The introduction of tax splitting for pension income has provoked debate over whether Canada should extend splitting to other types of income. Indeed, some of the same people who advocated pension splitting are leading the call for broader income splitting for couples. As once Tory, then independent and, as of writing, Liberal MP Garth Turner asserted, “Now that we’ve opened the door to pension-splitting for seniors, it’s only a matter of time before this principle extends through society” (Bailey 2007). Similarly, economist Don Drummond remarked, “I can’t see any particular logic of why, as a concept, it would be accepted for pension splitting and not be viewed as acceptable for general income splitting” (quoted in Whittington 2006). Popular debate over income splitting has been lively, with newspaper editorialists weighing in both strongly favourable¹ and clearly opposed.² Similarly, columnists and media commentators have taken positions both supportive (see Coyne 2006; Taylor 2006; Mrozek 2007) and critical (see Drache 2006; Philipps 2006; Cayo 2006; Weir 2007) of general income splitting.

The most common popular arguments for general income splitting are twofold. First, splitting would recognize the contribution of — and provide tax relief for — a one-earner couple’s at-home spouse³ who is caring for young children. Second, splitting would apply taxes more equitably between the spouses in two-earner couples who have the same total income but a different division of earnings. An additional, recent argument for general income splitting is that it is unfair to allow senior couples to split their pension income while prohibiting other couples from splitting other types of income. Others have pointed to the inconsistency of applying taxes on an individual basis while using joint income for income security programs.⁴ Thus, the arguments advanced most frequently to promote income splitting fall under the heading of fairness or equity.

These arguments might, at first blush, be appealing, but income splitting also raises intriguing political issues. In 1999, the Reform Party provoked sufficient concern over the asserted “unfair taxation” of one-earner couples to spur a Liberal government to undertake a parliamentary investigation (Canada 1999b).⁵ In 2000, Reform’s successor, the Canadian Alliance Party, advanced a flat tax proposal as a way to resolve the two main equity issues cited above. The Conservative Party’s 2005 inaugural policy declaration committed to income splitting as a way to eliminate inequities between dualand single-income couples, including senior couples with a single pension. Official approval of pension splitting in 2006, driven by the Harper government’s need to placate seniors riled over the imposition of a tax on income trusts, has encouraged advocates to push for broader income splitting. There is no denying the potential electoral appeal of general income splitting. Liberal finance critic John McCallum has conceded that income splitting “would be highly beneficial” to an important segment of voters, and Green Party leader Elizabeth May has endorsed the policy as pro-environmental.⁶ One observer has asserted, “As a green, tax-cutting, family-friendly policy, income splitting has the potential to become the 21st-century equivalent of a chicken in every pot” (Taylor 2007).

While recent events seem to be leading Canada toward the adoption of more general income splitting, there are good reasons to pause for careful scrutiny before proceeding. Income splitting, or some form of joint taxation of couples, is not a minor policy shift. Rather, it is a fundamental change to a basic design element of the direct personal tax system: the definition of the tax unit (the other elements being the tax base and the rate schedule). Income splitting would have a significant effect on the distribution of the tax burden, particularly for many top-income households. Extensive analyses of the behavioural effects and policy implications of alternative definitions of the tax unit are already available, mostly from US and European research and, to a lesser extent, from Canadian studies. Foreign experience with income splitting and other methods of joint taxation can be instructive for the Canadian situation. The issues involved in defining the tax unit are more complex and cross-cutting than most income-splitting advocates realize. Moreover, experience elsewhere suggests that general income splitting would raise even more controversial issues than it would resolve.

In this study, I provide a comprehensive assessment of the appropriate tax unit in Canada, whether the individual or the couple,⁷ with particular attention to income splitting as a special case of joint taxation for couples. My central criterion for assessing these choices is fairness, or equity — the focus of most popular dialogue. I also consider other major policy criteria, such as incentives for paid and unpaid work, marriage and separation; ease and costs of tax compliance and financial planning; economic efficiency; and revenue implications and distributional effects across income groups. I begin by carefully exploring the “horizontal equity” concept needed to give substance to “fairness”; I further consider gender equity and vertical equity and their relation to horizontal equity. I then examine alternative definitions of the tax unit and the associated issues of tax equity between couples and single persons. The heart of my analysis investigates each of three types of income — from pensions, investments and labour earnings — to assess what equity and the other criteria imply for the choice of tax unit and income splitting.⁸

Horizontal Equity: What’s “Fair”?

Assessing the equity of any public policy requires an objective criterion, since “fairness,” like beauty, hinges on the beholder. The horizontal equity concept provides such a standard — one that economists, tax law scholars and policy analysts have long used. Many synonymous formulations exist for horizontal equity, but the most common is “the equal treatment of equally situated individuals” or, in brief, “the equal treatment of equals.”⁹ A more pointed statement of the horizontal equity principle is “that those who are in all relevant senses identical should be treated identically” (Atkinson and Stiglitz 1980, 353; emphasis added). This version emphasizes the rejection of any arbitrary discrimination in public policies, although intentional differentiation in policy treatment (taxes or benefits) based on relevant characteristics might be acceptable. The appeal of the horizontal equity principle stems from a basic notion of fairness. As stated long ago by eminent economist Arthur Pigou, horizontal inequity creates “a sense of being unfairly treated [which is]…in itself an evil” (1949, 50).¹⁰

Horizontal inequity typically is not an intended part of tax legislation, but it is difficult to avoid, given the diversity of taxpayers, the complexity of economic and financial arenas and the multiplicity of tax policy objectives. Moreover, reforms that seek to improve horizontal equity among some “equals” often worsen it relative to other “equals.” For example, assume that individuals A, B and C are equally situated for taxation purposes but various differences among them mean that A pays more tax than B, who pays more tax than C. A prospective policy reform might reduce the horizontal inequity between B and C by reducing the tax on B, but if it cannot address the situation of A, it will simultaneously increase the inequity between A and B. If no feasible policy reform exists that can remove inequities across all these “equal” taxpayers, then whether the partial reform improves or worsens horizontal equity overall hinges on a value judgment.¹¹ This problem becomes manifest in my later analysis of specific income types. Given that tax policy has multiple objectives, any prospective gain in horizontal equity has to be weighed against other effects (Ravallion 2004, 18).

Before I apply the horizontal equity concept, let me dispense with one common misconception. Many advocates argue that income splitting for couples is implied directly from horizontal equity, since only under splitting (or a joint taxation variant) will couples with equal incomes be taxed the same. Yet, as an early analyst of this issue observed, “‘taxing [equally] families with the same incomes’ is not a criterion at all, but merely one of the possible conclusions that may or may not be reached after analyzing the basic principles” (Dulude 1985, 85). That is, taking the couple, rather than the individual — whether a single person or a spouse — as the unit of analysis predisposes the answer as to horizontal equity in the taxation of couples. Analysis of public policies must focus on the well-being of individuals, and the well-being of couples is simply some aggregation of the well-being of the partners.

Making it operational: who’s “equal”?

The most challenging aspect of making the horizontal equity concept operational is to find a standard for gauging when individuals are “equals.” Economists tend to prefer a utility measure of individual well-being, but that poses implementation problems since utility is not observable. What we require is a broad measure of total economic resources or real living standards — something that is both measurable and a good proxy for economic wellbeing. For purposes of the personal income tax, a natural candidate is pretax income: two individuals are thus deemed “equal” if they have the same amount of income before taxes are imposed.¹² However, while the income measure is commonly applied in assessing horizontal equity for income tax policies, it has major deficiencies. For individuals to be equal “in all relevant senses,” they must not only have the same realized incomes but also confront the same costs and opportunities and possess the same preferences.

Income can still serve as a useful starting point for assessing “equals,” so long as we can adjust for factors that vary across individuals in converting money income into real living standards. That is, we need methods to account for the heterogeneity of individuals in key dimensions. Exactly which characteristics should be recognized and which discounted for tax policy purposes entails value judgments that ultimately reflect society’s priorities at a particular time and place. For both pragmatic and ethical reasons, I ignore differences in individual utilities that are either idiosyncratic (such as a miser versus a bon vivant) or arise from personal status (such as the intrinsic utility of being married versus living alone). With appropriate adjustments to pretax incomes to account for relevant factors, a useful metric emerges for assessing when people can be regarded as “equals.” Various tax provisions are employed to refine the use of income as a measure of equals; as one analyst has stated, “[s]urely the aim of the tax-transfer policies whose horizontal equity is being assessed…is precisely to correct at least some of those inequities in the original income distribution?” (Le Grand 1987, 434).¹³

I focus here on three key respects in which individuals can differ: (1) in family size and, hence, in the size of their household budgets; (2) in preferences, in particular for market goods versus home-produced goods and leisure; and (3) in tax-saving opportunities that are open to some but not to others having the same income. Another important respect in which individuals differ is their gender; such differences enter my analysis later as another aspect of horizontal equity.¹⁴

Varying family size and scale economies

Any measure of “equals” has to adjust nominal incomes for factors that affect needs or living costs, the most prominent of which is variation in family size.¹⁵ When can we say that a family of size X with income IX is “equal” to a family of size Y with income IY? If two households differ in size, and thus in the number of members that their income needs to support, they will not be economic “equals” when they have identical nominal incomes. Economists have devised a solution to this problem in the concept of “equivalence scales.”¹⁶ By studying the budgets of households of different sizes, they have derived measures of the relative income needed for families of different sizes to achieve the same real living standard. These measures are stated as adult-equivalence scales, defined as the ratio of income a family of a given size needs to achieve the same real living standard as a single person. Equivalence scales provide useful guidance in setting the relative size of personal exemptions or credits — and the rate schedules — for families of varying sizes.

For two adults living together, the adult-equivalence scale is less than 2, because of the scale economies from such items as rent, utilities and furnishings. That is, two people living together have lower total costs than if the two lived separately. But the adult-equivalence scale for a couple is greater than 1, because scale economies are not unlimited. Typical estimates for the adult-equivalence scale of a couple range between 1.4 and 1.7; accordingly, I use a figure of 1.5, so that, in the jargon, a household of two contains 1.5 “adult-equivalents.” These scale values are used to convert the incomes of families of different sizes into a measure of adult-equivalent incomes. For example, the income of a two-adult household is divided by 1.5, which can be directly compared with a single person’s income. Households of differing sizes but with the same adult-equivalent incomes are taken as “equals.” This issue clearly is relevant to the issue of income splitting, since horizontal equity for taxpayers requires not only fairness among couples but also fairness between couples and singles¹⁷ — a theme I develop later in the study.

Different work preferences

One important, and common, way in which individuals and households can differ is in their relative preferences for work. More specifically, this choice is between paid work and the market goods and services it affords versus home-produced goods and services and leisure time. Two individuals with the same wage rate who allocate their time differently between the labour market and the home will have different levels of income, but they might still be regarded as “equals” insofar as their potential income would be the same if both worked for pay full time.¹⁸ Another way to look at it is that someone who chooses to spend more time at home tends to generate more income in kind than in cash. In its broadest sense, income is the consumption of real goods and services — both marketand home-produced — and leisure. Hence any attempt to measure real living standards across individuals, and thus gauge “equals,” ideally should adjust incomes for differences in working time.¹⁹

In practice, the income tax does not adjust for differences in working time across individuals. That goal could be achieved, but it would require additional reporting on employees’ work hours (or wage rates) as well as their total earnings. Pursuing that route would eliminate the largest economic distortion of income taxation, the bias between work and leisure, which yields disincentives to work. In a related context, one can approach more closely the economic ideal of measuring potential, rather than actual, income. For instance, when comparing one and two-earner couples, we can choose to recognize or we can choose to ignore the value of the additional home-produced goods and services and leisure of one-earner couples. But if we do the latter, we will underassess their real living standards relative to those of two-earner couples. My later analysis shows how this factor enters into evaluating the proper tax unit for couples.

Different tax-avoidance opportunities

Individuals with identical incomes can differ in yet another dimension that is relevant to assessing horizontal equity. Some people are simply better able to avoid paying tax because of their sources of income or their superior knowledge about or access to tax planning. Taxpayers, in fact, vary widely in their circumstances and ability to transfer assets or taxable income and thereby split their income with a spouse. Canada’s complicated tax system has unintentionally created numerous horizontal inequities in the pursuit of other tax policy goals.²⁰ If similar tax-avoidance opportunities were easily and equally accessible to all tax payers, they would not constitute horizontal inequities; it is the differential access to tax avoidance that makes for unfairness.

What should be done about tax-avoidance opportunities that give rise to horizontal inequity? The first-best policy response is to close off the opportunity directly through legislative or regulatory changes, if it can be done at an acceptable cost in terms of tax agency administration and enforcement and taxpayer compliance burdens. If this course of action is technically infeasible, too intrusive into people’s lives or unacceptable at the political level, then it might be possible to enhance horizontal equity by extending comparable tax-reduction opportunities to other groups similarly situated to those who are already avoiding tax. Often, this approach brings the further benefits of a simpler tax system, with cost savings for both administration and compliance. The risk here is that allowing one leaky tax provision might spawn others. The choice of policy response will also hinge, technically and politically, on whether the particular tax-avoidance practice has exceeded some critical threshold.

Gender equity and vertical equity vis-à-vis horizontal equity

Gender equity clearly is a relevant criterion for assessing income-splitting proposals, since splitting involves the tax treatment of women as spouses and cohabitants rather than as single individuals.²¹ Although not conventionally viewed in this way, gender equity can be regarded as analogous to the horizontal equity concept. Horizontal equity entails equal treatment of equals, but it also allows for situations where individuals do not possess the same tastes, needs or opportunities. Similarly, most formulations of gender equity also contain both a formal and a substantive dimension of equality (for official sources, see Canada 1998, 3; 2003, 8-9; 2006a, 13). For formal equality, public policy should treat women the same as men as long as their situations do not differ in “relevant respects.” For substantive equality, policy should recognize the different needs of women and men in overcoming barriers to equality of opportunity, such as institutional, cultural and biological factors that affect women’s educational, occupational and work/domestic/parental opportunities; their employment, training and promotion prospects; and their full equality and individual autonomy in conjugal relationships. That is, differential treatment is sometimes justified to provide women the same realworld opportunities as men to realize their full potential. This is similar to the adjustments for differing needs, tastes and opportunities in applying the conventional horizontal equity concept.

Gender equity is properly concerned with the opportunities and well-being of all women, but many proponents of gender equity stress the barriers to full equality faced by the most vulnerable groups. These include women with lower education, skills and earnings as well as visible minorities, single parents and the disabled. Proponents of gender equity argue that income splitting would do nothing for single women or for those who are part of a lower-income couple, since splitting requires the presence of two partners, at least one of whom earns above the bottom rate bracket ($37,178 for federal tax purposes in 2007). From this perspective, income splitting could never be desirable since it would benefit some married women at higher incomes but do nothing for those who are most needy. However, that is a very limited view of gender equity. A broader conception would have room for policies that augment fairness for women at middle and higher incomes as well as for needier women. No one policy can achieve both goals, and assorted measures in the tax, expenditure and regulatory realms are needed to address gender equity for women in diverse situations.

Vertical equity relates to the distribution of economic resources, the benefits of public spending and the burden of taxes across income classes. The optimal degree of vertical equity reflects both objective evidence on behavioural effects, such as incentives, and value judgments about inequality. Both gender-based analyses and “progressive” policy analyses often accord a higher priority to vertical equity than to horizontal equity. In a logical sense, though, horizontal equity must come prior to vertical equity, since horizontal equity concepts guide the metric for measuring relative real incomes of individuals, without which vertical equity is nonoperational. If a policy that brings greater horizontal equity to individuals at higher incomes also happens to worsen vertical equity, the ideal response is to pursue horizontal equity and to adjust the benefit or tax rate schedule to restore the desired degree of vertical equity. If the latter step is deemed to be politically infeasible, then a conflict can arise between the two types of equity. This conflict can be resolved only by reference to judgments about the prospective gains to fairness in the horizontal equity sense relative to the associated losses to fairness in the vertical equity sense.

“Control” versus “benefit” approaches to equity

Tax law scholars have used two different approaches in assessing the fairness of taxing couples: the “control” approach and the “benefit” approach.²² Under the control approach, the individual who earns the income or has accumulated savings is the one who should bear the tax, so that spouses should be taxed individually, and the income on assets transferred from one spouse to the other should be attributed to the transferor. Under the benefit approach, the tax system would recognize the sharing of material resources within the family, which would justify some form of joint taxation of couples or income splitting. Unlike the benefit approach, the control approach — in common with the gender equity approach — is concerned with which spouse has effective control over the use of a couple’s economic resources.

Both control and benefit approaches, however, are difficult concepts to translate into observable criteria that tax authorities can use in the operation of a real-world tax system (see Head 1996, 201-2). Couples vary widely in their control of resources and in how they share the benefits of their combined resources; I present evidence of these varying patterns later in the study.²³ For now, the key point is that tax policy cannot assume either that a couple’s joint resources are fully shared or that the spouse who earns the income or has legal title to an asset exerts full control without influence by the other spouse. Moreover, the Canadian personal “income” tax is, in fact, much closer to a tax on consumption, since it exempts or taxes lightly most savings and capital income for the overwhelming majority of taxpayers, and therefore is more attuned to a benefit approach to the well-being of couples.²⁴ As a result of these crosscutting considerations, neither the control nor the benefit approach is decisive in the choice of tax unit, even if each can usefully inform the discussion.

Properties of Alternative Tax Units

In addition to the horizontal, vertical and gender concepts of equity, we need to understand the mechanics and properties underlying the choice of tax unit in order to assess income splitting. The alternatives for defining the tax unit for the members of a couple are:

• to tax them as individuals who are not part of a couple;
• to allow them to average their separate incomes and to tax them using the same rate schedule as for individuals, which is “full” income splitting;
• to apply a form of joint taxation that allows splitting but also recognizes the scale economies in living costs that couples enjoy relative to singles; or
• to aggregate their incomes for taxation using the singles’ rate schedule.

Other issues related to the tax unit include combined versus separate filing of tax returns and provisions for the transfer between spouses of various tax benefits.²⁵ Table 1 summarizes the key properties of the various choices of tax unit, which I next examine in detail.

Taxing couples as individuals

The simplest approach is to treat all taxpayers as individuals and subject them to the same tax rate schedule based on their own incomes. This approach regards everyone as an autonomous individual and his or her marital relationship as irrelevant for tax purposes. Individual taxation is the most common approach within member countries of the Organisation for Economic Co-operation and Development (OECD), most of which recognize marital status only to the extent that one spouse’s income falls below the taxable threshold. With a progressive tax rate schedule, individual taxation means that the two spouses’ total tax liability depends not only on their total income but also on how that income is divided between them. Since they are taxed individually, each spouse’s marginal tax rate and related incentives are unaffected by the other’s income level. In contrast, income-splitting and joint taxation systems confront the lower-earning spouse with a higher marginal tax rate based on the couple’s joint earnings, which can pose disincentives for the lower earner (usually the woman) to enter the labour force or work more paid hours. Conversely, individual taxation provides spouses with divergent incomes and marginal tax rates an incentive to shift income to the lower-earning spouse, which the tax system requires special provisions to thwart.

Full income splitting for couples

Alternatively, couples can be taxed as two individuals who fully share their incomes. As tax analyst Joseph Pechman has stated, “[t]he classic argument in favor of income splitting is that husbands and wives usually share their combined income equally…[T]he tax liabilities of married couples should be computed as if they were two single persons with their total income divided equally between them” (1987, 103). Technically, this could be accomplished in either of two ways: (1) by using the same tax rate schedule as for single individuals and allowing the spouses to shift incomes between them until they are equalized; or (2) by providing couples with a special tax rate schedule for their combined income, but with the level of exemptions and all tax rate brackets twice as large as in the individual rate schedule.²⁶ The former approach could apply mandatory splitting or allow the couple to choose how much income to shift notionally for tax purposes. Unless there were interactions with other tax or benefit provisions, the spouses would normally choose to shift at least enough income to put them both in the same marginal tax bracket. This choice would yield them the same tax savings as full splitting — that is, shifting to the point that their taxable incomes were equalized.

These and other properties can be illustrated graphically for general income splitting — where all types of incomes can be shifted or fully split. For simplicity, I assume that the rate schedule for all taxpayers has just two positive marginal rate brackets above an exempt level (E), although the results could easily be generalized to Canada’s federal schedule of four rate brackets. I also assume that the higher-income spouse’s income (I2) is further removed from the boundary between the two tax brackets (B) than the lower-income spouse’s income (I1). Figure 1 shows the respective taxes on the two spouses, T1 and T2, in the absence of splitting.

Their average tax burden (T) arises at their average income (I), at the midpoint of the dotted line connecting the points of their original tax levels.

