide with opposing tastes					−0.179	0.740	−0.659	0.646
Jointly decide with fluctuating tastes					−0.145	0.286	−0.372	0.247
Jointly decide with indifferent tastes					0.284	0.441	0.007	0.385
Divorced							−0.127	0.140
Mean educational level couple							0.185**	0.032
Educational differenceb							−0.017	0.026
Mean income couplec							0.018	0.079
Income differenced							0.032	0.039
Female							0.044	0.076
Aged under 35							−0.188(+)	0.180
Aged over 65							0.367**	0.102
Children							0.081**	0.030
Home ownership							0.181*	0.089
Constant	5.671**	0.142	5.263**	0.172	5.701**	0.146	3.987**	0.240
R square	0.01		0.20		0.01		0.31

⁽⁺⁾p ≤ 0.10; * p ≤ 0.05; ** p ≤ 0.01

^aChurch attendance in times a year; ^b educational level husband − educational level wife; ^c mean income couple/1,000; ^d (income husband − income wife)/1,000

In model 4b in Table 4, we include the indicators for traditional gender role characteristics. After this inclusion, the difference in charitable donations between male and female deciding households disappears. Both belonging to a strict Protestant religious denomination and church attendance have a positive effect on charitable donations. There is no effect of traditional gender role values on the amount households donate to charitable organizations, when considering only donations over €50. Additional analyses reveal that specifically belonging to a strict Protestant religious denomination explains the higher donations by male deciding households compared to female deciding households.Footnote ⁷ These results provide additional support for the gender role hypothesis, specifically in the case of larger (>€50) donations.

In model 4c in Table 4 we include different tastes for giving. Whether partners have similar or opposing tastes do not matter for the amount households give, considering they donate more than €50. When controlled for tastes for giving, male deciding households still donate significantly higher amounts than female deciding households. These results do not support the hypotheses derived from both signalling theory and transaction costs theory, for donations over €50.

In the complete model in Table 4 (model 4d), the results resemble the findings in models 3b and 3c. There are no differences between decision making authorities in amounts donated over €50. However, in contrast to model 3b, in model 4d there is a slightly significant positive effect of traditional gender role values on donating over €50. This is both interesting and puzzling, as there is a negative effect of traditional gender role values on total donations. Furthermore, there are positive effects of mean level of education, age, the number of children, and being home owner on amount donated over €50.