Notes

^1.

Couples are defined as partners of opposite sex only. Same-sex couples were excluded from the analysis because of their small sample size (73 same-sex couples).

^2.

Under the assumption that fathers and mothers each keep their own incomes separate.

^3.

That is, that there is inequality in control over financial flows.

^4.

The Gini coefficient is a measure of inequality of a distribution, with a value of 0 meaning total equality and a value of 1 meaning total inequality.

^5.

Such as the roles that men and women are viewed as appropriately playing within the home or labour force.

^6.

Similarly, in terms of the operation of these strategies, Vogler, Brockmann, and Wiggins (2006, p. 459) review the literature and suggest that couples using a pooled approach seek "…to achieve equality of outcomes, in the sense of equal control over and access to joint money, even though they may make very different financial contributions to the pool," while couples using a separate approach "…are much more likely to define equality in terms of equal inputs, in the sense that both partners make equal contributions to collective expenditure ('going 50/50') despite often having very different levels of income."

^7.

Proxy responses were provided in 618 cases, or 2.6% of the sample.

^8.

This includes 63 respondents who reported that they use strategies 'other' than the five responses listed in the paper and 380 respondents who either did not know what type of strategy they used or did not answer the question. The incidence of item non-response on the question was comparable to that on standard questions, such as educational attainment.

^9.

However, note that respondents who did not answer GSS questions regarding income were flagged and retained in the analysis.

^10.

This resulted in the exclusion of 218 cases in which adult children (aged 25 or older) received income and 2261 cases in which other family members, such as children aged 15 to 24 or extended-family members, received income. The inclusion of these cases would have made it impossible to measure the relative income contribution of the spouses in the sample for this study—a key variable in the literature.

^11.

Again, only 63 respondents reported using another type of system, and these respondents have been excluded from the analysis. The categorization used in this study is more detailed than the two-category classification used in a number of studies. Vogler et al. (2008), Treas (1993), and Heimdal and Houseknecht (2003) combined the allocative and pooled strategies into a single category. In part, this was done for reasons of sample size, but Vogler et al. (2008) also argued that the allocative and pooled strategies are "…systems in which money is constructed as collectively owned and couples operate more or less as single economic units." This is different from the separate strategy, which reflects "individualized systems." (p. 120).

^12.

2002 ISSP data show that the 'allocative approach' is used by 3% to 5% percent of couples in Sweden, Finland, and Norway, by about 15% to 25% of couples in 'Anglo' countries (the United States, United Kingdom, Ireland, Australia, and New Zealand), by about 45% to 55% of couples in Brazil, Mexico, and Chile, and by 70% of couples in the Philippines and Japan. Canada is not included in the 2002 ISSP.

^13.

That is: both partners born in Canada (reference group); husband born in Canada and wife immigrated; husband immigrated and wife born in Canada; and both partners immigrated.

^14.

Since, in the GSS, education is a categorical variable, the wife's or female partner's relative education to that of her husband or male partner is defined as the wife's or female partner's level of education minus the husband's or female partner's level of education.

^15.

The GSS includes a continuous variable on respondent personal income, a categorical variable on household income and the number of income recipients in the household. From this information, it was possible to derive income for both the wife or female partner and the husband or male partner. Using lower bounds and middle points of the household income categories gave very similar regression results. The wife's or female partner's income relative to her husband's or male partner's is defined as the difference between wife's or female partner's income and husband's or male partner's income.

^16.

For example, one might hypothesize that receipt of Old Age Security benefits at age 65 provides elderly women with an independent, personal source of income over which they prefer retaining control and thereby increases the propensity to use at least a partially separate approach to finances. A variable identifying women who were younger than age 65 and women who were older than age 65 did not provide support for this hypothesis.

^17.

Among older couples, cohabitation (i.e., common-law status) may follow a previous marriage; this raises the possibility of overlap between these two variables. To assess this, the model was run with the common-law variable excluded and subsequently run with a variable combining common-law status and previous marriage. These specifications yielded the same result; that is, previous marriages remained insignificant.

^18.

Different reference categories were used for men and women on the income variable.

^19.

Dummy variables for all the provinces in Canada were used in an earlier version of the analysis, but yielded results similar to results yielded by the "Quebec–Rest of Canada" dummy. The latter was retained in the model for sake of parsimony and ease of presentation.

^20.

The negative effect of family composition reflects the fact that the presence of children reduces the likelihood of using a separate income management strategy, and this characteristic is less prevalent among common-law than married couples.

^21.

These decomposition results are based on the coefficients from the sample of married respondents. When coefficients from the sample of common-law respondents are used, these results are similar. When one uses 'common-law coefficients', compositional characteristics account for 16.7 percentage points, or 50%, of the difference in prevalence of the separate approach (compared with 14.0 percentage points, or 41%, when the 'married' coefficients are used). In both approaches, relationship duration accounts for the largest share of the explained component.