If the spouses were given the option of splitting by shifting part of their incomes, they clearly would shift income from the higher to the lower earner. To minimize total taxes, the couple would shift at least enough to bring the income of spouse 1 up to I1 andthat of spouse 2 down to I2. The couple’s resulting average taxes would thereby be reduced to T*, with  average savings per spouse of T — T* (the solid vertical line at I). Twice that amount is the couple’s total tax savings relative to the nonsplitting case, which is called a marriage bonus.

Alternatively, a special tax schedule could apply to the couple whereby all brackets are twice as wide as for a single taxpayer. This can be viewed in figure 1 as equivalent to requiring the two spouses to average their incomes and using the singles’ rate schedule. That would yield an average income for each spouse of I and the same total taxes per couple as the spouses would face if they could choose how much income to shift.²⁷ Hence, with full splitting, the results would be the same whether couples were allowed to choose how much income to shift or were required to split their total income equally.

In the illustrated situation, with the higher-income spouse further from the bracket boundary than the lower-income spouse, income splitting raises the lower earner’s marginal tax rate and leaves that of the higher earner unchanged. In the converse situation, splitting could reduce the higher earner’s marginal rate and leave the lower earner’s rate unchanged. If both spouses’ incomes placed them in the same tax bracket, however, there would be no gain from splitting and no effect on the marginal tax rate either spouse would face. With more than two tax brackets, splitting might even increase the lower earner’s marginal rate and decrease the higher earner’s rate. Moreover, with multiple brackets and many couples with substantially divergent incomes, the marginal tax rate on the lower-earning spouse is more likely to increase. In no situation would splitting decrease the lower earner’s marginal tax rate; it could only increase the rate or leave it unchanged.²⁸

Of the two cited methods for income splitting, only the shifting method would be feasible if splitting were limited to certain types of incomes, such as pensions. In the case of the restricted splitting that the tax system now permits, the logic also differs substantially. Now, the benefits of splitting hinge on which spouse has most of the income type that can be split. If that is the lower-income spouse, the couple might gain little benefit from splitting any of the eligible income even if the two spouses are in very different tax brackets; if a 50:50 mandatory split were imposed, the couple would suffer a penalty. And if the income that can be split belongs to the higher-income spouse, splitting might no longer put the two spouses in the same tax bracket if their other types of income are large relative to the income that can be split.

The full-income-splitting approach has several notable attributes. First, as Pechman notes, it assumes that the spouses fully share their individual incomes for common consumption purposes (1987). Second, while treating the couple as a sharing entity, it completely ignores the scale economies of living together. Third, with this approach to taxing couples there is no need to measure or track the separate incomes of the spouses. Fourth, full income splitting yields tax savings only when, and to the degree to which, the spouses’ own incomes place them in different marginal tax brackets under the individual rate schedule. Fifth, this approach to incomes splitting could produce only tax bonuses for marriage (or other recognized forms of union); it could never produce tax penalties. Finally, for most couples who would benefit from full splitting, the lower-earning spouse would face a higher marginal tax rate than would arise under individual taxation, which could exert adverse effects on labour force entry, work hours and economic efficiency. I examine these important issues in more detail later in the paper.

Joint taxation of couples

Joint taxation is a more general form of income splitting that taxes the couple’s combined income using a rate schedule reflecting the scale economies a couple enjoys. The tax rate brackets are less than twice as wide as those for individual taxes — such as the 1.5 ratio suggested earlier as the number of adult-equivalents for a couple. Joint taxation shares all the other attributes described above for full income splitting, with the notable exception that it can produce marriage tax penalties as well as marriage tax bonuses relative to individual-based taxation. If the two spouses have identical or very similar individual incomes, they would gain little or nothing from the splitting feature, but joint taxation’s recognition of their scale economies would mean they would be taxed more heavily than if they were not together. Couples with large tax savings from the splitting feature would have their savings partially offset by the recognition of their scale economies.

Figure 2 illustrates these effects of joint taxation of couples. This time, I distinguish between the tax schedule facing singles (Ts) and a schedule used to assess couples (Tc). Schedule Tc is presented as half the actual schedule for joint married filers and applies to half their combined income.²⁹ Assuming 1.5 adult-equivalents per couple, the exempt level (Ec ) and tax-bracket boundary (Bc ) are each three-quarters their size in the singles’ schedule. Consider a couple in which each spouse has the same income I. Each pays taxes of T“ (on the T schedule), whereas each would pay only T’ (on the T s  schedule) as a single. The difference, T “ — T ’, is the marriage penalty tax for each of the equal-earning spouses; the total marriage penalty for the couple is twice that amount.

Now consider a couple in which the spouses have divergent incomes of I1 and I2. If they were unrelated, their tax burdens (on the Ts schedule) would be T1 and T2, respectively, and their average taxes would be T (at the midpoint, or average income of I, on the dashed line connecting their individual tax liabilities). As a couple, the two individuals could combine or average their incomes using schedule Tc. This would yield a tax of T“ per spouse, which is less than the average taxes of T those individuals would pay if they were unrelated. The resulting tax savings per spouse, T — T“ (the solid vertical line at I), reflects the gains from income splitting offset in part by the scale economies in the joint rate schedule. Twice this amount is the couple’s marriage bonus. If the spouses’ incomes diverged but by not too much, they could still incur a marriage penalty.

To sum up, joint taxation of couples has great similarities to full income splitting; in fact, it is a more general case of splitting that recognizes the scale economies couples enjoy. As a result, joint taxation would provide couples less tax savings than full income splitting, and it would increase taxes on couples with spouses who have identical or not too divergent incomes. Hence, joint taxation produces marriage penalties as well as marriage bonuses. The bottom line in fiscal terms is that joint taxation would be considerably less costly than full splitting. If splitting were restricted to particular types of income, joint taxation would not be appropriate since it would apply a modified tax schedule to the couple’s entire income from all sources. Even with restricted splitting, however, one could contemplate introducing a device to account for the couple’s scale economies with respect to the income that can be split.

Full aggregation of couples’ incomes

Full income splitting is one polar case of the joint taxation of couples, in which couples are assumed to enjoy no scale economies relative to singles. At the other extreme is full aggregation of couples’ incomes using the singles’ tax rate schedule, which assumes that couples’ economies are infinite. That is, full aggregation assumes that “two can live as cheaply as one,” in effect deeming a couple to contain only 1.0 adult-equivalents. While full aggregation might not have much appeal today, some European countries once used this system (Ireland, before 1980; the United Kingdom, at least until 1971), and it was proposed as a reform to the US system in the 1940s, before that country’s move to full income splitting in 1948. Full splitting can produce only marriage bonuses, while full aggregation can produce only marriage penalties. All systems for taxing couples other than individual taxation can produce disincentives for paid work by raising tax burdens on labour force entry and extra earnings by the second earner, most often the wife.

Combined or separate filing?

Whether couples should be taxed on an individual, fully split, joint or fully aggregated basis is a logically distinct issue from whether spouses should be required to submit separate or combined tax returns. Some Canadian advocates of income splitting contemplate continued individual filing of returns but would give spouses the option to shift part of their income from the higher to the lower earner. More typically, countries that tax couples on a fully split or joint basis require the filing of a joint return that combines the reporting of the spouses’ income receipts. Common jargon for this practice is “joint filing,” but here I wish to distinguish clearly between forms of joint taxation (including the full-splitting and aggregation cases) and combined or separate filing of returns.

With joint taxation, several considerations arise in the policy choice between combined and separate filing. With combined filing

• spouses would no longer need to identify who received the income and thus couples’ financial lives would be simplified;
• each spouse would become jointly liable for payment of the other spouse’s outstanding taxes;
• wives’ financial knowledge might diminish, disadvantaging them if they became divorced or widowed, although it could work the other way if the spouses handled the tax return jointly; and
• if applied only to federal tax returns, it would be onerous for a province to retain the individual taxation of spouses, since different tax returns would be required for the two levels of government.

On the other hand, separate filing

• would afford each spouse’s financial affairs greater privacy and autonomy, although they would still have to share information on their taxable incomes to optimize their income shifting;
• could, with the option to shift income notionally, make one spouse liable for taxes on income he or she does not control; and
• would, with the option to shift income, give married taxpayers more flexibility in minimizing their taxes and maximizing their net benefits for tax and transfer provisions that have clawbacks based on individual rather than joint income.

Similar considerations also affect the choice between individual taxation, which typically entails separate filing by spouses,³⁰ and various forms of joint taxation.

Transferability provisions

Almost all systems for the individual taxation of couples provide for the transfer of basic tax exemptions, allowances, deductions and/or credits. Typically, this arises when the lower-earning spouse cannot fully use the provision to offset his or her own taxes and transfers the unused portion to reduce the tax liability of the higher-earning spouse. Such transferability constitutes a form of “jointness” in the taxation of couples but differs considerably from the ability to transfer or shift income to the spouse.³¹ In small measure, transferability recognizes the additional living costs of a couple relative to those of a single person, but it affords relatively little relief compared with most variants of joint taxation for one-earner couples, especially at median and higher incomes. Nevertheless, these provisions do engage issues arising for the choice of the tax unit, which I later discuss for Canada’s spousal and equivalent tax credit.

Practices in the United States and Other OECD Countries

In this section, I provide an overview of past and present practices in the tax treatment of couples in the United States — the most important tax comparator for Canada and a country where the taxunit issue has been extensively debated and assessed. I also briefly describe the diverse methods used for taxing couples in Ireland, the United Kingdom, Belgium, Spain and France. This institutional material offers valuable background for understanding my subsequent analysis of the equity and other properties of tax splitting for the three types of income. Almost all the issues for the tax treatment of couples in Canada have been confronted and addressed, one way or another, in foreign policies.

The tax unit in the United States

The United States offers an essential case study for the taxation of married couples. From its inception in 1913, the US income tax has traversed a tortuous course from the original system of individual taxation to the adoption of full income splitting in 1948 and subsequent reforms that reduced splitting to joint taxation and later restored full splitting only for low to middle-income earners.³² Here, I deal solely with the rate schedule and income-splitting aspects of the tax unit; I do not address other US tax provisions³³ or transfer programs³⁴ that also affect marriage penalties. The US experience is rich in both legal and economic studies on the tax treatment of couples, from which valuable policy guidance can be drawn. Even more instructive is the political lesson that initiating income splitting opens a Pandora’s box of interminable controversy over the equitable tax treatment of various family types.

At the outset, all US taxpayers, whether single or married, filed an individual return using a common rate schedule. Couples also had the option of filing a combined return using the full-aggregation method with the same rate schedule as singles; this allowed one-earner couples to access the personal exemption for both spouses. With the rising tax rates of the First World War, however, some couples residing in the eight community-property states of the Southwest and the Pacific Coast began filing separate returns in which each spouse claimed half the couple’s combined income. This “self-help” income splitting was justified by those states’ legal provisions deeming both labour earnings and investment income vested half in each spouse. The practice was challenged by the Congress and the Treasury Department but ultimately was upheld by the Supreme Court in 1930. Couples in the common-law states also sought to achieve income splitting by placing income-producing properties in joint tenancies or family trusts and by converting business proprietorships into corporations or family partnerships.³⁵ Those manoeuvres often did not succeed legally, so that from 1939 to 1947 several of those states switched to community-property regimes to assist their residents in reducing their federal tax burdens. Since the gains to splitting for one-earner couples, the norm in that period, rose with tax rates, the mounting tax revenues needed to finance the Second World War compounded all of these pressures.³⁶

Members of Congress were hard-pressed to respond to the growing horizontal inequities in tax burdens for couples, both among states with different property regimes and between couples with income from labour earnings as opposed to income from property and investments. In 1941, in an attempt to override the community-property states, the Treasury Department persuaded the House Committee on Ways and Means to advance a bill for mandatory joint filing with full aggregation by all married couples, using the same rate schedule as for single persons. This bill was attacked as “a tax on morality” for encouraging unmarried cohabitation and divorce and failed to gain political support.³⁷ The ultimate resolution, in 1948, was a reform that permitted all married couples to file a joint return with full income splitting, in effect validating what the shift to community-property regimes was achieving.³⁸ Congress passed this “far-reaching change in the income taxation of the family” and sustained it over a presidential veto (Surrey 1948, 1105). Married taxpayers filing a joint return computed the tax on one-half of their combined income, using the same rate schedule as for individual filers, and this figure was doubled to obtain their total tax liability.³⁹ As my earlier analysis of full income splitting showed, this system could create only marriage tax bonuses, and bonuses were far more common than no changes in tax liability because of the rarity of equal-earning spouses at that time.

No sooner had the United States introduced full income splitting than critics observed the new system’s tax penalties for single individuals. By its nature, full splitting implied that a single individual would pay substantially more tax than a married couple with the same income; this point was most salient when the couple’s income derived entirely from one earner. The first legislative response addressed the situation of unmarried individuals who had financial responsibility for dependants by using the same treatment as that accorded to a married individual with a dependent spouse. In 1951, a new tax rate schedule was introduced for “heads of households,” a category that initially covered single taxpayers who were supporting one or more dependent children; the schedule was extended in 1954 to include support of dependent parents. The head-of-household rate schedule roughly split the difference between the tax burden on single individuals and that on married couples at the same income levels.

With time, criticism also arose over the tax penalties for single individuals without any dependants. This included the growing numbers of unattached individuals plus unmarried cohabiting couples who in most states lacked the status of a common-law marriage.⁴⁰ These individuals could face a tax burden as much as 42 percent higher than that of a married couple with the same income. Congress eventually responded, in 1969, with another major rejuggling of the tax system that increased the number of rate schedules from two to four. Becoming effective in 1971, the new provisions reduced the basic tax rate schedule used by all filers other than heads of households so that the singles’ tax penalty could never exceed 20 percent of taxes paid by a married couple with the same income.⁴¹ Simultaneously, it reformulated the existing rate schedule by doubling the tax-bracket widths for use solely by married couples and applying them to the spouses’ combined income; this simplified their tax computations but left their tax burden unchanged. Married couples could not use the new, lower singles’ rate schedule but could opt for filing separately under the original but less advantageous rate schedule. In that way, Congress forestalled the ability of couples in community-property states to access the singles’ lower rate schedule and secure differential tax relief compared with couples elsewhere.

By preventing couples from using the singles’ rate schedule, the 1969 reform both reduced marriage bonuses and, for spouses with incomes not too divergent, created marriage penalties. Two-earner couples faced a marriage penalty when their incomes were more evenly divided than 20:80.⁴² Given the predominance of one-earner couples, however, marriage bonuses were much more frequent and in aggregate much larger than marriage penalties during this period. Even 10 years after the reform, the Treasury Department reported that marriage penalties constituted US$8.3 billion on 16 million joint returns, while marriage bonuses constituted US$19 billion on 24 million joint returns (Gann 1980, 22-3). In the ensuing years, with the ongoing rapid rise in married women’s labour force participation and growth in their relative earnings, the balance was destined to tilt much more toward marriage penalties and away from marriage bonuses. The measurement and assessment of marriage tax penalties and bonuses became a veritable cottage industry for US tax policy analysts.⁴³ Additionally, the adverse effects of joint taxation on wives’ work incentives attracted increasing scrutiny.

To address both marital bias and work incentive issues, a bill was introduced in Congress in 1974 to revert to individual taxation with a single rate schedule for all taxpayers. Despite its being sponsored by a politically diverse group of more than 150 congressmen and senators, the bill failed passage. In 1981, a provision for a second-earner deduction was approved to permit a married couple’s lower earner to deduct 10 percent of earnings up to US$30,000 per year beginning in 1983 (5 percent in 1982). The second-earner deduction provided substantial but incomplete moderation of both marriage penalties and work disincentives for two-earner married couples.⁴⁴

Although it enlarged marriage bonuses, it left the wife’s marginal tax rate tied to her husband’s income and did not relieve marriage penalties resulting from nonlabour income. The deduction was repealed in the sweeping 1986 tax reform, whose sharp rate cutting and bracket broadening directly moderated those problems but did not eliminate them.

Marriage penalties and work disincentives for married women grew again in 1993 with the Clinton administration’s income tax hikes for higher earners. These issues continued to attract political heat through the decade and, in the 2000 election campaign, the Republicans promised to address them. As of the 2000 tax year, the married joint rate schedule had an adult-equivalence scale for couples of 1.67 for the bottom two tax brackets, applying up to a combined income of US$105,950. For couples with higher incomes, the brackets’ implied adult-equivalent values were much smaller, so that the single and married joint rate schedules both reached the top marginal rate of 39.6 percent at an income of US$288,350. With the Bush administration’s tax cuts of 2001 and 2003, the married joint brackets were expanded to restore full splitting for low-to upper-middle-income couples; as of 2007, couples enjoy full income splitting for combined income up to US$128,500.⁴⁵ For couples with higher income levels, marriage penalties still exist. Doubtless, the cycle will turn once again, with pressures for reform likely centring on the growing singles’ tax penalty. As one prescient observer, an advocate of individual taxation, remarked in 1980, “legislative attempts to determine the appropriate relative tax burdens of single and married persons will continue to be what they have been: temporary, uneasy compromises that must yield to ever-changing legislative modifications passed in response to taxpayers’ complaints” (Gann 1980, 3).⁴⁶

Earlier analysts of the tax unit in the United States formed a near-consensus on the propriety of using the couple or family. In 1947, noted economist and future Nobel laureate William Vickrey wrote: “it is neither possible nor, in fact, desirable to attempt to consider each individual as an independent unit for tax purposes” (274). This view was later echoed by Groves, who stated that “[s]tudents of public finance with few exceptions regard the family rather than the individual as the proper unit for income taxation” (1963, 62). In an influential 1975 article, tax law scholar Boris Bittker acknowledged the advantages of separate filing by spouses but was unwilling to forego the principle of equal taxes for couples with equal total incomes.

Scholarly thinking on this issue has changed sharply over time, driven by major shifts in US demographic and economic patterns — rising labour force participation rates of married women, the increasing incidence of divorce and growing numbers of unmarried couples, single parents and other nontraditional family units. In 1977, economist Harvey Rosen stated that “if joint filing were eliminated, the federal income tax would become both more equitable and more efficient” (423). In subsequent years, support for individual taxation in the United States has swelled across a broad range of analysts.⁴⁷ Even conservative analysts, who tend to be most favourable to using the couple as the tax unit, have identified deficiencies in the existing US tax treatment of couples. Their solutions include adopting a flat rate income tax, replacing the income tax with a national sales tax and giving spouses the option of filing individual tax returns or joint returns (Bartlett 1998; Strassel, Colgan and Goodman 2006).⁴⁸

The tax unit in other OECD countries

The tax systems of the 30 member countries of the OECD are the most comprehensively documented in the world. Table 2, which groups those systems according to how they define the tax unit with respect to employment income,⁴⁹ clearly shows that individual taxation of couples is the predominant approach. Moreover, the trend since 1970 has been to move from various forms of joint taxation to individual taxation, as has occurred in Austria, Belgium, Denmark, Finland, Italy, the Netherlands, Spain, Sweden and the United Kingdom (OECD 1993). Ireland’s tax system, while still classified as joint, has also moved toward lesser jointness in the taxation of couples. Gender equity has been an important policy consideration in many countries that have moved closer to individual taxation.⁵⁰

Several OECD countries that apply individual taxation to income from employment use special forms of joint taxation for self-employment and/or investment income. In the Netherlands (and previously Belgium), investment income is assessed to the spouse with the higher earned income. The Netherlands and Belgium also permit a portion of self-employment income to be attributed to a “helping spouse.” Denmark attributes investment income over a specified threshold as taxable income to the higher-earning spouse (O’Donoghue and Sutherland 1999, 574). In 1991, Sweden adopted a dual tax that combines progressive rates on labour income with a flat rate of tax (equal to the corporate income tax rate) on investment income; Norway and Finland adopted similar schemes in the following two years (Sørensen 2007, 563).⁵¹ Since moving from joint to individual taxation in 1990, the United Kingdom has allowed couples to shift investment income between spouses to minimize their tax burdens, as I detail later.

