The institution of marriage is an economic good for its members and wider society. This paper contributes to the first stage of a Cardus project to explore the relationship between marital status and social-assistance participation. A substantial body of research in the United States has perused the hypothesis that some social-assistance policies discourage marriage among participants. The body of Canadian research is less robust. Canadian analysis provides mixed results, but shows that regional variations such as labour market conditions and wage growth are important considerations when exploring social-assistance participation and marital decision-making. The body of research also shows that significant policy adjustments like the drastic changes during the 1990s to Ontario’s spouse-in-the-house rules can alter martial decision-making behaviour, though such outcomes may be unintentional.

Social assistance remains a well-worn policy lever for addressing poverty in Canada. Policy-makers and social scientists have shown interest in understanding how social assistance interacts with family formation, stability, and dissolution. While significant work has been completed in the United States on this issue, less work has been undertaken using Canadian data. The social-assistance system in Canada is administered through the provinces and is distinct from the American model, meaning US data cannot be assumed for Canada. This report provides an overview of Canadian literature on social assistance and marital decision-making.

Do changes in social-assistance policies influence marital decision-making? Does the decline of marriage as an institution result in increased reliance on social assistance as a replacement? Does social assistance stabilize relationships or encourage dissolution?

This report reviews the development of Canadian literature, exploring the relationship between social assistance and marital decision-making. The Canadian research story is one of refined methodologies, program evaluations, debated outcomes, and the intersection of research, policy, and political ideals.

A well-known University of Virginia sociologist, Brad Wilcox, writes that the institutional model of marriage “seeks to integrate sex, parenthood, economic cooperation and emotional intimacy into a permanent union.” 1 1 The National Marriage Project, When Marriage Disappears: The New Middle America (Charlottesville, VA: The National Marriage Project, 2010), http://stateofourunions.org/2010/SOOU2010.pdf.  This thinking captures the interlinking relational and economic elements anchored in a stable union that generates social capital and overall well-being. Today, very few people would be able to articulate this. This is not helped by the reality that the institutional model of marriage in Canada is in decline as cohabitation and lone parenthood continue to increase. The interesting thing is that here, as in other Western countries, marriage has not declined equally by income or education levels. Those with higher educational attainment and incomes are more likely to be married. 2 2 When Marriage Disappears; Philip Cross and Peter Jon Mitchell, The Marriage Gap Between Rich and Poor Canadians: How Canadians Are Split into Haves and Have-Nots Along Marriage Lines (Ottawa: Institute of Marriage and Family Canada, 2014), http://www.imfcanada.org/sites/default/files/Canadian_Marriage_Gap_FINAL_0.pdf; Fraser Nelson, “Revealed: The Marriage Gap between Britain’s Rich and Poor,” The Spectator, November 15, 2014, https://www.spectator.co.uk/2014/11/marriage-is-becoming-a-preserve-of-the-rich/; Charles Murray, Coming Apart: The State of White America, 1960–2010 (New York: Crown Forum, 2013).  In fact, the economic well-being and social capital associated with marriage has led some observers to call marriage a key poverty fighter.

As the portion of lone-parent families has increased over time, the portion of these families headed by a never-married lone parent has grown dramatically. Lone-parent families remain over-represented among those using social assistance.

Canadian research demonstrates that economic behaviour among cohabiters differs from married couples. Cohabiters are less likely to pool income. Dana Hamplová, Céline Le Bourdais, Évelyne Lapierre-Adamcyk found that unmarried couples were four times more likely to keep money separate. 3 3 Dana Hamplová, Céline Le Bourdais, and Évelyne Lapierre-Adamcyk, “Is the Cohabitation–Marriage Gap in Money Pooling Universal?,” Journal of Marriage and Family 76, no. 5 (October 2014): 983–97, https://doi.org/10.1111/jomf.12138. Money management is important, as the researchers argue: “Furthermore, the adopted system of money allocation has important implications for social inequality or poverty levels in societies given that money is often not shared equally among family members.” 4 4 Hamplová, Le Bourdais, and Lapierre-Adamcyk, 3. Money pooling behaviours may reflect the established fact that cohabitating unions are less stable than marriage.

In their recent book Cohabitation Nation, social demographer Sharon Sassler and sociologist Amanda Miller argue that moderately educated Americans have fewer resources to transition from cohabiting relationships into marriage compared to their university-educated peers. Fewer economic opportunities may motivate people to cohabitate earlier in their relationships to save money, and displace consideration for long-term relationship planning. 5 5 Laurie DeRose, “Social Class Shapes the Experience of Living Together: A Review of Cohabitation Nation,” Institute for Family Studies blog, January 9, 2018, https://ifstudies.org/blog/social-class-shapes-the-experience-of-living-together-a-review-of-cohabitation-nation. Their research suggests that there is a class structure within cohabitation patterns.