The tax unit in Ireland

Since its inception, Ireland’s income tax system has applied three different methods of taxing couples (see Callan 2006; OECD 2007). From 1922 to 1979, the method was to aggregate spouses’ incomes but to offer no splitting benefits in the rate schedule — the same as that used for single taxpayers — although there was a “married man’s allowance.” A legal case in the 1970s attacked the resulting “marriage penalty” for two-earner couples on constitutional grounds and led to the adoption of full income splitting for couples in 1980. Over time, income splitting came to be seen as providing large benefits to many families not engaged in child care and no benefits to many families with heavy child care burdens. As analysts noted, full income splitting was an inefficient way to target fiscal resources at child care undertaken by married women in the home. In 2000, the system was modified toward a more “individualized” system with limited splitting; this greatly reduced the extent to which the benefits of the standard tax rate bracket were transferable when one spouse had low or no earnings.⁵²

The tax unit in the United Kingdom

At one time, the United Kingdom used the aggregation method, under which the wife’s income was added to the husband’s and taxed on his return.⁵³ There was also a “married man’s allowance” and a “wife’s earned income allowance,” the latter similar to the second-earner deduction once provided in the United States. In 1971, the UK introduced a “wife’s earnings election” that allowed a married woman to be taxed as a single person on her own labour earnings, but both spouses had to choose this option and the husband had to forego the married man’s allowance. In 1990 came a major switch to the individual taxation of couples, offering great flexibility in splitting investment income by transfers of assets between spouses. Such transfers could be made either as a gift (which incurs no capital gains tax) or by holding assets jointly with an option to divide the asset holding and associated income flow in any desired proportions.⁵⁴

The tax unit in Belgium

Belgium’s system is essentially the individual taxation of married couples (although they file joint returns), an approach that reflects reforms implemented between 2002 and 2004.⁵⁵ Prior to 1988, Belgium used the full-aggregation method for married couples. Dual-earner unmarried couples were not required to aggregate their incomes, so that the system created a steep marriage penalty for dual-earner couples. In 1988, the system changed to individual taxation, with a special provision for lower-earning spouses with earnings less than 30 percent of the couple’s combined earned income; this provision also applies to one-earner couples. This “nonearning spouse allowance” allows a notional transfer to the lower earner of up to 30 percent of the couple’s aggregate earned income, less the lower earner’s earnings and in 2006 subject to a 8,570 limit. This limited provision for splitting is akin to other countries’ provisions for transferability of a spousal allowance.

The tax unit in Spain

In Spain, the basic unit for taxation is the individual. Married couples can opt for either separate filing by spouses as individuals or joint filing as a couple on their combined income. Currently, married couples who file jointly face the same rate schedule as single filers, so that their only gain is the exempt amount for a nonearner spouse, similar to Canada’s provision for transferability of unused marital credits. In earlier years, however, many couples found joint taxation more attractive because it offered a separate tax rate schedule with partial income splitting.⁵⁶ The previous scheme constituted a variant of taxation of couples that differs from both full splitting (where only marriage bonuses can arise) and conventional joint taxation (where both marriage bonuses and marriage penalties arise). Spain’s earlier scheme allowed only the limited splitting of joint taxation but also avoided marriage penalties.

The tax unit in France

France offers an interesting variant on income splitting whereby the number of children as well as  adults affects the splitting calculation.⁵⁷ Its “family quotient” (quotient familial) system divides a family’s total taxable income by a quotient that depends on marital status and the number of children.⁵⁸ This “split” income is then used to calculate tax with a standard rate schedule, and the family’s total tax is the computed tax multiplied by its quotient. The quotient for couples is 2 (1 for each adult) plus 0.5 for each of the first two children and an additional 1 for each of the third and further children. For a single-parent family, the first child obtains a quotient of 1. The maximum tax reduction from the child part of the quotient in a two-parent family is limited; in 2005, the limit was 2,159 per half-quotient.

Canada’s Tax-Unit Practices

From its inception as the Income War Tax Act of 1917, Canada’s income tax system has always used the individual as the tax unit, with the same rate schedule for everyone, whether single or married. Initially, the finance minister of the day proposed to give all individuals the same personal exemption level irrespective of marital status or presence of a dependent spouse. This proposal was attacked in parliamentary debates as an unfair benefit to single men and “spinsters,” who would be taxed relatively “too lightly” (Lahey 2000, 1-2). In the end, the 1917 legislation provided tax exemptions for one-earner couples that were double those for single filers. A variant of this tax provision offering linkage between spouses has survived to this day — as a marital exemption through 1987 and since 1988 as a nonrefundable spousal tax credit. This provision, which operates only when one spouse has income below the taxable level, is now called the “spouse and equivalent-to-spouse tax credit” and is also available to single parents.

An interesting Canadian contrast to the US experience with community-property states arose in Quebec, which once had matrimonial property laws similarly derived from French law (Dulude 1985, 82). Under the standard Quebec regime, the spouses were nominally equal partners in their combined property, even though the husband typically wielded effective control. However, high-income Quebecers usually opted out of the community-property regime through prenuptial agreements, so that attempts at income splitting for tax purposes were not common. Then, in 1957, the Supreme Court of Canada ruled in the Sura case that spouses under the community-property regime could not each declare half of their combined income for tax purposes, arguing that the wife did not “have the exercise of the plenitude of rights which ownership normally confers…[T]he result is that the wife receives no income from community property,” so that the income could not be split to reduce the couple’s total tax burden.⁵⁹

The most focused early assessment of Canada’s taxation of couples was that of the Carter Commission (Canada 1966). The commission assumed that the appropriate unit for taxation was the family, not the individual, and it cited the administrative and compliance gains from not having to attribute investment incomes to each spouse in order to contain tax-motivated shifting. It recommended a system of joint taxation of couples that allowed for both income splitting and recognition of the scale economies enjoyed by families. The commission’s proposed system entailed a different rate schedule for married couples than for individuals. The government of the day rejected this component of the commission’s report, however, on the grounds that it imposed a tax on marriage and a barrier to the wife’s participation in the paid workforce by confronting her with the husband’s marginal tax rate (Canada 1969, 15).⁶⁰ Those concerns have recurred in subsequent research on income splitting, particularly with respect to labour income. Joint taxation under an elective scheme was supported by Canada’s Royal Commission on the Status of Women in 1970, but was officially rejected.

Further interest in joint taxation for couples in Canada arose in the mid-1970s with the federal Interdepartmental Committee on the Taxation of Women.⁶¹ Some committee members, led by those from the Department of National Revenue, supported joint taxation on the grounds of equity and administrative simplicity; other members opposed it on the grounds of its marriage penalty and work disincentives for wives. The latter group predominated, its views captured by the federal coordinator for the status of women:

Joint returns are an idea whose time has passed. During the 1950s and ’60s when most wives worked in the home, it would perhaps have been a better tax system than individual returns. In the 1970s, however, when government is making such efforts to improve the status of women by recognizing them as individuals outside the family unit, it would seem a retrograde step. (Quoted in Dulude 1985, 84-5)

The next hint of a policy change in the taxation of couples came in 1983, when the Minister Responsible for the Status of Women proposed to abolish the spousal exemption and use the funds to enhance child care — putatively the first step toward joint filing. Public backlash was quick and severe, suggesting concerns that the proposal failed to value women’s nonmarket work (Lahey 2000, 43-4). Then, after a relatively dormant period, political and scholarly interest in the issue of how to tax couples was renewed in the latter 1990s.⁶² Research published by Canadian think tanks from a wide range of perspectives has almost uniformly found individual taxation to be most appropriate.⁶³ Outliers in this body of research are studies by the Fraser Institute; for example, Veldhuis and Clemens assert that the Canadian “tax system is biased…against single income families,” but their preferred solution is a flat tax rather than income splitting (2004, 11-12).⁶⁴ Recent popular support for general income splitting has come predominantly from “family values” and faith-based advocacy groups.⁶⁵

Opposite-sex common-law partners and same-sex partners cohabiting for more than one year have been treated the same as married couples for tax purposes since 1993 and 2000, respectively. In addition to the spousal or equivalent credit, tax credits for charitable donations and medical expenses can be amalgamated between spouses in a way that is often favourable relative to single taxpayers (because thresholds for credits or credits at a higher rate need to be satisfied only once for the spouses jointly). Various nonrefundable tax credits not used by a nontaxable spouse can be transferred for use by the taxable spouse. Canada’s tax system also contains provisions that can operate against couples, such as the exemption of capital gains on a single principal residence for couples. Indeed, Lahey reports that the tax system has “some 100 provisions” in which marital status affects tax liabilities (2005, 33). The system seeks to apply tax to the individual incomes of spouses with rules to restrict income splitting, although sizable options for splitting exist, especially for savings and investment income, as I discuss later.

In the 2006 tax year, the spousal tax credit was a maximum of $7,505, while the basic personal credit was $8,839, which each filer with income above the taxable threshold could claim.⁶⁶ A low-earning spouse generated a spousal credit amount for the higherearning spouse equal to $8,256, less his or her net income up to a maximum of $7,505.⁶⁷ In effect, the lower earner could earn an initial $751 (that is, $8,256 minus $7,505) tax-free; any incremental earnings were deducted from the maximum amount and therefore faced the bottom marginal tax rate. This system for taxing couples, which is common in some variant to most countries that use individual taxation, poses a barrier to labour force entry and market earnings by a couple’s second worker, most often the wife. If spouses were taxed in a truly independent manner, each would be able to earn the full basic amount before facing any positive tax rate. However, the cited barrier for second earners is less severe than that arising under joint taxation; there, the initial earnings of the lower-earner would face the higher marginal tax rate of the primary earner.

For the 2007 tax year, the basic personal credit amount was indexed to rise to $8,929 and the 2007 federal budget raised the spousal tax credit to equal the basic personal credit amount. The finance minister asserted that he was “ending the marriage penalty for single-earner couples” by equating the two credit amounts (Canada 2007a, 13); most media coverage repeated the claim uncritically. Yet, the spousal credit provides a benefit to a couple for having a low or nil earner that would not arise if they were not married, since the credit is nonrefundable and can reduce the amount of tax owed only by the primary earner. Thus, it would be more correct to state that the budgetary initiative was “increasing the marriage bonus” for single-earner couples; it would be even more accurate to refer to the “marriage and cohabitation” bonus. With the raising of the spousal credit to the basic credit amount, all earnings of the lowerearning spouse now offset the spousal credit. That change has removed the tax-free range of initial earnings by the second earner, so that even the first dollar of earnings faces the bottom marginal tax rate. One could argue, however, that a lower spousal credit than basic credit is justified, given the scale economies that a couple enjoys relative to an individual. The fall 2007 federal economic statement boosted the basic and spousal credit amounts to $9,600 retroactively for the 2007 tax year and for 2008 (Canada 2007c).

Income-splitting behaviour and attribution rules

Canada’s system of individual taxation aims to assess separately all the economic resources of each filer to ensure the application of progressive rates. Typically, labour earnings of employees (the largest income source) are not susceptible to splitting with the earner’s spouse for tax purposes. Nevertheless, for incomes derived from sources other than employment, significant opportunities for splitting exist within couples. Two types of labour earnings are prone to splitting. First, individuals with unincorporated self-employment income can hire their spouses and thereby divert some of their earnings for tax purposes, although the tax regulations require that such payments be reasonable relative to the work performed by the spouse and at a rate that would be made to an arm’s-length employee.⁶⁸ Second, individuals with control over an incorporated business can similarly employ their spouse or alternatively divert their own returns to labour within the enterprise via dividend payments to their spouse as a shareholder. The tax authorities can contain only the more blatant abuses of excessive payments to spouses.

Greater income-splitting opportunities arise for Canadian couples in the areas of savings and associated investment income. Some of the more common methods are:⁶⁹

• the contribution of the higher-earning spouse to a spousal Registered Retirement Savings Plan (RRSP) to reduce the couple’s joint tax liability during retirement, or even sooner since withdrawals are allowed after three years from the deposit date;
• the assumption by the higher-earning spouse of all or most household expenses, including mortgage payments (since the ultimate capital gain on the home will be tax-free), leaving the lower-earning spouse to undertake all of the couple’s non-registered savings, with subsequent investment income taxed at the lower rate;⁷⁰
• the use of interspousal loans to finance the acquisition of assets that generate higher returns for the borrowing spouse than the interest paid to the lending spouse;
• the ability of spouses to split their combined Canada Pension Plan or Quebec Pension Plan (CPP/QPP) benefits;
• the use of a spousal RRSP to finance a Home Buyer’s Plan withdrawal (Steele 2007); and
• the use of spousal testamentary trusts to enable a bequest to be taxed separately, which thus allows splitting even after death.

All of these methods are fully legal and can yield large splitting benefits if pursued over an extended period. The latest addition to this list is the 2007 budget’s provision of pension income splitting for couples, which I describe in detail later.

Higher-wealth couples with different incomes often seek faster and larger income splitting through interspousal transfers of income-producing assets.

From its inception, Canadian income tax legislation has contained rules to attempt to thwart such splitting. These income attribution rules, which deem the income taxable to the transferor of the asset rather than to the transferee, have become more complex over the years through the accretion of legislative changes and court rulings. Nevertheless, one generation ago, an expert remarked, “[a]ny person with income from business or property, notwithstanding the attribution rules, is able to arrange his or her affairs so as to income split with family members” (London 1979, 7). Despite subsequent legislative tightening, most current observers would still find much truth in this statement. For example, income attribution rules are constrained in their application to the transfer of business interests as against financial assets. Among countries that use individual taxation, Canada is almost unique in the extent to which it seeks to curtail income splitting via asset transfers. Many analysts — including Donnelly, Magee and Young (2000); Young (2000); Philipps (2002); and Samtani (2006) — have critiqued the income attribution rules but with varying conclusions. I address this issue in my analyses of splitting for pension income and investment income.

Pension Income Splitting

As part of its Tax Fairness Plan announced in October 2006, the federal government allowed couples to split their pension-type incomes beginning with the 2007 tax year. Incomes eligible for the pension income credit can be split, but the precise types depend upon whether the transferor has attained age 65. Annuity payments from a Registered Pension Plan (RPP) can be split regardless of the transferor’s age. A spouse age 65 or over can also split annuity payments from an RRSP or Deferred Profit-Sharing Plan and payments from a Registered Retirement Income Fund (RRIF), locked-in RRIF (LRIF) or Life Income Fund (LIF).⁷¹ There is no age restriction for the spouse who receives the pension income allocation. A couple can jointly elect for a spouse to split up to one-half of his or her qualifying pension income with the other spouse; typically, these notional transfers will be from the higher-income to the lower-income spouse. Since nine provinces operate their income taxes using the federal definition of taxable income, pension income splitting will apply to them as well. While Quebec runs its tax system independently, it has introduced pension income splitting to parallel the federal provision.

At tax-filing time each year, a couple will have the option to engage in pension splitting. The chosen sum will be deductible to the spouse who makes the transfer and taxable to the other spouse. Therefore, both spouses must agree to the transfer. The amount transferred retains its character as income eligible for the pension income credit, so that where one spouse has little or no pension income the couple will now be able to tap the pension credit twice. Income splitting will usually be beneficial only when the spouse with the pension income to be split has larger taxable income and is in a higher marginal tax rate bracket. However, the split income can affect each spouse’s claim for the age tax credit and clawback of Old Age Security (OAS) payments, each of which is conditioned on the net income of the individual, not the couple.⁷² Therefore, many possible scenarios for splitting choices can arise depending on which spouse has more pension income, on the amount of each spouse’s nonpension income and on interactions with other provisions of the tax and benefit system. Some couples may even choose to transfer pension income from the lower-income to the higher-income spouse, where both were originally in the OAS clawback’s income range and one can be pushed above that range.⁷³ Although the calculations are complex, taxpayers will easily optimize their choice at filing time with tax preparation software.

The horizontal equity concept has been central in support for pension income splitting. Prior to the Tax Fairness Plan, the advocacy group Canadian Activists for Pension Splitting⁷⁴ argued:

We consider it unfair that some taxpayers are able to reduce their taxes by such means as spousal RRSPs or CPP splitting, while others do not have these available. Allowing splitting of all pensions would eliminate that unfairness. Our request is not merely another plea for a tax break. We consider this to be an issue of fairness, especially with respect to horizontal equity. (2006)

Spousal RRSPs are a major way for one-earner couples and spouses with divergent earnings to engage in tax splitting. Yet some workers — those with employers offering generous RPPs — have little or no access to RRSPs of any kind.⁷⁵ In contrast, other workers with similar earnings — the self-employed, business proprietors and employees with little or no RPP savings through their employer — have large opportunities to split incomes via spousal RRSPs. That discrepancy clearly gives rise to horizontal inequity that could justify pension income splitting. The ability to have CPP/QPP retirement benefits split between spouses, in contrast, does not create horizontal inequity because those plans cover all kinds of employment, regardless of their RPP coverage, as well as self-employment.⁷⁶

The new pension–income-splitting provisions will reduce substantially the incentives for using spousal RRSPs to split incomes. Some observers have even suggested that the spousal RRSP has become redundant. Yet, some circumstances will still favour couples who can access spousal RRSPs, such as:

• contributions can be withdrawn after three years without attributing the taxable income back to the contributor, thus allowing short-term splitting;
• spousal RRSPs are more flexible for splitting by workers choosing early retirement, since pension income splitting is allowed for RRSP proceeds only at ages 65 and above; and
• spousal RRSPs also provide more flexibility, since pension income splitting does not allow full equalization of spouses’ taxable incomes and marginal tax rates where the spouse with the higher pension income also has more income from sources that cannot be split; indeed, in this situation, the use of a spousal RRSP allows couples to exceed the splitting provision’s 50-percent limit.

To correct the remaining horizontal inequity, several measures could be considered. A minimal policy would be to attribute back to the contributor any withdrawals from spousal RRSPs until both spouses are at least age 60; this would eliminate the shortterm splitting advantage. A stronger policy would be to disallow any further contributions to spousal RRSPs.⁷⁷ Yet, dismantling existing spousal RRSPs and merging them with the contributor’s own RRSP would be politically difficult, and many would deem it unfair in view of savers’ long planning horizons.

The design of the pension-income-splitting provision raises questions of fairness. It is not a full or mandatory splitting scheme, in which each couple must equalize their combined pension income for tax purposes under what remains, essentially, a system of individual taxation. Rather, each couple can choose whether to split their pension income and how much to split, up to 50 percent of one spouse’s pension income. If the spouse receiving the notional transfer also has pension income, he or she can end up with more than 50 percent of the couple’s combined pension income.⁷⁸ Moreover, the optional nature of the splitting provision allows couples to “game” other tax and benefit provisions, most importantly their combined OAS clawback. It is unknown whether these properties were intended for the pension-splitting provision or the result of a hasty packaging of the Tax Fairness Plan. Revisions to the pension-splitting provision could address these problems. For example, couples wishing to use the provision could be required to split their combined qualifying pension incomes 50:50, which would reduce the opportunity for tax gaming. Alternatively, the OAS clawback on each spouse could be computed ignoring any pension splits.

Pension income splitting raises further issues of horizontal equity. As discussed earlier, splitting assumes that the couple is the appropriate tax unit. In that case, family-equivalence scales should be used to reflect scale economies of couples relative to singles. Adjusting tax rate brackets is easily done with general income splitting, but it becomes complex when a single income type is eligible for splitting. Hence, one needs to balance the gain in horizontal equity across pensioner couples against the loss in horizontal equity in comparing pensioner couples with pensioner singles.⁷⁹ Many observers would rate the pension-splitting provisions as a net gain for horizontal equity. Others might choose to constrain the allowable amount of pension income splitting to a ceiling of, say, $30,000 annually per couple; this reform would also limit the adverse effects of pension income splitting on vertical equity.

The effects of pension income splitting

Insofar as it reduces the attractions of spousal RRSPs, the pension-income-splitting provision will reduce the role for tax planning in couples’ savings decisions. Decreased expenditures on tax and financial advisors would constitute one form of efficiency gain from pension splitting. As for the effects on aggregate savings, however, the outcome is ambiguous. Couples who benefit from pension splitting will save considerable sums in federal and provincial taxes — in some cases, thousands of dollars per year. Thus, couples that are saving to meet target levels of retirement living standards could undertake less saving over their working years. Conversely, the lower effective tax rates that splitting affords them on consumption of lifetime savings in retirement provide an incentive for consuming less while working and saving more. The net effect on savings behaviour cannot be predicted without proper empirical analysis. Yet, if pension splitting increases the gap between marginal tax rates at the point of saving and the point of dissaving, it decreases the efficient lifetime allocation of couples’ consumption.⁸⁰

The federal government estimates the revenue cost of pension income splitting to be $675 million in 2007/08, the first full fiscal year of the provision’s operation (Canada 2007b). Provincial revenue costs will be an additional fraction of that figure. A primary critique of pension income splitting centres on the distribution of the tax relief (see Tomagno and Battle 2006; Woolley 2007). The benefits in terms of tax savings are heavily skewed toward upper-income couples, particularly where the two spouses’ lifetime earnings and registered savings have been divergent.⁸¹ For example, the splitting provision will save a senior couple with a single pension income of $100,000 about $7,300 per year in federal taxes, of which more than half stems from the reduced clawback of OAS benefits (Tomagno and Battle 2006, 4). Here, we have a case of improved horizontal equity but worsened vertical equity, a conflict that cannot be resolved by a hike in rate progressivity because it would need to be restricted to elderly taxpayers.

Gender equity and alternative policies

Pension income splitting has been shown to augment horizontal equity and compromise vertical equity, but what about the effects on gender equity? Most married women who are in or approaching retirement have had a constrained lifetime ability to earn income and therefore to accumulate savings independent of their husbands. Their situation is the lingering result of discriminatory economic and social practices in previous decades. At one time, for example, working women who married or became pregnant were obliged to quit their jobs. While this difference in the earnings and pension accruals of men and women is narrowing — and thus reducing the need for pension splitting in the future — it is a reality now. Pension income splitting helps to right gender inequities, in particular in the tax treatment of couples where the wife’s lifetime savings opportunities have been hindered. Pension splitting does nothing for single, separated or widowed women who have been similarly disadvantaged, but these groups account for a disproportionate share of OAS and Guaranteed Income Supplement expenditures, the largest of all federal programs.

There is another aspect of pension income splitting that relates to gender equity. The tax savings derived from splitting go to the higher-income spouse (usually the husband), while the incremental tax liability from splitting falls on the lower-income spouse (most often the wife). Gender equity is concerned with the equality of resources and opportunities between men and women, within conjugal relationships, and one cannot assume that the couple’s combined resources will always be shared equally. Clearly, for couples with relatively egalitarian financial relations, it will not matter which partner pays the additional taxes and which one saves taxes from pension splitting. The two have a shared interest in minimizing their joint tax liability and maximizing their joint resources. Pension splitting requires the explicit consent of the spouse who receives the notional transfer of pension income; that consent will be freely given by a spouse in an egalitarian financial relationship whenever there are joint tax savings to be had. The problem arises for couples who do not equally share total financial resources or decision-making. Even in these couples, the lower-income spouse will give consent for splitting only when she deems it to be in her own interests. The consent requirement improves her bargaining position, so that one would expect her to obtain full recompense for her incremental tax liability on the notionally shifted income plus part of the couple’s net gain from splitting. The wife might gain as much or even more from this form of splitting than from requiring an interspousal transfer of property title to split incomes; without a clear understanding of the couple’s decision process, the comparative outcomes are unknown.

The new pension-income-splitting provision has been severely criticized by economist Frances Woolley (2007) on equity, efficiency and administrative grounds. Her principal objections are that the provision shifts tax liability to women without shifting control over the associated income and, as previously noted, that its benefits are sharply skewed toward higher-income senior couples.⁸² Woolley offers several alternative policy responses, but here I discuss only three. One proposal is simply to live with the existing inequity, on the grounds that it should diminish over time with women’s rising relative earnings and pension accrual rates. Yet, that approach ignores the disadvantage faced by the current generation of older women, and it could appear overly sanguine about improvements for their successor cohort. Another proposal is to require spouses actually to transfer title to the pension asset that is generating the income in order to split it for tax purposes. While attractive in principle, federal and provincial pension laws, the Income Tax Act and provisions of each RPP agreement would forestall such an approach. It would be a daunting task to not only institute the requisite legislative changes but also amend the terms of every pension plan.⁸³ Allowing interspousal transfers of wealth without income attribution would be more attractive for nonregistered assets; I pursue this idea in my analysis of investment income splitting.

A third proposal of Woolley is to allow the transfer of unused room from the lower-income spouse’s bottom tax bracket to reduce the tax burden on the pension income of the higher-income spouse. She suggests this approach to address anticipated administrative problems involving an “innocent spouse” (where a spouse unknowingly incurs tax liability), incomplete take-up, the death of a spouse and remarriage, as well as to avoid shifting tax liability to the lower-income spouse. Her proposal to restrict tax-room shifting to the bottom bracket and to limit the tax savings to the rate differential between the bottom and second-lowest tax brackets would also reduce the benefits of splitting for higher-income couples.⁸⁴ This restriction would thus improve the vertical equity of pension income splitting while reducing its horizontal equity gains. Such an approach — possibly extended to cover shifting of unused tax room in all brackets — might warrant consideration if the posited problems are found to be material in practice.

Investment Income Splitting

T ax splitting for investment income⁸⁵ raises issues similar to those for pension income. The Canadian tax system already contains many methods by which astute, well-advised couples can legally split investment income, as I described earlier. If pursued systematically over an extended period, large amounts of investment income can be shifted between spouses. Hence, for horizontal equity, the key question is whether fairness would be improved by removing the income attribution rules or by allowing all couples to access simple provisions for splitting their combined investment income. Reforms of these kinds would improve horizontal equity for couples who are not well informed or well advised in tax and financial matters. They could also bring other benefits: the use of fewer resources by the tax authorities to police the system; reduced expenditure on financial, tax and legal advisory services; and a sharper focus on maximizing investment returns rather than tilting portfolios and trading strategies to minimize the couple’s joint taxes. Insofar as these changes would improve the efficiency of capital markets, benefits would arise for society at large as well as for the couples involved.

One might argue that allowing tax splitting for investment income while denying it for labour income would itself constitute a form of horizontal inequity. However, this view ignores the inequities that already exist between couples who engage in tax planning (typically those with higher incomes) and those who do not. Full splitting of investment income might also seem unduly generous to couples relative to singles since it fails to account for the scale economies couples enjoy. Yet, dealing with scale economies becomes relatively difficult in a scheme that permits only certain types of income to be split. The Meade Committee, which examined tax reform in the United Kingdom, considered a “partial quotient system” for splitting investment income between spouses while maintaining the separate taxation of their labour income (Meade 1978, 389-90). This method would reflect the scale economies of couples with a quotient of 1.5 on their split investment income, but its operation would be complex.

With joint taxation of couples (embodying either full or partial splitting), the allocation of each spouse’s investment income raises no issues; their total investment income is amalgamated with labour income for tax purposes. But if couples are taxed along with singles in a system of individual taxation, rules are needed for the treatment of their investment income. All countries that use individual taxation have faced this problem with varying solutions, including:

• the unrestricted division of investment income between spouses by allowing asset transfers (United Kingdom);⁸⁶

• the allocation of all investment income to the spouse with the higher labour earnings (the Netherlands);

• the allocation of all investment income over a specified threshold to the higher-earning spouse (Denmark);

• the taxation of all investment income at a flat rate independent of the progressive rates applied to labour income (Finland, Norway, Sweden);
• the division of all investment income equally between the two spouses (Belgium);⁸⁷ and
• the taxation of investment income to the spouse who generated the savings (Canada).

Which of these approaches best meets horizontal equity concerns, minimizes resource costs and investment distortions, and maintains gender and vertical equity across individuals? The UK system is the most lenient, as it gives couples unlimited ability to shift investment income to the lower-earning spouse, thereby reducing the tax system’s effective progressivity. Conversely, allocating all of a couple’s investment income to the spouse with the higher labour earnings, as several countries do, is unduly harsh and creates a substantial marriage penalty for two-earner couples in which both spouses save. Canada’s system in some sense takes a middle ground by seeking to tax the spouse who generated the original savings, but its deficiencies — in terms of horizontal inequity, enforcement problems, inefficiency and tax planning costs — are large. Either of two alternative policies could offer a significant improvement over Canada’s current income attribution regime.⁸⁸

Tax reform for investment income

One alternative would be to allocate half of the couple’s total investment income to each spouse. This approach would greatly simplify income attribution, and it would vastly reduce the resources and time currently spent on tax avoidance and enforcement related to interspousal transfers.⁸⁹ It could actually yield more progressive outcomes in cases where couples have shifted more than half their total investment income to the lower-earning spouse. Applying this method in Canada would require separate splitting for each of the types of investment income that are taxed differently: interest and income trust distributions, capital gains and dividends. Note that such an approach would entail a formal allocation rule, not simply an option for couples to split part or all of their combined investment income. Couples in which one spouse has little or no labour earnings would still have an incentive to shift more than half their total investment income to the lower-income spouse, thus necessitating some anti-avoidance measures.

An alternative reform would be simply to lift the attribution rules that prevent interspousal splitting of investment income and to allow spouses to transfer assets to their partner without tax consequence. The associated investment income would be taxable to the spouse holding title to the asset.⁹⁰ In a variant of this approach, interspousal transfers would be deemed to be realized dispositions with respect to capital gains tax but all subsequent investment returns would be taxable to the transferee.⁹¹ This reform has two potential advantages over the first alternative. First, it would impose the tax only on the partner who owned the asset and thus directly controlled the income flow. Second, it might encourage the more egalitarian allocation of total wealth between spouses. Conversely, it would allow unlimited shifting of a couple’s investment income to the lower-income spouse, which could undercut the tax system’s vertical equity to a greater degree. The actual extent to which spouses would choose to shift assets to their lower-earning partners under such a regime is an empirical question for which data and estimates are lacking.⁹² Relative to the first alternative, this approach might not decrease the returns to and costs of tax planning as much, but it would reduce them considerably from the current regime.⁹³ Thus, differential outcomes could still arise for “equal” couples. Table 3 summarizes key differences between the status quo and the two proposed alternative policies.

Gender equity and vertical equity effects

The gender equity criterion would lead to a strong preference for lifting the attribution rules and allowing the tax-free transfer of assets between spouses over both allocating their total investment income equally between them and maintaining the status quo. Two concerns motivate this preference. Foremost is the desire to equalize economic resources, opportunities and decision-making between women and men. Allowing the free transfer of assets between spouses without income attribution would encourage greater shifts to lower-earning spouses, whereas the 50:50 allocation approach would discourage the tax-motivated interspousal asset transfers that occur even under the present regime. The other concern relates to taxing spouses on income over which they have no control, which would arise under the first alternative — a point Woolley cites with respect to the new pension income-splitting provision (2007). Feminist tax policy analysts in Canada have supported relaxing or abolishing the attribution rules to promote gender equity goals.⁹⁴ As tax law scholar Lisa Philipps asserts, “the time is ripe for a thorough debate on whether the spousal attribution rules have become outmoded in view of changes in the economic status of women and in social norms surrounding gender equality” (2002, 1035).

The issue of control over transferred assets is critical in a gender equity assessment of changes to the attribution rules. As both Lahey (2000) and Philipps (2002) note, the transfer of the title of an asset to a lower-earning spouse is not necessarily equivalent to the transfer of control over the asset, depending on the dynamics of power between the two spouses. Asset transfers motivated purely by tax avoidance would be facilitated by the abolition of attribution, which would reduce the tax system’s vertical equity. Philipps proposes that the attribution rules not be abolished but be waived only when there is clear evidence of the transfer of effective control over the asset as well as legal title.⁹⁵ Meeting this goal would entail the tax authority’s looking into the couple’s relationship in ways that would be difficult in practice, not to mention highly intrusive. At most, the rules could apply attribution on an ex post basis to flagrant cases, such as where the transferor took back the assets over the objections of the transferee.

For most practical purposes, then, the attribution rules would be abolished. As for the effect on vertical equity, normally a concern for feminist policy analysts, Philipps notes: “Some may be prepared to live with this regressive impact if it results in wealth and income being shared more equally by women” (2002, 1035).⁹⁶

Revenue and economic effects

Whether we allocate half of the couple’s total investment income to each spouse or allow spouses to transfer assets and the associated tax liability to their partner, either proposal likely would entail a net revenue cost for government. Since no information is available on the extent of current tax splitting of investment income between spouses, no aggregate estimate of that cost is possible. Some couples do not engage in tax planning on investments to the full legal extent, and both proposals would expand splitting for other couples. A 50:50 allocation rule would bind couples who have exploited tax avoidance extensively and would see them facing higher taxes. The revenue cost of allowing interspousal asset transfers without income attribution might be greater, but it could be partially offset by deeming gains to be realized and taxable at the time of the transfer. Few couples except those at the highest income levels have substantial non-tax-sheltered savings outside their pension plans, RRSPs and home equity, which would also limit the total revenue cost.

The distributional effect of both proposals clearly would be to reduce vertical equity, but the effect would be less than one might imagine due to the effort currently devoted to applying (and avoiding) the income attribution rules. The maximum annual savings in federal income taxes would arise for a couple who had previously pursued no splitting strategies and was able to shift assets generating $121,000 annually of taxable income (the threshold for the top rate bracket in 2007) to a spouse who otherwise had no income of his or her own. That amount of income would represent an asset shift of at least $1.5 million at an 8 percent rate of return. The annual savings for such a couple would be $9,200 (see table 4 on page 29), a modest fraction of the couple’s total federal tax liability of at least $58,000 per year — and typically much more.⁹⁷ It is improbable that many such high-income couples have not already substantially exploited their legal tax-splitting opportunities. Hence, most of the gains would go to upper-middle-income couples with large non-tax-sheltered assets.

The two proposals could also have effects on aggregate savings and economic efficiency. By lowering the tax bite on some couples’ investment income, either proposal would reduce their effective marginal tax rate, increase the net return to saving and encourage greater savings.⁹⁸ Conversely, couples who are “target savers” with specific retirement savings goals might reduce their savings with the higher return to saving. Regardless of whether aggregate savings rise or fall, the economy’s efficiency would unambiguously improve. Decreasing the effective marginal tax rate on the return to savings would shift the personal tax base further from income and toward consumption, which would increase efficiency (Kesselman 2004). A 50:50 allocation rule would create further gains in efficiency by saving resources now spent on tax planning and getting couples to focus on maximizing their investment returns rather than minimizing their tax burden.

Labour Income Splitting

For the great majority of nonelderly couples, labour earnings constitute by far the largest portion of their total income. Hence, it is not surprising that tax splitting of labour income has attracted the greatest scrutiny in previous research. Like the analysis of splitting for pension and investment income, horizontal equity plays a key role in the analysis of labour income splitting. Unlike income from pensions and investments, though, the issue of differential tax avoidance does not play a significant role in assessing horizontal equity for labour income splitting. The bulk of labour earnings are wages, salaries and fringe benefits for employees, and little scope exists for shifting tax liability for those items to the worker’s spouse. Self-employment offers greater opportunity for shifting income to the worker’s spouse — beyond the spouse’s true contribution to the enterprise.⁹⁹ An empirical study of illegal income shifting by self-employed Canadian workers found the activity to be “non-trivial” but nevertheless very small; its estimates imply that only about 0.5 of 1 percent of the labour force is engaged in this type of tax avoidance (Schuetze 2006).¹⁰⁰ The phenomenon thus cannot justify a horizontal equity argument for permitting broader splitting of labour income. Additionally, the free entry of workers into self-employed occupations suggests that any gains from differential tax avoidance relative to employees would be dissipated through competition.¹⁰¹

The choice of the appropriate tax unit with respect to labour income has evoked heated controversy on account of strongly held views about the role of women in the home and in the labour market. This debate has been lively not only in Canada but also in other countries where the issue has been examined. For example, in Ireland a judge argued that “[i]ndividualisation [of the tax system] is a naked attempt to increase female participation in the labour force and to coerce women, who might otherwise opt to work at home while their children are young, into the labour force” (quoted in Callan 2006, 17). Taking the opposite perspective, a Dutch economist asserted that “[j]oint or split taxation tends to conserve sex roles and make women more dependent on their husbands by decreasing married women’s economic remunerations from participating in the labor force and mak[ing] market-related human capital investments” (Gustafsson 1992, 82). As detailed in my review of the US tax treatment of couples, joint taxation has raised endless controversy over distortions to couples’ marital and working choices.

In assessing the horizontal equity of labour income splitting, I consider three distinct cases. (For simplicity, I abstract from the presence of income from any sources other than labour.) The first case is that of a one-earner couple versus a two-earner couple, both with the same total income; this is the conventional example that has motivated much debate over income splitting in Canada. The second case is that of two-earner couples with the same total income but with different splits in the spouses’ earnings; this situation has also attracted much attention in public discourse. The third case is that of a one-earner couple in which one spouse serves in a supportive role to make possible the other spouse’s high earnings; this situation has attracted comparatively little discussion. I conclude the section by examining the revenue, distributional, behavioural, parenting and gender effects of alternative choices of the tax unit in relation to labour income.

Case 1: One-earner versus two-earner couples

The classical argument for general income splitting, including all labour income, hinges on a comparison of oneand two-earner couples with the same total income. A progressive tax rate schedule clearly implies that, with individual taxation, the one-earner couple will pay more tax than the two-earner couple. The one-earner couple gets the benefit of the lower tax brackets just once, while the two-earner couple benefits twice.¹⁰² The argument is that differential taxation of couples with the same income is unfair in that it violates horizontal equity. Contemporary proponents of income splitting have expanded this argument to include considerations of caring for children at home or in daycare. As MP Garth Turner has expressed this view, “[i]ncome-splitting would increase household cash flow, encouraging many households to consider having children by lessening the financial burden. It would even out the tax disparities between single and dual-income families and give much needed monetary value to the work of unpaid caregivers” (2006). The one-earner versus two-earner argument for splitting is typically set in terms of couples with children, but it also applies where one spouse stays home to engage in the production of nonmarket goods and services besides child care.

The fundamental deficiency of this argument for income splitting is that it regards money income from market work as the measure of “equals.” It ignores the fact that real living standards, a better measure for “equals” and thus for assessing horizontal equity, are also affected by factors that differ between one-earner and two-earner couples. Economists and other policy analysts have recognized this point in the research literature on the tax unit for decades,¹⁰³ and it has also been acknowledged in a report of the House of Commons Finance Committee:

a dual-earner couple with the same total income as a single-earner couple is not as well off as the latter. Not only are there additional employment related expenses that must be incurred with respect to the second worker, the value of unpaid work in the home, or leisure, must also be taken into account. (Canada 1999b, 12)

In other words, in assessing horizontal equity in the taxation of single-versus dual-earner couples, one needs to consider the incremental work-related costs and the loss of valued home-produced goods and services associated with the second earner’s market work.¹⁰⁴ (Henceforth, I use the term “second earner” to denote the spouse who is more likely to stay out of the work force or to have the lower earnings. In most cases this is the wife, but increasingly it is the husband.¹⁰⁵)

Costs related to the second earner’s employment include transport to work, clothing and personal care needed for the job, meals away from home, union and professional fees and the like. Where children are present in the household, daycare expenses might also be incurred when the second earner works. Additionally, there are income and payroll taxes on the second earner’s income. Lost value from in-home production when the second earner goes to work include cooking, cleaning, decorating, gardening, laundering, mending, caring for children and other services. The household with a second earner needs to purchase many of these items, or substitutes, in the market rather than self-supplying them, and these expenses eat into the additional earnings along with direct work-related expenses. One study of couples without children found that the average two-earner couple required about 30 percent more money income to achieve the same real living standard as a one-earner couple (Lazear and Michael 1980, 207). This differential widens further if one considers the daycare costs incurred with a second earner or, alternatively, the loss of self-provided child care. A study of two-parent families with children found that dual-earner families spent up to eight times more than single-earner families on daycare and babysitting. All expenses incurred by dual-earner families diminished their income advantage over single-earner families by 46, 57 and 68 percent, respectively, at low, middle and high incomes (Hanson and Ooms 1991, 632).¹⁰⁶

Clearly, if horizontal equity is to be judged by real living standards, one cannot compare two-earner couples and one-earner couples with the same money income alone. One-earner couples have potential income that they could generate if the second spouse went to work, yet the second spouse’s valued home-produced goods and services are not counted as part of the family’s money (or taxable) income. Wives induced by the tax system — such as by the increased marginal tax rate on second earners under US joint taxation — to reduce or quit their paid work have been found to devote their incremental time to household production rather than to leisure (Hunt, DeLorme and Hill 1981). By virtue of their choosing to keep one spouse at home rather than in the paid labour force, the single-earner couple reveals that the value of additional work in the home exceeds the value of the foregone second earnings.¹⁰⁷ That is, the one-earner couple has “real income” (including the value of household production) that exceeds its money income. For that reason, it is improper to undertake a horizontal equity assessment of one-earner and two-earner couples having the same money income; they are, in fact, not “equals” in their real income. For this case, the tax unit satisfying horizontal equity is the individual, not the couple.

Case 2: Two-earner couples with different earnings splits

Another long-standing argument for income splitting is that couples having the same total income but with earnings split differently between spouses should pay the same total tax. This case differs from case 1 in that both spouses are assumed to be working full time, so that none of the couples being compared has additional time for household production or leisure activities, and there are also no differences in their work-related expenses. The argument proceeds from the assumptions that the family or couple is the economic unit and that two-earner couples with the same total income have the same ability to pay. Differing tax burdens based on the division of earnings are thus deemed to constitute a horizontal inequity, and full income splitting is asserted to be requisite for horizontal equity. However, even in this case, unless the two spouses’ earnings diverge sufficiently to put them in different tax rate brackets, they would not gain anything from income splitting. To put this into context, using the 2007 federal rate brackets, income splitting would benefit neither couples in which both spouses have taxable incomes of less than $37,178 nor those in which both have taxable incomes between $37,178 and $74,357.