Canadians rely on social-assistance programs for numerous reasons. Academics Ron Kneebone and Katherine White argue that provincial labour markets are a significant force in social-assistance use, particularly since the 1990s. 6 6 Ronald Kneebone and Katherine White, “An Overview of Social Assistance Trends in Canada,” in Welfare Reform in Canada: Provincial Social Assistance in Comparative Perspective, ed. Daniel Béland and Pierre-Marc Daigneault (Toronto: University of Toronto Press, 2015), 53. To measure the prevalence of social-assistance use, Kneebone and White evaluate the number of social-assistance users compared to the size of the population of people zero to sixty-four years old. Examining the period between 1969 and 2012, the researchers note that social-assistance use increased significantly during the 1990s, when the economy contracted, and then declined to half of its peak use during the following period of economic growth (FIGURE 1). 7 7 Kneebone and White, 56. Kneebone and White state that rate of wage growth and employment opportunities influence provincial social assistance rates. They note that changes in social assistance policies can also increase rates of social assistance use.

National trends provide an overview of social-assistance use, but Kneebone and White argue that provincial-level analysis is important because of the significant provincial influence on welfare use. How a province is doing economically matters. While trends don’t always follow provincial economic patterns, indicators such as wage growth and employment opportunities have a significant influence on benefit reliance. 8 8 Kneebone and White, 57.

Kneebone and White state that determining the demographic characteristics of those who rely on social assistance and the various program descriptions and policy changes over an extended period can be difficult. British Columbia has among the most developed demographic information, and Kneebone and White point to a remarkable finding in that province. Between 1995 and 2012, the number and portion of two-parent and single-parent families on social assistance decreased significantly from 34.9 percent to 20.2 percent. The number of children in families receiving social assistance declined. Meanwhile the portion of single recipients increased. 9 9 Kneebone and White, 71.

Finally, though this report does not examine long-term disability, it is important to note there has been a large growth in the portion of people with disabilities who use social assistance. In British Columbia, for example, the portion of social-assistance users with a disability grew from 7.3 percent in 1995 to 55.7 percent in 2012. 10 10 Kneebone and White.

Studies examining the impact of welfare on family structure and family dissolution emerged during the 1960s and 1970s. Derek Hum and Saud Choudhry identified two perspectives present in the early studies. One perspective argued that negative economic forces such as underemployment or poverty weaken union stability, and increase the risk of dissolution. The other perspective argued that the interaction of various economic factors incentivized or dissuaded entrance into intimate partnerships. For example, increased welfare benefits or marriage penalties were believed to incentivize the growth of female-headed households. 11 11 Derek Hum and Saud Choudhry, “Income, Work and Marital Dissolution: Canadian Experimental Evidence,” Journal of Comparative Family Studies 23, no. 2 (1992): 250.

The debate during the 1970s was whether income levels influenced family stability or, conversely, whether stability created better opportunities for steady income. Researchers expanded their work to include the impact of assets and debt on family stability. Although the findings varied, researchers generally accepted that income correlated with decisions to separate or divorce. Some researchers speculated that increased welfare payments afforded women less economic reliance on a partner, incentivizing lone parenthood. A significant limitation was that many studies relied on cross-sectional data, meaning that marital-status data and income data were collected at the same point of time, frustrating the ability to determine the direction of causality. 12 12 Hum and Choudhry, 252.

Economist Douglas Allen’s 1993 study is among the first Canadian studies to estimate the impact of welfare policy on family structure. 13 13 Douglas W. Allen, “Welfare and the Family: The Canadian Experience,” Journal of Labor Economics 11, no. 1, Part 2 (January 1993): 202, https://doi.org/10.1086/298333.  Allen hypothesized that welfare policy influences work incentives and family stability. The aim of his study was to estimate the severity of the impact welfare policy has on family-structure decisions. 14 14 Allen, 202.  The economist concluded that even small changes in welfare-benefit levels influence unwed childbearing, single parenthood, and divorce. 15 15 Allen, 218.  Allen contended that increasing benefits reduces the economic benefits of partnership.

Allen argued that increasing enrollment in social assistance is not problematic if it increases prosperity. His concern was that welfare may provide disincentives to forming institutions like marriage that promote social-capital investment. This effect could cause long-term harm.