Whether the couple can be considered the economic unit for tax purposes hinges on whether spouses fully pool their combined income and act as a unitary decision-maker. One empirical test of that hypothesis has examined the effect of a change in the payment of a public child benefit from the father to the mother (Lundberg, Pollak and Wales 1997). It found that, holding total family income constant, the income each spouse receives has a significant effect on expenditure patterns: increasing the mother’s income raises the amount the couple spends on her and the children. Other studies have uncovered extensive evidence that many couples do not fully pool or share their earnings or other economic resources.¹⁰⁸ In their study, Vogler and Pahl find that “the orthodox model of households as egalitarian decision making units, within which resources are shared equally, applied to only one-fifth of the households in the sample, those using the joint pool system” (1994, 285). In fact, many spouses keep their earnings in separate accounts, and even when spouses do pool their income, they might not share equally in decisions on spending. Tax law scholar Marjorie Kornhauser reports that “the amount of money one partner earned relative to the other determined relative power and control over resources” (1993, 89). And although the nonearner spouse might manage the couple’s finances, the earner spouse tends to control the actual spending beyond routine outlays.

Even if couples fully shared their earnings, horizontal equity between couples and singles would still require recognition of the scale economies that couples enjoy relative to singles. Therefore, the proper tax treatment would not be full income splitting but joint taxation with tax brackets less than twice the width of those for single taxpayers. That system has the well-known marriage penalty for couples where the spouses earn incomes that do not diverge widely. Moreover, the practical issue arises of how tax policy could distinguish between couples in case 1 and those in case 2. In fact, there is a continuum between case 2 couples (where both spouses work full time) and case 1 couples (where one spouse works full time and the other is out of the labour force). Yearly earnings do not provide sufficient information to distinguish between a spouse who works full time, full year at a modest pay rate and a spouse who works part time and/or part year at a high pay rate. Some spouses earn amounts that might appear to come from full-time work while actually working part time and/or part year to reduce their work-related expenses or have extra time for household production. Without knowing the time worked per year, tax policy cannot distinguish properly among couples. Thus, it is not feasible to implement individual taxation for case 1 couples and joint taxation for case 2 couples, even if that were appropriate.

Case 3: One-earner couple with an assisting spouse

The third case involves a one-earner couple in which one spouse has opted out of the paid labour market to provide supportive services that enable the other spouse to generate higher earnings.¹⁰⁹ If one spouse has a highly paid, demanding job, the couple might rationally decide that the other spouse should play an assisting role. Indeed, the support of such “informal market work” could be critical to the success of the spouse who supplies the formal market work. These supportive services can be wide ranging: “(i) direct substitution, in which she does work that could be done by a paid employee (e.g. clerical work, deliveries); (ii) indirect support, including social hosting, that deploys relationship building skills; (iii) consulting, that is listening, advising and helping the employee with judgment calls and decisions; and (iv) emotional aid, providing encouragement and moral support about job challenges.”¹¹⁰ In effect, the two spouses are working jointly for one paid job and one unpaid job.¹¹¹ In a commentary on income splitting, Vancouver Sun columnist Don Cayo writes, “My wife[’s]…support at home freed me up for a career that included not only a full-time job, but also a lot of freelancing” (2006) — a pattern that is common among top executives and professionals in diverse occupations.

Unlike case 1, in case 3 the at-home spouse is a productive input to the market services supplied by the “employed” spouse. The at-home spouse does not necessarily have more time to produce in-kind goods and services; rather, “her” output takes the form of higher market earnings received by, and taxable to, the employed spouse. If one spouse is self-employed or operates a closely held business, the supportive spouse can be paid appropriately and taxed on that income. If the first spouse is an employee, however, this type of income-splitting opportunity typically does not exist. One could justify the introduction of an income-splitting provision for this case on grounds of both horizontal equity and efficiency,¹¹² but the splitting usually should be less than 50:50, and there is no universal guide as to the proper proportions. Given couples’ incentives to minimize their total taxes, they cannot be expected to self-report trustworthy figures for the shares of earnings to allocate to each spouse. The best solution for case 3 would be to retain individual taxation on labour income but allow employers to make direct payments to spouses of employees for supportive services (see Philipps 2007). This approach would also address gender equity concerns that women be compensated for their work and not be taxed on income they do not control.

Revenue and distributional effects

A study undertaken by the Library of Parliament estimates the total federal revenue effect of full income splitting by all couples to be $5.0 billion for the 2007 tax year (Laurin 2006, 2). Of that total, $2.2 billion is attributed to nonelderly couples with children, $2.1 billion to nonelderly couples without children and $700 million to elderly couples.¹¹³ Since the 2007 budget raised the spousal tax credit to equal the basic filer credit, the related $270 million annual revenue cost would reduce the net cost of full splitting by that amount. Of the remaining $4.0 billion revenue cost of splitting by nonelderly couples in 2007, the great bulk would be attributable to labour income rather than investment income (though no breakdown is available). Another Library of Parliament study put the 2005 federal revenue cost of splitting for two-parent families with children younger than 18 at $1.6 billion, with $1.1 billion for dual-earner couples and $500 million for single-earner couples (Bergevin, Laurin and Kitching 2006, 8). The $1.6 billion figure was reduced by $400 million if splitting were restricted to families with at least one child age 12 or under.

The distribution of savings from the $1.6 billion of tax relief for couples with children in 2005 would be just 13 percent to families with total income of $60,000 or less and 37 percent to families with total income exceeding $100,000. On a per family basis, by far the largest tax savings, an average of $3,362, would go to one-earner families with income above $120,000, versus an average of $177 to one-earner families with income between $20,000 and $40,000 (Bergevin, Laurin and Kitching 2006, 8-9). Most of the savings attributed to tax splitting for the latter families, in fact, was eliminated by the equalization of spousal and basic tax credits beginning in the 2007 tax year. Similarly, the tax savings from full income splitting for elderly couples are highly concentrated in the highest-income groups. Elderly couples with total income over $90,000 almost entirely accruing to one spouse are estimated to save an average of $3,292. In contrast, all elderly couples with total income below $30,000 save an average of just $215 (Laurin 2006, 7); this figure is explained by the spousal credit differential that has now been eliminated.

Table 4 illustrates the distributional pattern of federal tax savings in 2007 from full income splitting for a range of assumptions about a couple’s income — both total income and the split between spouses. If the provinces went along with a general income-splitting scheme at the federal level, the tax savings would be compounded.¹¹⁴ As expected, the table shows only positive figures representing marriage bonuses and no marriage penalties. These results reflect the proximity of the spouses’ incomes to the thresholds of the various rate brackets and their differential marginal tax rates. One can observe the following patterns:

• an earnings split of 50:50 yields no tax savings,  since incomes are already equally divided;
• no tax savings arise for any earnings split with total income of $30,000, since this figure is already in the bottom rate bracket;
• with a 40:60 split, no tax savings arise at some income levels because the two spouses are already in the same rate bracket, and the tax savings overall are surprisingly modest compared with more divergent splits;
• except for the most divergent split (0:100 for oneearner couples), tax savings do not rise uniformly with total income;
• at any level of total income, the largest tax savings from splitting accrue to one-earner couples;
• by far the largest tax savings arise for couples at the highest incomes; and
• the peak tax savings of $9,203 per year arise for one-earner couples with income of $241,774 or higher, which is two times the top bracket income threshold.

Any irregularities in the patterns of tax savings are not inequitable per se; they might simply reflect the effect of removing horizontal inequities in the existing tax structure, if the assumptions for income splitting are correct.

Table 5 illustrates the comparative tax savings and tax costs arising for couples under a system of joint taxation with partial splitting, where the couple’s equivalence scale is taken to be 1.5. That is, all the 2007 federal tax brackets have been expanded by 1.5 for joint filers, and joint filing using this schedule is assumed to be mandatory for couples.¹¹⁵ The same variations in combined income levels and assumed splits of earnings are used in table 5 as in table 4, except that the top level of total earnings shown is $182,000 — 1.5 times the top bracket threshold.¹¹⁶ Compared with full income splitting, joint taxation reflecting couples’ scale economies is like an upward shift in the tax schedule. Therefore, most figures shown in table 5 equal the corresponding figures in table 4 minus a constant that hinges on the couple’s total earnings; for example, at total earnings of $100,000 each figure is reduced by $1,301. Joint taxation creates tax savings for one-earner couples at almost all levels of total earnings. As expected, joint taxation creates marriage penalties (as shown by negative figures) for couples in which the spouses’ earnings do not diverge greatly. Unexpectedly, these marriage penalties arise even for splits as wide as 25:75 for couples with total earnings of $75,000 and higher.

Behavioural effects of the choice of tax unit

The choice of tax unit can affect the behaviour of individuals and spouses in a variety of ways. Here, I focus on several key behavioural issues that arise with respect to labour earnings (although some also apply to other types of income): effects on the labour supply of spouses (particularly wives); effects on efficiency and vertical equity, and what those responses can tell us about the optimal tax unit; the distribution of resources between spouses; and effects on marriage and divorce decisions.

Effects on labour supply

Considerable research effort has been devoted to understanding workers’ labour supply behaviour; this includes both the decision to participate in the labour force and the choice of how many hours to work. Particular attention has been given to the behaviour of married women relative to married men, an issue of clear relevance to income splitting. Both full splitting and joint taxation’s partial splitting can reduce the marginal tax rate a husband faces (assuming he is the higher-earning spouse) and raise the marginal tax rate of the wife. Evidence shows that the work hours and labour force participation of wives are much more responsive than those of husbands to variations in the net wage rate, such as those caused by changes in marginal tax rates. Over the past 30 years, the responsiveness of married women to net wage rates or tax rates has declined but it is still larger than for married men.¹¹⁷ Hence, the expectation is that individual taxation will be more favourable than joint taxation or full splitting for the workforce participation and working hours of most married women. One confirmation comes from Crossley and Jeon’s study of Canada’s 1988 shift from a spousal tax exemption to a tax credit, which reduced the second earner’s marginal tax rate on initial earnings. They found that this reduced “jointness” in the tax system’s treatment of couples was associated with a 9 to 10 percentage point increase in the labour force participation of women with low education married to higher-earning men (2006).

As for the labour supply responses of married women to changes in the tax unit, LaLumia (2005) examined the effect of the 1948 US introduction of joint taxation for couples by comparing behaviour in the newly affected common-law states with community-property states. She found that the tax change was associated with a 0.9 to 1.6 percentage point decline in married women’s probability of employment; she also found that married women were less likely to have nonlabour income after 1948 (2005). Hausman projected significant increases in wives’ labour supply if joint taxation were converted to individual taxation (1981). Feenberg and Rosen simulated the effect of the United States’ shifting from joint taxation to full splitting for couples and found a modest increase in work hours of second earners (1983). Simulating the effect of giving couples the option between joint taxation and filing as individuals using the singles’ rate schedule yielded larger increases in wives’ work hours, particularly among two-earner couples. Leuthold forecast a 25 to 35 percent increase in labour force participation by married women if the United States eliminated the partial splitting of joint taxation; moreover, the effects would be larger for the wives of higher-income men than those of lower-income men (1984).

In the European context,¹¹⁸ Gustafsson compared Sweden, which had individual taxation, and West Germany, which had income splitting for couples. His simulations found that imposing the West German tax system on Swedish wives would reduce their labour force participation rates by 20 percentage points, while imposing the Swedish tax system on West German wives would raise their rates by10 percentage points (1992). Callan estimates the effects of increasing the individualization of the Irish tax system by eliminating interspousal transfers of the standard tax rate band and returning the revenue gains via cuts to tax rates. He finds that husbands’ participation rates would be virtually unchanged while those of wives would rise by about 2.5 percentage points (2006). An OECD study estimates positive effects on married women’s participation rates from the full individualization introduced in Belgium’s 2001 tax reform and from prospective elimination of the “couples” component of France’s family quotient system (2006). O’Donoghue and Sutherland estimated the effects on effective marginal tax rates of British taxpayers — who were subject to individual taxation — of imposing the various degrees of splitting used in France, Germany and Spain. They found, in every case, that husband’s marginal rates would decrease while wives’ rates would rise between 4 and 7 percentage points, with clear implications for labour supply (1999).

Effects on economic efficiency and social welfare

Efficiency is a fundamental concept in economic policy analysis, and it relates to how resources would be allocated in a hypothetical perfectly competitive economy. Economic inefficiencies therefore relate to the distorting effects of tax, subsidy and regulatory policies on behaviour from what it would otherwise be in this idealized economy. Tax policies that reduce tax rates on more responsive factors (such as the labour supply of one group of workers) and raise them on less responsive ones (such as the labour supply of another group) can increase total economic efficiency, since they reduce the net distortion to economic behaviour. The differential labour supply responses of married men and women to changes in their effective marginal tax rates imply that a change in the tax unit might affect economic efficiency. Policy changes that increase the economy’s efficiency do not necessarily raise the level of social welfare; the latter concept also incorporates value weights on the groups that lose as well as those that gain from the change.

Rosen computed that shifting from joint to individual taxation of spouses would improve the economy’s efficiency and possibly raise social welfare (1976), and Hausman suggested that joint taxation caused a loss of economic welfare due to the high tax rates imposed on the labour supply of married women (1981). A succeeding series of theoretical economic studies has assessed the choice of tax unit in the broader framework of optimal taxation.¹¹⁹

They balance efficiency effects against vertical equity effects to maximize society’s welfare, and their results are expressed in terms of the optimal relative tax rates on a couple’s primary and secondary earners.¹²⁰ The seminal contribution by Boskin and Sheshinski found that the secondary earner (usually the wife) should face a tax rate only about half that of the primary earner (usually the husband) (1983).¹²¹ Neither joint taxation nor income splitting would realize this outcome, since they would impose the same tax rate at the margin on both spouses, but individual taxation with a sufficiently progressive rate schedule would suffice.

Some of Boskin and Sheshinski’s restrictive assumptions have been addressed in more recent research. In a pure efficiency analysis, Piggott and Whalley challenged the earlier analysis for ignoring distortions of couples’ household production choices (1996).¹²² While the system of differing tax rates under individual taxation is more efficient with respect to spouses’ differing labour supply responses, joint taxation’s uniform rate on both spouses is more efficient with respect to their household production choices. Piggott and Whalley’s analysis showed that it is theoretically possible for the latter consideration to be more important, with married women choosing to do too little work in the home sector under individual taxation. This finding indicates the potential superiority of joint taxation, but the research has been critiqued on both analytical and quantitative grounds.¹²³ In the end, the Piggott-Whalley analysis implies lesser difference in the optimal tax rates on primary and secondary earners than suggested in the earlier analysis, but it does not unambiguously support joint taxation.

Several recent contributions have used the optimal tax approach to assessing the tax unit.
Kleven, Kreiner and Saez extend the Boskin-Sheshinski analytical model to consider a fully general income tax rate schedule rather than a linear schedule.¹²⁴ They find that negative jointness between spouses is optimal — that is, the higher the primary worker’s earnings, the lower should be the secondary earner’s tax rate. Joint taxation, in contrast, has positive jointness, with each spouse’s tax rate rising with the other spouse’s higher earnings. Hence, Kleven and associates find that the individual tax unit is optimal for income taxes, while the couple’s combined income is optimal for the income tests in transfer programs (2006). That finding answers a question I posed at the start of this study about the inconsistency of taxing spouses as individuals while treating couples jointly for transfers. Brett models the choice of the tax schedule for couples as a multidimensional screening problem. He concludes that, in the optimum, spouses should face different marginal tax rates, so that individual taxation is almost always superior to joint taxation (2007). Alesina and Ichino find that the optimal tax rate of married women should be much less than that of their spouses. They support not just individual taxation with the same rate schedules for men and women but gender-based tax rates that favour women — a policy that would run afoul of conventional notions of equity (2007).¹²⁵

Effects on distribution within the couple

The choice of tax unit potentially could affect the distribution of economic resources between spouses as well as across couples and singles. Evidence shows that sharing of income between spouses is far from universal or complete. An empirical study of the United Kingdom’s provision for unconstrained splitting of investment income found a significant increase in the share claimed by wives. It also found that “an overwhelming majority of households do not fully exploit their ability to reduce their tax burden” (Stephens and Ward-Batts 2004).¹²⁶ Thus, most couples do not behave as unitary decision-makers who optimize their combined taxes and equally share their joint income. Theoretical research has also investigated the effects of the choice of tax unit on the relative well-being of husbands and wives based on intrahousehold trade responses (Apps and Rees 1999b) or bargaining/sharing behaviour (Gugl 2007). The precise results hinge on various assumptions, but the general findings favour individual over joint taxation to achieve greater income equality between spouses. A shift to joint taxation, while holding tax revenues constant, could even hurt the lower-earning or athome spouse, usually the wife. Of course, operational features such as source-withholding methods, tax-filing requirements, tax refunds and joint liability for payment could also affect the outcome.¹²⁷

From the gender equity perspective, concern over intracouple distribution extends beyond the division of a couple’s joint income. Equally important is the division of a couple’s time between market work and home responsibilities, both routine household duties and child care. Many public policies are germane to this issue (see Kershaw 2005), with the choice of tax unit high on the list. Previously cited empirical findings on the adverse effect of income splitting on the labour force participation and paid work hours of married women has strong implications for this issue. Choosing individual taxation rather than any form of joint taxation would increase the wife’s time in the paid market, decrease the husband’s time in paid labour and potentially allow time for his greater participation in both home duties and child rearing. Greater labour force activity by married women would also promote their earnings capacity and economic self-sufficiency in the event of subsequent marital dissolution — with positive consequences for the well-being of the children of such families. A theoretical analysis suggests that moving from individual to joint taxation would harm the spouse who is more productive in household production by discouraging premarital investment in human capital (Wrede 2003).

Effects on marriage and divorce

Empirical studies of the effects of the choice of tax unit on marital and divorce decisions have exploited the changing pattern of tax penalties and tax bonuses accruing to marriage under the US income tax.¹²⁸ One study found no statistically significant effect on the fraction of unmarried women over age 15 who marry in each year (Sjoquist and Walker 1995). Another study, using similar data but a slightly longer period and different methods, found a small but statistically significant effect: as the marriage tax increases and the marriage bonus decreases, the aggregate marriage rate also decreases (Alm and Whittington 1995). Both studies found evidence that higher marriage penalties cause some couples to delay marriage from late in the calendar year until the following spring to avoid the tax bite for one year. A study of Canada’s 1986 change in how the second spouse’s income was applied against the spousal exemption — from just income earned after the marriage to the spouse’s income for the full year — also found a shift in the timing of marriages away from late in the year (Gelardi 1996). Another study found that the US tax treatment of married couples affected divorce in the expected direction, although the response was small and more significant for women than for men (Whittington and Alm 1997).

The relevance of US findings to the Canadian context needs to be qualified. The Canadian tax system treats married couples and common-law partners the same, whereas in the United States only married couples face mandatory joint taxation. Hence, assuming that the authorities could determine when two people were in a common-law relationship, the application of joint taxation in Canada would not necessarily have an effect on marriage behaviour. The US marriage penalty might simply increase the rate of unmarried cohabitation among two-earner couples. But the US finding of an effect on divorce behaviour might still apply to Canada if it adopted joint taxation. Note that full splitting, which the United States allowed between 1948 and 1969, exerts behavioural effects in a direction similar to that of joint taxation except that no marriage penalties arise. Canada’s system of individual taxation imposes no distortions on marital or divorce decisions.

Splitting of labour income and parenting

Evidence from these diverse studies of the behavioural effects of income splitting and joint taxation further supports my finding based on horizontal equity for labour income: the individual is the best tax unit. Still, many advocates of income splitting see it as a way to encourage greater parental in-home child care, which they deem superior to care that others provide; some would allow income splitting only for couples with young children.¹²⁹ Regardless, this approach has three major difficulties. First, income splitting (or joint taxation) provides the greatest benefits to one-earner couples who need it least (the highest earners) and the least or no benefits to couples who need it most (the lowest earners). As several analysts have noted, income splitting is a poorly targeted method of supporting one-earner couples engaged in home child care (Cooper 1995; Callan 2006; Weir 2007). Second, income splitting would benefit nonelderly couples without dependent children more than couples with children (Laurin 2006, tables 2 and 8). Third, other programs are better suited to support families at all income levels in keeping a parent at home to care for children if that is their preference. One could consider enriching employment insurance parental leave benefits, National Child Benefits (particularly the supplement portion) or Universal Child Care Benefits.

Splitting of labour income and gender equity

Because labour earnings are by far the largest component of total incomes for most working-age couples, allowing this type of income to be split would have the biggest effects on gender equity. From almost any perspective, these affects would be adverse. Splitting would distort cohabitation and marital incentives, create disincentives for women to enter the labour force and bias spouses’ choices of the mix of household production and market work — tending to keep women in the home and performing most household work. The gender effects of splitting would also carry over from women’s short-run decisions into their longer-run life paths. Joint taxation and income splitting would pose hurdles to women’s taking part-time work and thereby tend to make their choice between work and home all-or-nothing. Extended time out of the labour force would interrupt their work experience, depreciate their market skills and possibly signal to potential employers a lack of commitment that might affect future job options. In contrast, individual taxation reduces the financial hurdles facing married women who wish to take part-time work, encouraging earlier re-entry into the labour market after maternity-related absences. Inequalities in market earnings between spouses also affect the dynamics of power over spending within the household. All of these disadvantages follow women throughout their lives, their effects accumulating at both the individual and societal level. They are also accentuated when marriages dissolve.

Related Issues

Several additional issues related to the tax treatment of couples warrant brief review: the spousal and equivalent tax credit; the relationship between the tax unit and flat taxes; the implications of the choice of tax unit for provincial income taxes; and the politics of and public attitudes toward income splitting.

The spousal and equivalent tax credit

The Canadian income tax’s key provision linking spouses is the spousal and equivalent tax credit, which in some form has been part of the tax system since its inception. Calls for the abolition or reform of this credit (and its predecessor exemption) have been made by the Royal Commission on the Status of Women (Canada 1970), the Law Commission of Canada (2001) and several feminist tax policy analysts.¹³⁰ Their arguments are similar and often parallel my reasons for opposing the splitting of couples’ labour income: the value of household production by an at-home spouse, the incremental work-related costs of the second spouse and work disincentives for the second earner. Moreover, the spousal credit is available regardless of whether claimants have dependent children. Indeed, the Law Commission of Canada says that more than half of spousal credit claimants have no dependent child and even fewer have a preschool child (2001, 76-7). As the Royal Commission on the Status of Women stated:

We believe that a woman does not become economically dependent by virtue of her marriage…A childless couple has the right to decide that the wife will devote all her time to homemaking but there is no reason why the State should attach an advantage to this choice by giving the husband a married status [tax] exemption. (Canada 1970, 299)

Recommended alternatives to the spousal credit have included restricting it to families with young children, making the credit refundable and payable directly to the at-home spouse and replacing it with an enriched National Child Benefit. A few countries that use an individual tax unit have abolished similar spousal provisions, but 21 of the 30 OECD countries have counterparts to Canada’s spousal tax credit (OECD 2005, 34-6).

Are there any arguments for retaining the spousal and equivalent tax credit? Ascribing free choice to the at-home spouse in the decision to work at home or in the labour market is not always accurate. Some spouses cannot readily find market work on account of a poor local economy for their skills, extended illness or other personal factors. Older women in traditional marriages, who have been out of the labour force for many years or were never active in it, are likely to find work only with great difficulty. Additionally, one could argue that married and common-law spouses have support obligations toward their spouse, mandated under provincial family laws, which creates a charge against the supporting spouse’s ability to pay taxes (Duff 2002). More pragmatically, attempts to abolish the spousal credit would encounter fierce political opposition (as did a similar suggestion by the Minister for the Status of Women in 1983). To meet the concerns of gender equity, the credit could be made refundable and payable directly to the athome spouse, but it would not eliminate that spouse’s work disincentives. And unless restricted to the presence of a dependent child, that change would raise the equity issue of eligibility by lowincome singles.

Splitting and flat taxes

Without progressivity of the tax rate schedule, the choice of tax unit would have no effect on the tax liabilities of couples versus singles or one-earner couples versus two-earner couples. A pure flat tax system — or a single-rate tax with a spousal exemption or credit equal in value to that of the basic filer and transferable to the filer — would treat all couples with the same income identically.¹³¹ Alberta’s single-rate income tax fits this bill precisely. Interestingly, the Canadian Alliance’s 2000 proposal for a flat tax was motivated by the argument of horizontal equity for single-earner couples versus dual-earner couples with the same total income (Canadian Alliance Party 2000, 17, 22). As noted earlier, conservative analysts in both Canada (Veldhuis and Clemens 2004) and the United States (Bartlett 1998) have supported the flat tax as their preferred solution. And in a frontpage newspaper column extolling the virtues of income splitting, Andrew Coyne has also cited the flat tax alternative (2006). Of course, abandoning rate progressivity, and thus vertical equity, in personal taxation in order to correct an alleged horizontal inequity is akin to throwing out the baby with the bathwater.¹³²

Implications of splitting for provincial taxation

If the federal tax system adopted income splitting for couples, this could have important implications for provincial income taxes. But the exact form of federal splitting would make a big difference. If the federal tax instituted a joint return for couples, then the nine provinces (all except Quebec) that piggyback their income tax onto the federal return under a tax collection agreement would be constrained to use the same joint return. Provinces would be able to retain individual taxation only by having their own tax returns, in which each spouse would separately identify his or her own income, thus raising taxpayers’ filing costs. However, if the federal income splitting were implemented via transfers of taxable income between spouses, while retaining individual filing, then each province could still choose to apply its tax on an individual basis without the need for separate tax returns. Either approach to retaining individual taxation at the provincial level would leave intact some incentives for couples to engage in tax planning to split incomes for provincial tax purposes. The revenue effect on the provinces of mandatory joint returns would hinge on the extent of splitting permitted — with full splitting entailing a substantial revenue loss unless provincial rate schedules were adjusted.

US practice at the state level is instructive on this issue. None of the states has a tax collection agreement similar to Canada’s that permits the use of a unified federal-state tax return. The 42 states that impose income tax use widely varying methods (see United States 1997, 61-2; Michael and Manzi 2007, 27-8). Six states employ flat rate taxes and therefore have small or no marriage tax penalties or bonuses.

Nine states require individual filing based on each spouse’s own income, similar to the US federal system in common-law states prior to 1948, or a combined return but with taxes applied to the spouses’ separate incomes as if they were single. Those states face the problem of how to prevent the shifting of assets between spouses to avoid tax; some allow couples to allocate income freely from jointly owned property. Thirteen states offer couples pure income splitting, with married joint tax brackets twice the width of singles’ brackets, like the US federal tax from 1948 through 1969. Seven states have a system of joint filing for couples with the widths of progressive tax rate brackets less than twice those for single filers, like that of the US federal tax for most of the period since 1969. Two states have hybrid systems with only some brackets for couples twice those for singles, yielding marriage penalties across all incomes. Five states offer a credit to offset the marriage penalty from their progressive rate structure.

Public attitudes and politics

Two polls have found that Canadians heavily support the introduction of general income splitting — with about three-quarters of respondents favourable to the proposal.¹³³ One pollster described income splitting as

a pretty simple concept and people find it a popular notion. We found in the past that with tax cuts in many, many instances, when traded off against health care and other social programs, and attach a number to it, it was rejected. But not in this case. We didn’t see it and I thought that was kind of surprising. (quoted in Aubry 2007)

These polling results appear to confirm the earlier characterization of income splitting as “the 21st-century equivalent of a chicken in every pot.”

Yet the depth and strength of support for income splitting has not been truly tested in the political arena. In early 2000, the Reform Party introduced its flat tax proposal, which initially garnered significant public interest and support. After the plan’s extreme distributional impacts — with tax relief heavily tilted toward the wealthiest taxpayers — were publicized in the run-up to the fall 2000 federal election, Reform’s successor, Canadian Alliance, modified the proposal. Its dual-rate tax plan substantially moderated the flat tax scheme’s distributional tilt (see Kesselman 2000). If a political party were to come forward in support of general income splitting, it seems likely that history would repeat itself. Opposition parties would publicize the fact that income splitting benefits only a minority of Canadian households, and even within that group the benefits are heavily skewed toward the highest earners. Electoral concerns could press any such proposal to be restricted to families with young children and/or limited in the amount of splitting allowed.

Summary and Recommendations

Ibegan this study with a question: “Is income splitting fair?” To answer, I developed a framework for assessing fairness, centred on the concept of horizontal equity and taking “equals” to be taxpayers with the same real living standards. I assessed the key factors needed to translate money income into real living standards and considered how these factors apply to the major types of income. The translation is affected by family size and work preferences. The differential opportunity of some groups to avoid taxes also enters the analysis, a factor that affects the assessment of horizontal equity for the splitting of each income type in different ways. Hence, it is not surprising that the answers to the question vary by the type of income. Table 6 shows the relevance of each factor to my assessment of splitting for each income type. My recommendations are also influenced by the criteria of gender and vertical equity, incentive, efficiency, simplicity and operational considerations. My findings take guidance from countries that use the individual as their basic tax unit for labour income but that apply variants of splitting for other types of income.

Pension income

Pension income splitting can be justified by the horizontal inequity arising from working couples’ differential access to spousal RRSPs. The self-employed and employees with little or no employer pension savings can use spousal RRSPs to undertake lifetime income splitting, but employees with good employer pension plans cannot. The introduction of pension income splitting reduces the appeal of spousal RRSPs. Any remaining horizontal inequities should be controlled by attributing back to contributors for tax purposes withdrawals prior to age 60 (rather than the current three-year window) and/or disallowing any future spousal RRSP contributions. The pension-splitting provision itself could be improved by, for example, constraining pension splitting to a 50:50 split between the spouses of their combined eligible pension income (to minimize gaming the OAS clawback), disregarding split pension income in the computation of each spouse’s OAS clawback or setting an upper bound to the amount of pension income that can be split annually (in order to constrain horizontal inequity between senior couples and senior singles).¹³⁴ Pension-income splitting also could be replaced by applying the lower-income spouse’s unused room in the lower tax brackets to the other spouse’s pension income, a preferable option for those concerned by the gender equity issue of avoiding taxation on income not controlled by a spouse.

Investment income

Investment income splitting can be justified by the horizontal inequity arising from the different use of tax avoidance mechanisms by couples with respect to shifting their savings and financial assets. Canadian couples have numerous legal means to shift their lifetime non-tax-sheltered savings to reduce their total tax burden. Attempts to constrain such actions consume tax enforcement resources and lead to higher tax planning costs and distorted investment patterns that reduce economic efficiency. All countries that use individual taxation of labour earnings have faced the difficult issue of how to tax couples’ investment income. Canada could pursue either of two solutions: impose a mandatory 50:50 allocation rule on couples’ total investment income or permit the transfer of financial assets between spouses and tax investment income to the spouse who holds legal title to the asset. The former approach would be simpler, avoid the costs and distortions of tax and financial planning by couples and allow couples to focus on maximizing their investment returns rather than minimizing their tax burden. But if gender equity is a primary concern, the latter approach is preferable because it would encourage the transfer of property to lower-earning spouses and would not impose a tax liability on spouses who do not control the income flow.

Labour income

Given its predominance in the budgets of most nonelderly households, labour income is pivotal in any analysis of income splitting. Unlike the horizontal equity assessment of pension and investment income splitting, differential opportunities for tax avoidance are not a major consideration for labour income outside the limited realm of the self-employed and proprietors. My analysis of the horizontal equity of labour income splitting hinges on three cases, which differ in terms of how much time the couple’s second earner spends in the paid labour market versus home production versus informal market work. Any assessment of horizontal equity must adjust money income for factors such as home production and work-related costs to determine couples’ relative real living standards. The first case is that of a one-earner couple, where the at-home spouse can produce additional nonmarket goods and services and avoid the costs related to working. The notion that this couple is an “equal” for tax purposes to a two-earner couple having the same money income is quite unfounded. Rather, the one-earner couple is comparable to a two-earner couple with the additional earnings that the second earner could generate in the paid labour market. Alternatively, one needs to add the value of the at-home spouse’s additional home production and leisure to the other spouse’s market earnings to gauge the couple’s total income. For this case, the individual is the most appropriate tax unit, with no income splitting.

The second case is that of two-earner, full-time working couples with the same total income but different splits in the spouses’ earnings. Here, an equity case for some income splitting might be made. All such couples face the expenses of both spouses working and none has additional time for household production. However, this case assumes that the couple is the tax unit and that the spouses share their income fully, so horizontal equity with singles would require a joint rate schedule that reflects the scale economies of couples rather than full income splitting. This would yield marriage penalties as well as bonuses — something that Canadian advocates of splitting appear not to have contemplated. Moreover, many spouses do not fully share their income, and it would be impractical to distinguish between second earners who work full time and those who work part time at high hourly pay rates (and have more time for household production). Thus, even for this case, individual taxation is the best policy choice.

The third, and strongest, case for labour income splitting involves one-earner couples in which the second spouse remains at home to provide assistance to the earning spouse, thus enabling him/her to garner a higher salary. In this case, the couple does not have additional time in total for producing nonmarket goods and services at home. In effect, the couple has two workers, one at home providing “informal market work” and one in the market, together generating a single high salary. This situation could justify income splitting, albeit limited to the portion of the salary that could be attributed to the at-home spouse’s contribution. For the self-employed and proprietors, payments to the spouse for work performed in the business are already allowed; the counterpart payments to the assisting spouses of employees are highly restricted. Such spousal payments should be allowed, but given the incentives for an employed spouse to exaggerate the contribution of the at-home spouse and the monitoring difficulties for tax authorities, it would be better for employers to make payments directly to, and taxable to, the assisting spouse. Another problem is that allowing spouses to split income in this case but not in the other cases would fail the test of public acceptance. In short, the third case of labour income, like the other cases, is best handled by individual taxation.

The evidence on distributional effects and empirical findings on the labour supply, economic efficiency, intrafamily distribution and marriage effects of alternative tax units reinforces the choice of individual taxation of labour income on horizontal equity grounds. Many of these findings also strongly reject joint taxation or income splitting on gender equity grounds. This is a remarkable convergence of diverse types of analyses and evidence to sustain such a policy conclusion. Groups that advocate income splitting out of concern for couples who

care for young children at home would meet their objective more effectively and fairly through alternative policies. Good candidates include enriching employment insurance parental leave benefits, National Child Benefit Supplements and Universal Child Care Benefits. Much of the tax savings from labour income splitting, in contrast, would be dispersed to couples without dependent children and concentrated among a relatively small group of high-income couples.

Final thoughts

My analysis supports limited splitting of pension income and mandatory splitting of investment income or free interspousal asset transfers while retaining the individual tax unit for labour income. Labour earnings constitute the primary source of income of the great majority of non-elderly couples, so my primary message is: use the individual tax unit for labour income. In addition to equity gains among couples and between couples and singles, individual taxation of labour income promotes gender equity and horizontal equity, and it avoids the significant reduction in vertical equity that would accompany full splitting. It also avoids distortions of marital and cohabitation choices that arise through marriage tax bonuses and penalties, and it maintains incentives for second earners to enter the labour force and to choose their desired mix of household production and market work. Use of the individual tax unit further circumvents the need to adjust rate schedules for scale economies, since all taxpayers face the same schedule.¹³⁵ Finally, the individual tax unit carries additional advantages of personal autonomy and financial privacy for each taxpayer.

The proposed package of individual taxation for labour income and other methods for pension and investment income also satisfies the tax policy criteria of simplicity, operational ease and economic efficiency. It facilitates simplicity and operational ease by taxing labour income on an individual basis and by radically easing income attribution rules. Individual taxation of labour income is also more efficient than joint or split taxation because of differential work responses of married men and women. The efficiency and savings effects of pension splitting are ambiguous, but at least it should reduce tax planning costs. Investment income splitting through either of my proposed methods would decrease the effective tax rate on non-tax-sheltered investments and thus would carry the greater economic efficiency of a consumption-based personal tax. My proposals for pension and investment income might be criticized on the grounds that they would reduce vertical equity, but the loss of vertical equity would be limited given the splitting that already takes place. Any loss of vertical equity needs to be balanced against the gains in horizontal and gender equity and in simplicity and efficiency. In short, aside from limited forms of income splitting, the Canadian personal tax system should retain the individual as its basic tax unit.