Pierre Lefebvre and Philip Merrigan’s 1997 study challenged Allen’s findings. They argued that the weakness in using cross-sectional data is that this method does not determine the direction of causality. They noted that Allen did not distinguish the timing of marital dissolution and whether respondents experienced a previous union dissolution. 16 16 Pierre Lefebvre and Philip Merrigan, “Social Assistance and Conjugal Union Dissolution in Canada: A Dynamic Analysis,” The Canadian Journal of Economics 30 (1997): 129, http://www.academia.edu/30337284/Social_Assistance_and_Conjugal_Union_Dissolution_in_Canada_A_Dynamic_Analysis. Lefebvre and Merrigan used data from the General Social Survey on Family and Friends (1990) and employed a Cox proportional hazard model.  Lefebvre and Merrigan accounted for this in their model to estimate the influence of various economic variables including social assistance on marital decision-making.

Lefebvre and Merrigan estimated the impact of earned income and social assistance on dissolution of first marriages and unions. They argued the methodology must account for a variety of changing economic variables. They found that earned income had a positive effect on union stability while social assistance had no impact on rates of relationship breakdown when a number of economic variables are considered. 17 17 Lefebvre and Merrigan, 112.  The authors concluded, “Welfare benefits do not seem sufficient to counter economic disincentives to divorce. It seems that it would be more relevant to treat welfare policy for what it truly is, that is, public child support for single-parent families. Therefore, the recently popular view within certain political circles that more generous welfare programs induce marital instability is likely mistaken in Canada.” 18 18 Lefebvre and Merrigan, 132.  Lefebvre and Merrigan’s comments show the political and policy tensions of the time. It is revealing that the authors interpret the pragmatic function of social assistance as primarily a child-support system.

Lone-parent families, and in particular lone mothers, are over-represented among social-assistance users. Lone parents are also more likely to experience longer spells of social-assistance reliance. For this reason, economists have focused on this population.

Economist Martin Dooley examined social-assistance enrollment trends among Canadian lone mothers between 1973 and 1991. His study was the first Canadian exploration of social- assistance participation to utilize a methodology relying on a time series of cross-sections. Dooley examined the rise in social-assistance participation by lone mothers during the 1970s and 1980s. He discovered a distinct difference in social assistance use by age. Dooley found increasing reliance on social assistance among lone mothers under age thirty-five, but decreasing dependence among lone mothers age thirty-five and over. Dooley’s work reveals that the growth in social assistance among lone mothers under thirty-five correlated with increasing welfare benefits relative to decreasing labour-market wages. This trend occurred during a demographic shift toward smaller families and an increasing portion of never-married lone mothers. The decrease in social-assistance use by lone mothers over the age of thirty-five correlated with increased market wages and educational attainment, and decreasing family size. 19 19 Martin D. Dooley, “The Evolution of Welfare Participation Among Canadian Lone Mothers, 1973–1991,” The Canadian Journal of Economics / Revue Canadienne d’Economique 32, no. 3 (1999): 589–612, https://doi.org/10.2307/136439.  In short, older lone mothers had access to greater educational attainment and market wages, consistent with factors that lower social-assistance participation as noted above in Kneebone and White.

Lefebvre and Merrigan examined the effect of social assistance on marital decision-making among lone mothers with particular interest in remarriage. Utilizing the Family History Survey (1990) from the General Social Survey, the authors determined that lone motherhood is not a static state. About six in ten lone mothers in the sample moved on to married or cohabiting unions.

The authors evaluated the effect of a number of sociodemographic characteristics on partnership behaviour of lone mothers, including social-assistance use. They found that provincial social-assistance regimes had an important effect on the decision by lone mothers to partner. Lefebvre and Merrigan also found that social-assistance policies benefiting couples with children increased the portion of women moving out of lone motherhood. With the understanding that lone parenthood is often not a static state, they speculated that providing robust social-assistance benefits for couples with kids could produce shorter durations of lone motherhood, meaning less funds required for social assistance for lone parents. They acknowledged that any savings would likely be offset by increased spending on couples with children and that offering robust benefits could risk creating an incentive for couples to participate in social assistance. 20 20 Pierre Lefebvre and Philip Merrigan, “The Impact of Welfare Benefits on the Conjugal Status of Single Mothers in Canada: Estimates from a Hazard Model,” The Journal of Human Resources 33, no. 3 (1998): 755, https://doi.org/10.2307/146340.

Lefebvre and Merrigan joined Dooley and Gascon for a 2000 study to examine further lone mothers in the social-assistance system. The collaboration tested two variables. They examined the impact of social-assistance benefits on lone and two-parent families with young children, and the impact of earnings opportunities for mothers and their potential partners.