1. See “Couples’ income-tax sharing should be welcomed,” editorial, Orangeville Citizen, November 30, 2006; “The case for income splitting is compelling,” editorial, Vancouver Sun, November 22, 2006, p. A22; “Income splitting should benefit all,” editorial, Calgary Herald, February 4, 2007; and “Income-splitting for all,” editorial, National Post, January 30, 2007, p. A16.
2. See “Income splitting ‘gift to the rich,’” editorial, Leader-Post (Regina), December 13, 2006; “Incomesplitting genie out of bottle,” editorial, Toronto Star, November 2, 2006, p. A6; “Swing axe slowly on income-splitting,” editorial, Edmonton Journal, February 1, 2007; and “No to income-splitting,” editorial, Times Colonist (Victoria), February 13, 2007.
3. Throughout this study, I use the term “spouse” to include common-law as well as married partners whenever referring to the Canadian tax system. The terms “wife” and “husband” also include common-law partners and the counterparts for same-sex partners, whether married or common-law. Many other countries do not treat married couples and common-law partners equally for taxation.
4. For a recent popular articulation of this view, see Taylor (2006); for a more scholarly statement, see Boessenkool and Davies (1998).
5. The inquiry concluded that the critique was based on a failure to consider the in-kind production by at-home spouses and the added work expenses of second earners (1999b, 12). Analysis by the Finance Department supported this conclusion (Canada 1999a).
6. See Riley (2006) and Taylor (2007), respectively. The “green” support for income splitting emanates from its potential to encourage less out-of-home work, more part-time and home-based work, less commuting and more volunteering and parenting. See the official statement in Green Party of Canada (2007, 64-5).
7. I consider the couple, rather than the family unit, as I do not cover issues concerning the tax treatment of dependent children or family members other than the spouse; my focus is on couples versus singles.
8. I do not cover splitting of transfer incomes, for several reasons. First, transfers targeted at those with low incomes (such as the Guaranteed Income Supplement and provincial income assistance) are based on a couple’s combined income and go mainly to nontaxable households. Second, Quebec and Canada Pension Plan benefits can already be split for tax purposes. Third, clawbacks of other transfer benefits from higherincome earners (such as Old Age Security and employment insurance) are based on individual income consistent with social insurance principles.
9. An alternative, and clearly different, notion of horizontal equity is that tax policies should not alter the ordinal ranking of individuals: the posttax ranking of individuals should be the same as the pretax ranking. This notion is less appealing in the context of income splitting. For discussion and measurement of the ranking version of horizontal equity, see Atkinson (1980); Kaplow (1989); Jenkins and Lambert (1999); Duclos and Lambert (2000); and Auerbach and Hassett (2002). Duclos provides a penetrating analysis of both notions of horizontal equity (2006).
10. In a more recent rendition, “large differences [in tax burdens] among similar individuals…might be viewed as intrinsically arbitrary, and therefore more costly to the social fabric” (Auerbach and Hassett 2002, 1117). See Ravallion for extensive discussion of how horizontal inequity can undermine requisite public support for development policies (2004); similar points apply to tax and other policies in developed economies.
11. However, some quantitative guidance can be taken from the relative numbers in the various groups and the relative magnitudes of the horizontal inequities reduced and increased. That is, if a provision favouring one group cannot be removed, then extending similar favourable treatment to another group might improve horizontal equity overall, even if it leaves still another group disadvantaged. Note that this problem is analogous to the problem of “second best” in economic theory: if one distortion cannot be removed, it might be economically more efficient to increase other distortions rather than to reduce them.
12. I ignore the fact that the base of the personal “income” tax in Canada is much closer to consumption than income — because of the provisions for tax-sheltered retirement savings and tax-free capital gains on homes. For discussion of horizontal equity for these two tax bases, see Musgrave (1976, 9-13) and Feldstein (1976, 87-9). McIntyre explicitly considers tax-unit issues for consumption-based personal taxes (1990).
13. Similarly, Kaplow asserts, “[horizontal equity] demands equal treatment only when there is no legitimate basis for [unequal treatment]” (1989, 149).
14. My development of the horizontal equity implications of heterogeneous preferences, costs and options in the next subsections builds on general concepts in Feldstein (1976); Musgrave (1976); and Duclos (2006).
15. Other aspects of needs or cost of living are differentials by location, age or health status, which are not relevant to the issue of income splitting.
16. For a volume devoted to studies of equivalence-scale methodology and application, see Dagum and Ferrari (2004); for studies that use the concept in assessing the Canadian income distribution, see Green and Kesselman (2006). Note that this literature also considers families of differing composition between adults and children.
17. The role of equivalence scales in assessing the horizontal equity of married joint tax-filing provisions in the US income tax system is carefully assessed in Auerbach and Hassett (2002). McIntyre argues that scale economies should be disregarded in the taxation of couples (1988, 197-8); see also Minarik (1983). Bittker discusses how scale economies are likely to vary by a couple’s income level and are especially difficult to assess for very high incomes (1975, 1422-5). The Law Commission of Canada asserts that it is inappropriate for tax policies to account for scale economies enjoyed by married couples differently than those arising from other relationships (2001, 80-1).
18. The concept of “potential” or “full” income is commonly used in economic analysis to measure the earnings of an individual, standardized for full-time work.
19. As Le Grand has observed, “realized income has an important characteristic that makes it unsuitable for equity purposes: its voluntary nature” (1987, 435).
20. Common examples would be the differential taxation of particular types of income such as dividends and capital gains on equities to reflect the fact that the income has already been taxed at the corporate level. The relationship among tax-avoidance opportunities, tax complexity and horizontal equity is developed at length in Kesselman (1994). Here, I use “tax avoidance” to refer to both legal tax planning and activities that extend beyond the law — such as self-employed persons’ excessive splitting of business income with a spouse — but that are difficult for the tax authorities to monitor or control.
21. For US research on income splitting in which gender equity issues are highlighted, see Gann (1980); Davis (1988); Jones (1988); Kornhauser (1993); and McCaffery (1993, 1997); for Canadian studies, see Maloney (1994); Brooks (1996); Lahey (2000, 2005); Freiler, Stairs and Kitchen (2001); Law Commission of Canada (2001); Philipps (2002); Samtani (2006); and Woolley (2007).
22. For the “control” approach, see Dulude (1979); Kornhauser (1993); and Brooks (1996); for the “benefit” approach, see McIntyre and Oldman (1977); Minarik (1983); and McIntyre (1996); and for a critical assessment of these approaches, see Head (1996).
23. Davis (1988) and Law Commission of Canada (2001), among others, have noted not only that full sharing is lacking among many couples but also that sharing is found in many other nonconjugal relationships that would equally warrant income splitting if it were allowed for couples. Philipps notes that the transfer of legal title over assets between spouses does not necessarily mean transfer of effective control (2007).
24. The control approach relates more closely to the traditional definition of income as an individual’s consumption plus change in net worth. In contrast, the benefit approach is more compatible with the use of a consumption base for the direct personal tax. The principal means by which the tax base is transformed from income to consumption are: tax-deferred savings in Registered Retirement Savings Plans and Registered Pension Plans, tax-free capital gains on home equity and nontaxation of the imputed rental value of owneroccupied housing and favourable tax treatment of dividend income and capital gains. See Kesselman (2004) for further analysis.
25. Here, I do not consider tax-unit issues with respect to dependent children or other family members. For studies that focus on the appropriate principles for taxing families, including children, see Boessenkool and Davies (1998) and Vincent and Woolley (2000). Tax avoidance by splitting incomes with children is another policy concern; see Donnelly, Magee and Young (2000).
26. Note that the rate schedule would be twice as wide because, with scale economies ignored, a couple is assumed to contain two adult-equivalents. A third method of full income splitting is to allow transfers of unused room in the lower brackets between spouses; as I discuss later, a limited form of this method has been proposed as an alternative to pension income splitting.
27. Intuitively, once the spouses have shifted enough income that they are in the same rate bracket, any further shifting will change the allocation of taxes between them but not their total taxes.
28. Technically stated, the lower earner’s marginal tax rate will increase from splitting if and only if the couple’s average income (I) equals or exceeds the top end of the lower earner’s tax bracket.
29. This approach facilitates the analytical comparisons with the singles’ tax rate schedule. Joint taxation systems typically express their rate schedules as applicable to the spouses’ combined incomes and use a different rate schedule for single filers.
30. In contrast, Belgium and Greece use individual taxation but require couples to file a joint tax return.
31. Callan aptly draws this distinction in the context of reforms to the Irish tax treatment of couples (2006).
32. This account of US policy history relies most heavily on Surrey (1948); Groves (1963); Bittker (1975); and McCaffery (1997, chaps. 2 and 3), and secondarily on Brazer (1980); Gann (1980); Munnell (1980); and O’Neill (1983). Jones offers a fascinating gendered analysis of the social and attitudinal changes surrounding the adoption of income splitting in the United States (1988).
33. For example, if provisions such as the standard deduction, zero-bracket amounts, personal credits or capital gains loss offsets for married couples are less than twice those for single filers, additional marriage penalties can arise, while other provisions can create marriage bonuses; see United States (1996; 1997, 15-25). McCaffery also cites as a marriage bonus the US taxfree treatment of employee fringe benefits that extend to the employee’s spouse (1993, 1010-3); Lahey reiterates this theme in the Canadian context (2005).
34. Marriage bonuses and penalties and disincentives for second workers in US income security programs often exceed those of the income tax. Most noteworthy are provisions in the US social security program for spousal retirement and survivor benefits (Boskin and Puffert 1987), and the earned income tax credit (Holtzblatt and Rebelein 2000). Canada does not have comparable issues in its Canada Pension Plan retirement benefits, but the National Child Benefit Supplement does contain similar penalties and disincentives. Since Canada treats married and commonlaw couples identically, the family effects relate to cohabitation and truthful reporting of conjugal status rather than marriage per se. Welfare programs have even more extreme marriage penalties and work disincentives for second earners.
35. From the outset it was established law that couples could use interspousal gifts of income-producing assets to shift the tax liability so long as they were willing to relinquish control over the assets (Bittker 1975, 1403).
36. Top marginal tax rates rose to above 90 percent during the 1940s, and coverage of the adult population by the US federal income tax rose from just 5 percent in 1939 to 74 percent in 1945, while income tax revenues rose from 1.2 percent to 10.0 percent of personal income over that period (Munnell 1980, 255).
37. See the account in Bittker (1975, 1408-10, 1412) and Groves (1963, 63). In a repeated attempt to pass this legislation in 1942, a provision for relief on the earnings of a second spouse was added.
38. The reform fell slightly short of full income splitting on account of a standard deduction of $1,000 per filer regardless of whether it was an individual or a couple filing jointly; previously, the standard deduction had been $500 per return, but a couple filing two individual returns effectively obtained a total of $1,000. Surrey states that this fixed standard deduction “can be accepted as a device to compensate for these economies [of scale for a couple’s living costs]” (1948, 1108).
39. Groves attributed the origin of this income-splitting method to Fabian Society leader Sidney Webb (1963, 64; see also Webb 1916, 236-8). See my later description of the United Kingdom’s system of tax aggregation for that period.
40. Federal tax law defers to state law on the issue of eligibility for married joint-filing status; only legally married couples plus partners with a common-law marriage recognized in the state where the couple resides are eligible. Currently just 10 states plus the District of Columbia recognize common-law marriages contracted within their borders, while another five states recognize such relations if contracted prior to a specified date. Same-sex couples cannot file jointly for federal tax purposes under any circumstances.
41. Some observers also attributed a congressional intent to recognize the scale economies enjoyed by couples; compare Gann (1980, 28-9); Cohen (1983, 30); and McIntyre and McIntyre (1999, 919).
42. See Gann (1980, 22), and also compare this with the results of my later computations for a Canadian system of partial income splitting, with couples assigned 1.5 adult-equivalents, in table 5.
43. See, for example, Rosen (1976); Feenberg (1983); Feenberg and Rosen (1983, 1995); and Eissa and Hoynes (2000). For the most extensive analysis of marriage bonuses and penalties, see United States (1997); for alternative ways of defining and measuring these items, see Bull, Holtzblatt and Nunns (1999); McIntyre and McIntyre (1999); and Esenwein (2001).
44. Feenberg estimates that the second-earner deduction shifted many couples from the penalty to the bonus category, produced small increases in married women’s work hours and reduced the efficiency cost of the income tax (1983). See also Gann for extensive assessment and critique of the deduction (1980, 36-9; 1983).
45. Here, I ignore the fact that the married joint filers’ standard deduction is less than double that of single filers. Rate schedules for all four filing types for tax years 2000 through 2007 are available at www.moneychimp.com/features/tax_brackets.htm
46. A more acerbic version of this concern was that “the two-job married couple has moved to the center stage in the ever changing but never ending drama entitled ‘Victims of Tax Injustice’” (Bittker 1975, 1431).
47. See Brazer (1980); Gann (1980); Munnell (1980); O’Neill (1983); Davis (1988); Kornhauser (1993); and McCaffery (1993, 1997), all of whom have been concerned with both equity and incentive issues. The most indefatigable yet nuanced defender of full income splitting among US analysts has been Michael McIntyre — see, for example, McIntyre and Oldman (1977); McIntyre (1988, 1990, 1996). See also Minarik for a compact but spirited defence of income splitting (1983).
48. Allowing spouses to file individually (using the singles’ rate schedule) would eliminate all marriage penalties and leave only marriage bonuses. For an assessment and costing of this approach, see United States (1997, 97); Bartlett notes its complexities (1998, 15-6).
49. Some countries, such as Portugal and Spain, that offer variants of joint taxation for married couples require individual taxation for cohabiting couples; see O’Donoghue and Sutherland (1999, 597). See also Sommerhalder for a detailed description and history of the tax treatment of couples and families in selected European countries (1996).
50. Lahey who tracks the history of the choice of tax unit in six countries, concludes that “joint models of the tax unit tend to be replaced with individual models as relationships recognized in law become more diverse, as sex roles become more egalitarian, as social values become less ‘traditional’ and [more] inclusive, as women gain direct legal interests in incomes and assets, and as the state assumes more responsibility for the welfare of children” (2000, 25).
51. Even before its 1991 tax reform, Sweden taxed investment income jointly (and separately from labour earnings) because of concerns about tax avoidance and, according to a Swedish analyst cited in Dulude because “the introduction of completely separate taxation [for such income was] not deemed necessary from a labour market point of view” (1985, 81).
52. In the 2006 tax year, the standard tax bracket (20 percent rate) was 32,000 for single taxpayers, 41,000 for one-earner couples and up to 64,000 for twoearner couples (a minimum of 41,000, then increased by the amount of the lower earner’s income to a maximum of 23,000); any income above the standard bracket was taxed at 42 percent (OECD 2007, 257).
53. Sources for the UK system are Sommerhalder (1996) and Stephens and Ward-Batts (2004); see also Dulude for a more detailed history (1985, 76-9).
54. This applies to income from real property as well as financial income in the form of interest and dividends; the United Kingdom imposes a separate tax on capital gains.
55. My sources for the Belgian tax system are Sommerhalder (1996); OECD (2005, 61-2; 2007, 142-3); and private correspondence with Geoffroy van Kan and Lieven van Belleghem.
56. Joint filing with partial splitting combined with optional separate filing by spouses using the single rate schedule has also been considered as a reform option in the United States (see United States 1996, 36; Esenwein 2001, 13).
57. Sources for this account of the French approach are OECD (2007, 200-1) and O’Donoghue and Sutherland (1999, 573, 582-3); see also Dulude for a more detailed history (1985, 71-3).
58. Full income splitting follows the family quotient format but with a fixed quotient of 2 per couple irrespective of the presence or number of children; Portugal and Germany use this approach. Germany adopted income splitting in 1957 based on a Supreme Court finding of a constitutional requirement to ensure equality of treatment of taxpayers before and after marriage (Sommerhalder 1996, 185). France adopted its quotient system in 1945 to encourage family formation and larger families (Pechman and Engelhardt 1990, 9).
59. Sura v. Minister of National Revenue, [1957] 62 D.T.C. 1005, at 1008-9.
60. Dulude reports that the then Deputy Minister of Finance confided to her that high governmental officials were also influenced by their wives’ opposition to the proposed joint taxation (1985, 84).
61. This history is documented by Dulude (1985, 84-5); see also Lahey (2000, 40-4) for additional background
62. Some of the notable Canadian research on this topic during the intervening period includes Dulude (1985); Lahey (1988); and Maloney (1989).
63. See Maloney (1994); Boessenkool and Davies (1998); Canada (1998); Krashinsky and Cleveland (1999); Lahey (2000, 2005); Vincent and Woolley (2000); Young (2000); Law Commission of Canada (2001); and Freiler, Stairs and Kitchen (2001). Unlike the others cited here, Boessenkool and Davies conclude that joint taxation is superior on horizontal equity grounds, but ultimately they prefer individual taxation on social, behavioural and operational grounds.
64. See also Emes and Clemens, who compute how a flat tax would reduce the differential tax burdens of singleand dual-earner couples with the same total income (2001); Boessenkool (2001) and Rabushka and Veldhuis (2008) for other flat tax proponents citing the inequity for single-earner couples; and my later discussion of the flat tax approach.
65. These groups include the Institute of Marriage and Family Canada, the Work Research Foundation, the Institute for Canadian Values and REAL Women of Canada; see Coggins (2007) and the Web sites of the individual organizations. The House of Commons Standing Committee on Finance also lists groups testifying in support of income splitting (Canada 2006b, 42). These groups have focused on the system’s alleged inequities for single-earner couples with a stay-athome parent and for dual-earner couples with identical total incomes but different splits.
66. All the credit amounts cited in this paragraph must be multiplied by the federal bottom-bracket marginal tax rate to obtain the federal tax reduction (that rate was 15.25 percent in the 2006 tax year and was retroactively reduced to 15 percent for 2007). Also, these results are compounded by provincial income taxes, although the amounts of basic and spousal credits vary by province.
67. The lower-earning spouse was still required to file a separate tax return. This is often advantageous in any event in order for that individual to claim a refund of withheld taxes and to establish eligibility for various refundable tax credits (federal and provincial) that are conditioned on the joint incomes of couples.
68. Until 1980, Canadian tax law prohibited deductions even for payments reflecting the actual value of work performed by a wife in the taxpayer husband’s sole proprietorship or partnership; these payments were attributed back to the husband for tax purposes (Lahey 2000, 29, n. 41).
69. Another splitting device, available for owner-managers to split both capital and labour returns, is the Employee Profit Sharing Plan. For longer lists of income-splitting opportunities for couples (and the use of children for splitting), as well as the income attribution rules used by the authorities in attempts to control splitting, see Donnelly, Magee and Young (2000), and Fiscal Agents Financial Services Group (2007).
70. With a rising proportion of married women working, this method has become feasible for more couples even though the tax savings from splitting have diminished with growing equality in the partners’ earnings.
71. The Finance department Web site explains this age differentiation as targeting the pension income credit to retired individuals: “individuals have much greater personal control over the timing of withdrawals under RRSPs, RRIFs and LIFs compared to RPPs” (Canada 2007b).
72. The OAS clawback operates in addition to the taxability of OAS benefits. While OAS benefits are paid out universally to all longer-term residents of Canada beginning at age 65, they are phased out incrementally according to an individual’s income beyond the $63,511 threshold (for 2007) and fully eliminated when income reaches $102,865. Shifting income between spouses can reduce the OAS clawback for one by more than it increases the OAS clawback for the other, thus raising the couple’s total net income.
73. The OAS clawback is imposed at the steep rate of 15 percent of the individual’s net income beyond the threshold amount. Even after considering the taxability of OAS, the effective marginal clawback rate of 9 to 10 percent is larger than any of the interbracket tax rate differentials, while income above $102,865 remains in the same marginal rate bracket. In contrast, the phaseout rate for the age tax credit amount (not the credit itself) is 15 percent for net income between $30,936 and $65,449 in 2007, but the impact on the individual’s effective marginal tax rate is just 2.25 percent (that is, 15 percent times the bottom-bracket rate of 15 percent).
74. The group listed 23 separate organizations belonging to the Common Front for Pension Splitting, mostly retiree and pensioner groups (such as Police Retirees of Ontario).
75. Interestingly, neither the Canada Revenue Agency nor Statistics Canada publishes figures on the extent or distribution of spousal RRSP contributions; these are amalgamated with regular RRSP contributions.
76. Note that CPP/QPP benefit splitting is restricted to benefits arising from contributions made while the spouses were living together, and both spouses must be at least age 60.
77. A companion policy change could be to reduce the age limit threshold from 65 to 60 for splitting of pensiontype incomes other than annuities from RPPs.
78. This choice would not be an uncommon one in cases where the spouse with the higher pension income also has more income from other sources (such as investments or employment) than the other spouse.
79. This case is a concrete illustration of my earlier discussion of the problem of simultaneously improving horizontal equity between B and C but worsening it between A and B, and assessing the net change in horizontal equity.
80. See Kesselman and Poschmann for parallel analysis of the intertemporal inefficiencies that can arise with taxdeferred savings plans such as RPPs and RRSPs (2001).
81. Laurin reports the distributional effect by income class (2006), but these results need to be adjusted to account for the 2007 federal budget measure raising the spousal tax credit to the level of the basic credit, which reduces the gains from pension splitting and particularly eliminates many gainers at low and moderate incomes.
82. Feminist tax law scholars have criticized the provision for similar reasons (see, for example, Philipps 2006). More than a quarter-century ago, Moerschbaecher argued that “no valid tax policy or reason exists for allowing a hypothetical income split with a spouse, usually a wife, when no real split of income or control of assets has occurred” (1981, 135).
83. By law, the pensioner cannot assign the right to receive pension benefits except upon death (and then normally to the surviving spouse).
84. Extending this proposal to full shifting of unused tax room in all brackets would add to the complexity of tax computations, but these could be easily handled by a supplemental worksheet and tax preparation software.
85. This includes income from financial assets of publicly traded or arm’s-length entities (interest, dividends and capital gains) and from real property (net rents and capital gains). It excludes the returns from unincorporated businesses and incorporated but closely held businesses, which combine labour and capital incomes of the owners; these are treated in part in my section on labour income splitting.
86. At one time, the UK practice was to tax all of a couple’s property income to the husband irrespective of which spouse actually owned the assets (Brazer 1980, 243).
87. This approach was also proposed for the United Kingdom by the Meade Committee (1978, 385-90). In Belgium, it applies only to couples married with “shared patrimonium”; spouses married with “separate patrimonium” are taxed on the income of investments they own individually or their share of jointly owned assets. This situation is analogous to the pre-1948 distinctions between community-property and common-law states in the United States.
88. Another, more radical solution would be that of the Nordic countries’ dual tax, which combines a flat rate tax on investment income with progressive rates on individual labour earnings; see Sørensen (2007). The US federal income tax since 2001 has similarly applied a flat rate (15 percent) on long-term capital gains and dividend incomes of most taxpayers above the lowest brackets.
89. Munnell observes: “Although taxing the couple for unearned income [as in a 50:50 allocation rule] would violate the principle of individual taxation, the administrative difficulties probably outweigh the desirability of individual taxation in this area” (1980, 274). Note that tax planners might still seek to allocate more than 50 percent of a couple’s total investment income to the lower-earning spouse — for example, by characterizing rent from real property as business income instead of property income.
90. An important caveat here is the need for a companion anti-avoidance rule that would allow the tax authority to disregard an interspousal transfer if the transferee did not acquire beneficial ownership of the asset as well as legal title.
91. Currently, in Canada, interspousal asset transfers are nontaxable unless a special election is made; gains realized on the subsequent sale of the assets are attributed to the transferor.
92. Asset shifting in the United Kingdom under a similar reform was found to be considerably less than one would have predicted based on the minimization of joint taxes by couples (Stephens and Ward-Batts 2004). Shifting of assets might be much more pronounced under this reform in Canada, which, unlike the United Kingdom, imposes the splitting of assets upon dissolution of marriage.
93. Because income splitting could be achieved through simple, direct transfers, this would eliminate many corporate, trust and other complex intermediary structures currently used to circumvent the attribution rules.
94. See, for example, Dulude (1985); Brooks (1996); Lahey (2000); Vincent and Woolley (2000); Young (2000); and Philipps (2002).
95. Lahey similarly suggests the need for “tests designed to identify genuine economic autonomy” but does not offer an operational proposal (2000, 119). Head however, criticizes such an approach, noting that “[i]t remains unclear how the true pattern of ‘control’ could ever be determined in practice or made the basis for the design of a feasible tax unit system” (1996, 202).
96. Brooks expresses similar views: “recognizing the autonomy of women for tax purposes is important, regardless of the distributive consequences across income classes” (1996, 74). Lahey argues that “after economic relations between men and women have been equalized, we can take up the issue of equalizing them between women” (2000, 300).
97. The maximum savings would be less for an elderly couple on account of the impact on OAS clawbacks; also, the stated figures apply to most forms of taxable income other than dividends.
98. As I described earlier, increased splitting could increase the marginal tax rate of the lower-earning spouse, which could have adverse effects on that spouse’s incentives to undertake paid work.
99. Owners of unincorporated and incorporated businesses can also use Employee Profit Sharing Plans (EPSPs) to shift labour earnings to a spouse; a recent Tax Court of Canada ruling has upheld the ability of EPSPs to avoid payment of CPP premiums on the payments (Greber v. Minister of National Revenue, [2007 TCC 78]) .
100. Schuetze estimates that just one out of 18 selfemployed men engaged in this activity; he found no evidence of any illegal shifting of income by selfemployed women to their spouses (2006). I arrive at the figure cited in the text based on these findings plus the 15 percent of the labour force that is self-employed.
101. This outcome follows the findings of my general equilibrium economic analysis of income tax evasion (Kesselman 1989).
102. For estimates of the differential effects on after-tax incomes of one-earner versus two-earner couples with thesametotalincomeatdifferentincomelevelsandfor each OECD country in 2002, see OECD (2005, 55, 58-60).
103. As Lazear and Michael note, “[i]n order to make comparisons across households, however, it is important to understand the mapping from dollar income into some notion of ‘real income’ and how that mapping is affected when one moves from one-earner to twoearner families” (1980, 203). O’Neill observes: “Since one-earner couples systematically have a larger component of unmeasured income than two-earner couples, a policy that seeks to equalize the tax burdens of couples whose pooled money incomes are the same violates the more fundamental principle of ability to pay.” O’Neill also cites supportive contemporary studies on the value of household production by working and nonworking wives and on work-related expenses (1983, 6-9). Krashinsky and Cleveland provide detailed examples to demonstrate the horizontal equity issues involved with respect to one-earner versus two-earner couples in Canada; they also show the important role played by the child care expense deduction (1999).
104. Some analysts have argued that it is not proper to consider the imputed value of home production under an income tax and, hence, the differential opportunities for home production between oneand two-earner couples; see, for example, McIntyre and Oldman (1977, 1609-18).
105. See OECD for a similar definition of the “secondary worker” (2005, 37).
106. The study’s sample included all families with an oldest child under age 18 and second earners who worked either part time or full time. The results would have been even more dramatic if it had focused on families with preschool children and full-time second earners.
107. The value of the incremental household production exceeds the individual’s potential net-of-tax labour market earnings but not necessarily potential gross (before-tax) earnings; thus, the tax system still exerts an economic inefficiency. Vincent and Woolley detail alternative methods of gauging the value of an at-home spouse’s household production (2000, 35-6).
108. See Kornhauser (1993); Cantillon and Nolan (1998); Freiler, Stairs and Kitchen (2001); and studies cited therein.
109. Analysts have long tried to determine why married men earn more, on average, than unmarried men — does being married raise a man’s earnings or are higherearning men more marriageable? Antonovics and Town offer an empirical analysis using a sample of monozygotic male twins (to eliminate nonobservable interpersonal differences). They find that marriage does cause men’s earnings to rise, although they do not address the issue of whether having a wife who stays out of the labour force to provide supportive services raises the earnings of married men (2004).
110. The quotation is from Philipps (2007), summarizing a passage in Kanter (1977, 110-11).
111. Thissituationisaptlycharacterizedinajudicialdecision: “The company…does not hire an applicant as a salesman…if his wife does not meet the company standards… Canada Life regards the combination of the husband and wife as the selling unit in the business and takes active steps to foster the wife’s participation, but they do not pay her”; see Hale v. Canada, [1969] 1 Ex.C.R. 259 (Exchequer Court), at 261. Bromwich (2007), taking a self-described “alternative feminist perspective,” makes essentially this case for income splitting. However, she focuses on the at-home mother’s assuming disproportionate household and parenting responsibilities, relieving the partner to spend more time in paid employment.
112. In a recent noteworthy case, the Tax Court of Canada upheld the deductibility of an employee’s payments to his wife as an assistant; the taxpayer was a salesman required by an employment contract to pay his own expenses. See Longin v. The Queen, [2006 TCC 335].
113. Note that the $700 million attributed to elderly couples somewhat exceeds the $675 million federal revenue cost of pension splitting alone because it applies to all incomes received by elderly couples.
114. However, the additional tax savings at the provincial level would not be proportional to federal tax savings, because each province imposes its own rate schedule with different threshold levels for the rate brackets. None of the provinces has a top income bracket threshold as high as the federal one. Carson and Morris compute the maximum annual tax savings for a nonelderly single-income couple from fully shifting $115,739 of taxable income to the second spouse at $14,367, of which $8,652 is federal tax savings and the balance is Ontario tax savings net of the couple’s Ontario health tax liability (2005, tables 1 and 2).
115. Note that, unlike in figure 2, I have not applied the equivalence scaling to the basic filer or spousal tax credit but left them at their full values for joint filers.
116. Note that the maximum tax savings per couple arise with a 0:100 earnings split at total earnings of $182,000 and higher. For other earnings splits, the tax costs grow beyond the tabulated figures for total earnings above $182,000; at $242,000 with a 50:50 split, the tax cost becomes –$4,602.
117. See reviews of this evolving research in Mroz (1987); Heckman (1993); Eissa (1996); Blundell and MaCurdy (1999); and Heim (2005). Heckman observed: “Whether labor supply behavior by sex will converge to equality as female labor-force participation continues to increase is an open question” (1993, 118).
118. Of course, caution is needed in interpreting the findings from other countries with widely varying social policies such as subsidized child care that might interact with tax policies in labour market effects.
119. McCaffery refers extensively to the optimal tax literature and draws inferences for family tax policies (1997, 16799). As he noted in an earlier study, the optimal taxation framework neatly encompasses the issue of household production via each spouse’s labour supply response (1993, 1040). Kaplow offers an interesting utilitarian analysis that focuses on alternative assumptions about how resources are shared within the household and which members benefit from scale economies (1996)
120. This strand of research assumes a unitary utility function for couples that does not differentiate the well-being of the individual spouses.
121. Their analysis assumes that households have identical preferences (with Stone-Geary utility functions) and that tax schedules are linear (but differing between the primary and secondary earner). The covariance between the wage rates of the husband and wife also enters the analysis (so-called assortative mating).
122. Their analysis uses specific functional forms for the couple’s utility and for household production; it also takes a single representative household and thus avoids distributional considerations. Their quantitative findings for Australia show efficiency differences across the alternative tax treatments of couples amounting to only a small fraction of 1 percent of incomes.
123. See Apps and Rees (1999a) and Gottfried and Richter (1999). One cited deficiency is that the analysis makes discrete comparisons between joint and individual taxation without allowing for intermediate results with smaller but positive differences in the tax rates on the spouses. Criticisms also arise with the assumptions made in the authors’ quantitative work. For the authors’ responses, see Piggott and Whalley (1999).
124. Their model does not incorporate the household production choices of the Piggott-Whalley model. Like the other models, theirs also has restrictive aspects such as assuming no income effects on labour supply, separability in the disutility of labour for the two spouses and fixed costs of working for the secondary earner that are distributed independently of that spouse’s wage rate.
125. The authors contend that gender-based tax rates would be a superior substitute for other policies, such as affirmative action, that also discriminate by sex. One should note that their analysis proceeds as if all women and men are married and ignores the empirical finding that the labour supply elasticity of single women is very similar to that of men. Hence, their policy would need to discriminate even further between married and single women.
126. A similar policy introduced in Canada would likely stimulate more extensive asset shifting. Unlike in the United Kingdom, when Canadians divorce, there is an automatic 50:50 division of property accumulated during the marriage. Therefore, Canadian husbands might be less reluctant to make such transfers to their wives.
127. The importance of these operational features should not be underrated. For example, Gustaffson describes the West German system that withheld tax at source from the “main breadwinner” using a low rate schedule assuming that he or she was the only earner, while the second worker’s earnings had taxes withheld using a high rate schedule that reflected the tax on incremental earnings (1992, 70).
128. For a broad review of the effect of marriage penalties from both the tax and transfer systems on marital behaviour, see Alm, Dickert-Conlin and Whittington (1999). Chade and Ventura offer an interesting theoretical analysis that includes assortative mating behaviour; they find that a switch from US-style joint taxation to individual taxation would decrease the correlation of the partners’ productivities as measured by wage rates but, because of the pattern of labour supply effects, increase the correlation of their labour earnings (2002).
129. For example, the Library of Parliament analyses of income splitting (Bergevin, Laurin and Kitching 2006; Laurin 2006), both done at the request of MP Garth Turner, differentiate between nonelderly couples with and without children; the Institute for Canadian Values urges that any income splitting be limited to couples with dependent children (2007); while a Senate committee advocates looking at income splitting for families with autistic children in which one parent is often required to remain at home (Canada 2007d).
130. See, for example, Dulude (1985); Maloney (1994); Brooks (1996); Lahey (2000, 2005); Young (2000); Vincent and Woolley (2000); and Kershaw (2002). Woolley estimates the distributional effects of replacing the spousal credit with an enriched Canada Child Tax Benefit (2002).
131. Groves recognized this point long ago: “The issue [of the tax unit] is associated with progressive taxation and would be of little or no importance under a proportional tax” (1963, 56). Bartlett and others have advocated a flat tax in the United States, in part because it would eliminate the marriage penalty (Bartlett, 1998).
132. For a detailed critique of the “equitable family taxation” argument for a flat tax, see Kesselman (2000, 36-42).
133. The first poll, undertaken by Ipsos Reid in February 2007, found 77 percent in favour, with the strongest support in Alberta (85 percent) and Quebec (77 percent) (Aubry 2007). The second poll, undertaken by the Strategic Counsel in March 2007, found 74 percent in support; 16 percent opposed; 10 percent “don’t know” (2007, 11).
134. Under the last option, it would be essential to prohibit any future spousal RRSP contributions.
135. As Brennan and Brooks argue, it is economically inefficient for the tax system to tax the economic gains from scale economies of shared living expenses (1983, 125-6).

Alesina, Alberto, and Andrea Ichino. 2007. “Gender Based Taxation.” Draft paper, May 25. Cambridge, MA: Department of Economics, Harvard University.

Alm, James, Stacy Dickert-Conlin, and Leslie A. Whittington. 1999. “Policy Watch: The Marriage Penalty.” Journal of Economic Perspectives 13 (3): 193-204.

Alm, James, and Leslie A. Whittington. 1995. “Does the Income Tax Affect Marital Decisions?” National Tax Journal 48 (4): 565-72.

Antonovics, Kate, and Robert Town. 2004. “Are All the Good Men Married? Uncovering the Sources of the Marital Wage Premium.” American Economic Review 94 (2): 317-21.

Apps, Patricia, and Ray Rees. 1999a. “Individual versus Joint Taxation in Models with Household Production: Comment.” Journal of Political Economy 107 (2): 393-403.

_____. 1999b. “On the Taxation of Trade within and between Households.” Journal of Public Economics 73 (2): 241-63.

Atkinson, A.B. 1980. “Horizontal Equity and the Distribution of the Tax Burden.” In The Economics of Taxation, edited by Henry J. Aaron and Michael J. Boskin. Washington, DC: Brookings Institution.

Atkinson, Anthony B., and Joseph E. Stiglitz. 1980. Lectures on Public Economics. New York: McGraw-Hill.

Aubry, Jack. 2007. “Canadians overwhelmingly support income splitting: Poll.” canada.com, February 26. www.canada.com

Auerbach, Alan J., and Kevin A. Hassett. 2002. “A New Measure of Horizontal Equity.” American Economic Review 94 (4): 1116-25.

Bailey, Sue. 2007. “Income-splitting warning issued.” Winnipeg Free Press, January 30.

Bartlett, Bruce. 1998. “The Marriage Penalty.” Policy Backgrounder 145. Dallas: National Center for Policy Analysis.

Bergevin, Philippe, Alexandre Laurin, and Andrew Kitching. 2006. “Allowing Income Splitting for Two-Parent Families with Children and for Senior Couples.” Ottawa: Library of Parliament, Parliamentary Information and Research Service.

Bittker, Boris I. 1975. “Federal Income Taxation and the Family.” Stanford Law Review 27 (6): 1389-463.

Blundell, Richard W., and Thomas MaCurdy. 1999. “Labor Supply: A Review of Alternative Approaches.” In Handbook of Labor Economics, vol. 3A, edited by Orley C. Ashenfelter and David Card. Amsterdam: Elsevier Science.

Boessenkool, Kenneth J. 2001. “Give Mom and Dad a Break: Family-Friendly Tax Options for Canada.” Policy Options 22 (8): 30-4. Montreal: IRPP.

Boessenkool, Kenneth J., and James B. Davies. 1998. “Giving Mom and Dad a Break: Returning Fairness to Families in Canada’s Tax and Transfer System.” C.D. Howe Institute Commentary 117. Toronto: C.D. Howe Institute.

Boskin, Michael J., and Douglas J. Puffert. 1987. “Social Security and the American Family.” In Tax Policy a the Economy, vol. 1, edited by Lawrence H. Summers. Cambridge, MA: National Bureau of Economic Research, MIT Press.

Boskin, Michael J., and Eytan Sheshinski. 1983. “Optimal Tax Treatment of the Family: Married Couples.” Journal of Public Economics 20 (3): 281-97.

Brazer, Harvey E. 1980. “Income Tax Treatment of the Family.” In The Economics of Taxation, edited by Henry J. Aaron and Michael J. Boskin. Washington, DC: Brookings Institution.

Brennan, Geoffrey, and Michael Brooks. 1983. “Towards a Theory of Family Taxation: The Equity Dimension.” In Taxation Issues of the 1980s, edited by John G. Head. Sydney: Australian Tax Research Foundation.

Brett, Craig. 2007. “Optimal Nonlinear Taxes for Families.” International Tax and Public Finance 14 (3): 225-61.

Bromwich, Rebecca. 2007. “Income Splitting — An Alternative Feminist Perspective.” Paper presented at the conference “It’s Fair to Share: Reducing Taxes in Canada by Sharing Income,” January 30, Ottawa.

Brooks, Neil. 1996. “The Irrelevance of Conjugal Relationships in Assessing Tax Liability.” In Tax Units and the Tax Rate Scale, edited by John G. Head and Richard Krever. Sydney: Australian Tax Research Foundation.

Bull, Nicholas, Janet Holtzblatt, and James R. Nunns. 1999. “Defining and Measuring Marriage Penalties and Bonuses.” OTA Working Paper 82. Washington, DC: US Treasury Department, Office of Tax Analysis.

Callan, Tim. 2006. “Gender and Income Tax in Ireland: A Tale of Three Systems.” Paper presented at the conference “Gender, Tax Policies, and Tax Reform in Comparative Perspective,” Levy Economics Institute, Bard College, May 17-18, New York.

Canada. 1966. Royal Commission on Taxation Carter Commission. 1966. Report, vol. III. Ottawa: Queen’s Printer.

_____. 1969. Department of Finance. Proposals for Tax Reform. Ottawa: Queen’s Printer.

_____. 1970. Royal Commission on the Status of Women. Report. Ottawa: Queen’s Printer.

_____. 1998. Status of Women Canada. Gender-Based Analysis: A Guide for Policy-Making, revised edition. Ottawa.

_____. 1999a. Department of Finance. Presentation to the Standing Committee on Finance, Sub-committee on Tax Equity for Canadian Families with Dependent Children. Ottawa, April 21.

_____. 1999b. Parliament. House of Commons. Standing Committee on Finance. For the Benefit of Our Children: Improving Tax Fairness. Ottawa.

_____. 2003. Health Canada. Exploring Concepts of Gender and Health. Ottawa: Health Canada, Women’s Health Bureau.

_____. 2006a. Expert Panel on Accountability Mechanisms for Gender Equality. Equality for Women: Beyond the Illusion, Final Report. Ottawa.

_____. 2006b. Parliament. House of Commons. Standing Committee on Finance. Canada: Competing to Win. Ottawa.

_____. 2007a. Department of Finance. The Budget Speech 2007: Aspire to a Stronger, Safer, Better Canada. Ottawa, March 19.

_____. 2007b. Department of Finance. Qs and As on the Tax Fairness Plan. Ottawa. www.fin.gc.ca/pensioncalc/QA_e.html.

_____. 2007c. Department of Finance. Economic Statement. Ottawa, October 30.

_____. 2007d. Parliament. Senate. Standing Committee on Social Affairs, Science and Technology. Pay Now or Pay Later: Autism Families in Crisis. Ottawa.

Canadian Activists for Pension Splitting. 2006. CAPS Position Paper. www.pensionsplitting.ca.

Canadian Alliance Party. 2000. Tax Reduction and Tax Reform. Ottawa.

Cantillon, Sara, and Brian Nolan. 1998. “Are Married Women More Deprived Than Their Husbands?” Journal of Social Policy 27 (2): 151-71.

Carson, Donald E., and Iain D. Morris. 2005. “Income-Splitting Savings.” Canadian Tax Highlights 13 (3): 6-7.

Cayo, Don. 2006. “Income splitting seemed like a good idea.” Vancouver Sun, December 29, D3.

Chade, Hector, and Gustavo Ventura. 2002. “Taxes and Marriage: A Two-Sided Search Analysis.” International Economic Review 43 (3): 955-85.

Coggins, Jim. 2007. “Christians Assess the Federal Budget.” Canadian Christianity.com. Christian Info Society, Langley, British Columbia. www.canadianchristianity.com/cgi-bin/na.cgi?nationalupdates/070322budget

Cohen, Edwin. 1983. “Commentary on O’Neill.” In Taxing the Family,edited by Rudolph G. Penner. Washington, DC: American Enterprise Institute.

Conservative Party of Canada. 2005. “Policy Declaration.” March 19. www.centralnova.com/policy-declaration.pdf

Cooper, Tony. 1995. “Taxing the Family Unit: Income Splitting for All?” Revenue Law Journal 5 (1): 82-99.

Coyne, Andrew. 2006. “A policy that’s fair, and it wins votes: income-splitting allows all couples to be treated the same.” National Post, November 22, A1.

Crossley, Thomas F., and Sung-Hee Jeon. 2006. “Joint Taxation and the Labour Supply of Married Women: Evidence from the Canadian Tax Reform of 1988.” QSEP Research Paper 404. Hamilton, ON: Research Institute for Quantitative Studies in Economics and Population, McMaster University.

Dagum, Camilo, and Guido Ferrari, eds. 2004. Household Behaviour, Equivalence Scales, Welfare and Poverty. Heidelberg, Germany: Physica-Verlag.

Davis, Laura Ann. 1988. “A Feminist Justification for the Adoption of an Individual Filing System.” Southern California Law Review 62 (1): 197-252.

Donnelly, Maureen, Joanne Magee, and Allister Young. 2000. “Income Splitting and the New Kiddie Tax: Major Changes for Minor Children.” Canadian Tax Journal 48 (4): 979-1018.

Drache, Arthur. 2006. “The slippery slope of income splitting: full income splitting brings philosophical, technical complexities.” Financial Post, November 28.

Duclos, Jean-Yves. 2006. “Innis Lecture: Equity and Equality.” Canadian Journal of Economics 39 (4): 1073-104.

This publication was produced under the direction of France St-Hilaire, Vice President, Research, IRPP. The manuscript was copy-edited by Barry Norris, proofreading was by Mary Williams, production was by Chantal Létourneau, art direction was by Schumacher Design and printing was by AGL Graphiques.

Copyright belongs to IRPP. To order or request permission to reprint, contact:

IRPP
1470 Peel Street, Suite 200
Montreal, Quebec H3A 1T1
Telephone: 514-985-2461
Fax: 514-985-2559
E-mail: irpp@nullirpp.org

All IRPP Choices and IRPP Policy Matters are available for download at irpp.org

To cite this document:

Kesselman, Jonathan R. 2008.  “Income  Splitting and Joint Taxation of Couples: What’s Fair?” IRPP Choices 14 (1).

About the Author

Jonathan R. Kesselman is a professor in the graduate Public Policy Program at Simon Fraser University and holds the Canada Research Chair in Public Finance. He has a BA (Hon.) in economics from Oberlin College and a Ph.D. in economics from MIT. He served as principal investigator of the Social Sciences and Humanities Research Council of Canada’s Major Collaborative Research Initiatives project on “Equality/Security/ Community.” His research focuses on applied economic theory and policy analysis of taxation, income security, social programs and social insurance. Dr. Kesselman’s work has been recognized by the Reserve Bank of Australia’s Professorial Fellowship in Economic Policy (1985), the Canadian Tax Foundation’s Douglas J. Sherbaniuk Distinguished Research Award (2002) and the Canadian Economics Association’s Doug Purvis Memorial Prize (1998 and 2007).

Acknowledgements

The author thanks many people for their input into this study but cites the following for particularly helpful comments and discussions: Don Cayo, Paul Kershaw, Ed McCaffery, Kevin Milligan, Lisa Philipps, Bill Robson, France St-Hilaire, Frances Woolley and two anonymous reviewers. Support from the  Canada  Research Chairs Program is gratefully acknowledged. All findings and any errors are the author’s sole responsibility.