The authors argued that Allen’s use of single cross-section data was problematic for the reasons noted in other studies above. The authors used time series, cross-sectional data from the Survey of Consumer Finance 1981–1993, employing a similar time-series methodology used in the influential American study by Moffitt. Moffitt had previously used single cross-section data with similar observations to the later study by Allen. 21 21 Robert Moffitt, “The Effect of the U.S. Welfare System on Marital Status,” Journal of Public Economics 41, no. 1 (1991): 101-124.  Yet when Moffitt used time-series data, the link between welfare benefits and lone motherhood appeared much weaker. 22 22 Robert Moffitt, “Welfare Effects on Female Headship with Area Effects.” The Journal of Human Resources 29, no. 2 (1994): 621-636.  The authors summarized their approach writing, “In this paper, we apply Moffitt’s test to Allen’s finding using Canadian data.” 23 23 Martin Dooley et al., “Lone Female Headship and Welfare Policy in Canada,” The Journal of Human Resources 35, no. 3 (Summer 2000): 590.

When the authors applied provincial fixed effects to their time-series sample, they found the coefficients for welfare benefits were small and statistically insignificant. They argued that socioeconomic factors like education, age, and wages appear to be a more significant influence. Finally, the authors acknowledged that research in this area in Canada was still in the early stages, and they looked forward to the potential use of longitudinal data from the Survey of Labour and Income Dynamics (SLID). 24 24 Dooley et al., “Lone Female Headship.”

Several social-assistance pilot projects conducted over the previous decades provided opportunities to explore the impact of policy alternatives on marital decision-making.

In the early 1990s, Human Resources Development Canada funded an experimental project called the Self-Sufficiency Project (SSP). The intent of the project was to observe the effects of an earning supplement for lone parents who qualified for Income Assistance. The project funded participants up to three years if they found and maintained full-time employment. The experimental project was tested in British Columbia and New Brunswick.

Kristen Harknett and Lisa Gennetian examined the SSP data to determine if the supplement affected rates of marriage and cohabitation among lone mothers compared to a control group. The authors noted the comparative wealth of American literature on welfare and conjugal relationship formation and dissolution. They acknowledged earlier research suggesting that other experimental projects such as the negative income tax (minimum income) in the 1970s were abandoned in part because it was thought to have contributed to marital dissolution. 25 25 Kristen Harknett and Lisa A. Gennetian, “How an Earnings Supplement Can Affect Union Formation among Low-Income Single Mothers,” Demography 40, no. 3 (2003): 451, https://doi.org/10.2307/1515155.

Harknett and Gennetian’s work uncovered a fascinating observation. The SSP experiment correlated to very different results between the two provinces. For example, there was an increased marriage and cohabitation rate among the New Brunswick sample compared to the control group. In British Columbia, there was a reduction in the marriage rate and no effect on cohabitation compared to the control group. In fact, by the thirty-six-month mark of the program there was a 16 percent drop in the marriage rate among participants compared to the control. In New Brunswick, however, there was a 22 percent increase in the marriage rate compared to the control group at the 36th month. Analysis at the subgroup level confirmed the contrasting results between the two provinces.

The authors considered various explanations but concluded that local labour markets and policy contexts likely mediated the effect of the SSP on marriage and cohabitation among lone mothers. The initial enthusiasm for the SSP dissipated when the long-term effect of the program on social-assistance use appeared to fade compared to a control group. 26 26 Chris Riddell and W. Craig Riddell, “When Can Experimental Evidence Mislead? A Re-assessment of Canada’s Self Sufficiency Project,” Discussion Paper Series (Bonn, Germany: IZA, May 2016), ftp://repec.iza.org/RePEc/Discussionpaper/dp9939.pdf. Riddell and Riddell suggest the long-term impact of the SSP should be reconsidered as policy changes during the pilot project may have influenced the behaviour of the control group influencing the long-term evaluation.

The portion of couples living in cohabiting relationships has increased ever since Statistics Canada has been keeping track, beginning with the 1981 Census. Research demonstrates differences in economic behaviours and other outcomes of cohabiting couples and their children compared to married couples. Much of the Canadian research has not distinguished between the two family forms when examining the role of social assistance on family stability.

The work of Emily Louise Hanna warrants examination in some detail, as it distinguishes between marriage and cohabitation, and for her use of data from the SLID. Hanna’s interest was due in some part to her speculation that cohabitation was possibly becoming a “functional substitution” for marriage.

Hanna utilized time-series cross-sectional data from the SLID to explore the correlation between social assistance and marital decision-making. The study encompassed the period in which Ontario implemented a reformed social-assistance policy, allowing Hanna to estimate the effects of this reform on marital decision-making.

Hanna summarized the previous literature, including many of the American studies. She concluded that the previous body of work “suggests that social policy and taxation have small and mixed effects on marital status and living arrangement decisions.” 33 33 Emily Louise Hanna, “Marital Status Decisions and Canadian Social Assistance Policy” (PhD diss., University of Toronto, 2007), 21–22.  Hanna argued that the mixed effects shows not that policy lacks influence on these decisions, but that the policies examined have been “uninformative.” In short, studies have not focused on the policy changes that influence marital decision-making. She stated that her work gave evidence that social policy can inform partnership outcomes. 34 34 Hanna, 22.  As an example, she argued that significant changes to the spouse-in-the-house rules in Ontario had particular influence on marital decision-making.

The 1995 the government of Ontario made changes to social assistance, including when cohabiting partners are considered spouses. Previously, couples were not considered spouses unless they had been co-residing continuously for three years. Prior to the three-year mark, non-married conjugal partners could receive social assistance as individuals without consideration for the other partner’s income. The policy was a disincentive toward marriage because of the reduction of benefits for cohabiting couples transitioning into marriage. The 1995 reform considered conjugal partners to be spouses the moment they moved in together. After an Ontario Court of Appeals ruling, the government instituted a 2002 policy defining cohabiting couples as spouses after three months of continues co-residency. The new rules reflected a similar policy implemented in Saskatchewan in 1997. 35 35 Hanna, 38-39.

Hanna examined the effect of the change to the spouse-in-the-house rule.

The 1995 Ontario reforms correlate with a 2.5 percent reduction in the probability of cohabiting for women who dropped out of high school and are over age forty, and a 3.7 percent reduction among their male peers. The reforms were associated with an increased probability of marriage among the same cohort. High school dropout males over age forty had nearly a 6 percentage-point increase in the probability of being married while their female counterparts had a nearly a 3 percentage-point increase. 36 36 Hanna, 65, 68.

Hanna’s results suggest that spouse-in-the-house reforms in Ontario encouraged cohabiting couples to marry but did not motivate single individuals to marry. Hanna suggested that the reforms also discouraged singles from entering cohabiting relationships.

Hanna’s work illustrates how public policy can influence behaviour, and it serves as a caution to policy-makers about the unintended impact of policy changes.

Hanna argued that cohabitation may be viewed by some as an institution, but the family structure’s status is inhibited by the lack of legal protection in the areas of social benefits, property rights, support obligations, and inheritance law. She acknowledged that the economic behaviour of cohabiting couples regarding household production, economies of scale, and pooling of resources differs from married couples. 37 37 Hanna, 8, 162.

Finally, Hanna argued that social-assistance policies that move eligible participants from dependence on the state to dependence on a mate undermined the goal of assistance policies to reduce dependency. Hanna argued that her findings suggested this was not the case with the 1995 Ontario policy change. 38 38 Hanna, 139.  It should not be assumed that policies that encourage marriage are necessarily encouraging dependence, but may in fact encourage economic cooperation. This cooperation, as Hanna acknowledged, distinguishes itself from cohabitation in behaviours such as pooling of resources.

Marriage as a social institution is economically beneficial for its members and larger society. The stability associated with marriage contributes to the common good, yet the institution has been in decline in recent decades, particularly among those who need it most: low-income Canadians. Whether poverty-alleviation policies contribute to the decline of marriage among low-income Canadians is an important question. This paper is the initial step in a planned project to explore this question.

Within the current Canadian policy environment there is little evidence that policy-makers distinguish between cohabitation and marriage, or consider the impact of policy on family formation.

The body of Canadian research on the impact of social assistance on marital decision-making is far less robust compared to the research in the United States. Canadian research has produced mixed results regarding the relationship between social assistance and marital decision-making, with little work undertaken in the last ten years. Existing research suggests regional wage and labour-market conditions influence social-assistance rates. How might these conditions also influence pathways into marriage?

The studies reviewed in this paper show that variations in policy such as the drastic change in Ontario’s spouse-in-the-house rules in the 1990s are important variables when considering the relationship between social assistance and marital decision-making.

Recent work using social-assistance user data (not included in this review) suggests that many people move in and out of the system. Those who are long-term participants tend to transition to disability benefits. The relationship between marital status and long-term disability is a potential area for further exploration as well.

As the studies above demonstrate, partnership status does not remain static. Further exploration could examine whether income level correlates with frequency of partnership change.

Finally, the guaranteed income discussion has returned to public discourse. Further work could investigate the current validity of concerns raised in the 1970s about the impact of that policy on marital dissolution.