MORE OR LESS UNMARRIED. THE IMPACT OF LEGAL SETTINGS OF COHABITATION ON LABOUR MARKET OUTCOMES

. We study how diﬀerent levels of protection upon separation aﬀect the labour market behaviour of unmarried cohabiting partners. In Canada, unmarried cohabitation becomes a legal status after one year of relationship. Most provinces automatically expand couples’ rights and responsibilities after several years of cohabitation: some provinces allow cohabiting partners to claim for alimony upon separation, while others consider cohabiting couples to be equal to married couples. Using cross-province variations in legal settings and minimum eligibility duration, we ﬁnd that eligibility for a regime making cohabiting partners equal to married partners increases men’s labour supply and decreases women’s labour supply and earnings while eligibility for a regime allowing for post-separation transfers between ex-partners decreases women’s earnings only. Our results show that eligibility aﬀects within-household allocation of earnings and hours of work, and reinforces existing inequality. Our results contribute to the ongoing public debate regarding the legal recognition and level of protection that should be given to unmarried cohabiting partners.


C. Common trend tests
Our estimation strategy relies on the common trends assumption.To test if the common trend assumption is likely to hold true, we test whether the treated group behaves differently from the control group before treatment in an event study approach.
This method is not straightforward to implement in our setting, for two reasons.First, we estimate simultaneously the impact of two different reforms.Second, there are several ways to become eligible for a protective regime: some couples become eligible because of the introduction of a new regime (in the paper, we call this group "couples eligible at the moment of the reform") and some couples become eligible when they reach the required number of years (in the paper, we call this group "couples eligible after the reform").
Notice that these groups are exclusive within regimes (the same couple is either eligible at the moment of the reform or after the reform) but not between regimes.For instance, in Manitoba and Saskatchewan, couples eligible after the introduction of the marriage-like regime are also formed after the introduction of the alimony regime but some couples formed after the introduction of the alimony regime are directly eligible at the introduction of the marriage-like regime.
For the event-study analysis, we calculated two variables: one variable measures time to the reform, the other measures time to eligibility for the regime.For both, we cannot observe many periods before the reform or before eligibility.For couples eligible at the moment of the reform, time to eligibility corresponds to the time to the reform.The maximum number of periods observed depends on the year of the reform but also on the beginning and end years of the rotating panels.As we observe five 6-years panels and one 4-years panel (1993-1998; 1996-2001; 1999-2004; 2002-2007; 2005-2010; 2007-2011) We perform several series of tests (on men and women separately, and on each outcome separately), that we present in this online appendix: (1) Effect of the marriage-like reform: we observe up to four periods before marriagelike reform (and three after) (fig.C1) (2) Effect of eligibility for alimony regime on couples eligible after the reform: we observe up to three periods before eligibility for the alimony regime (and four after) (fig.

C2)
(3) Effect of eligibility for marriage-like regime on couples eligible at the moment of the reform: we observe up to four periods before eligibility for the marriage-like regime (and three after) (fig.C3) (4) Effect of eligibility for marriage-like regime on couples eligible after the reform: we observe up to three periods before eligibility for the marriage-like regime (and four after) (fig.C4) We first present the results of our test for the common trends assumption in our analysis of the effect of the marriage like reform.Figure C1 shows the results.We find no difference in the labour market outcomes of men and women before the introduction of the marriagelike regime, in treated provinces compared to other provinces.We now present the results of our tests of common trend assumptions in our analysis of the effect of the eligibility status, separating couples formed before the introduction of the reform from couples formed after the introduction of the reform.We find no differences in the labour market outcomes of individuals before eligibility between treated and untreated individuals, except for the labour income of men-before becoming eligible for the alimony regime, men in a relationship formed after the reform tend to have higher labour earnings than men who are in the same duration of cohabitation but who are not about to become eligible for the alimony regime.However, we do not find any significant effect of eligibility on men's labour income after the period.
-  Note: Data are from the 1993-2011 SLID panel data.The sample consists in individuals living in cohabitation for less than 10 years, aged between 18 and 50 years old in Canada.All regressions include controls for individual fixed effects, relationship duration fixed effects, year fixed effects, a dummy indicating having a child, year dummy interacted with a dummy for having a child, relationship duration fixed effects with an interaction for having a child, province fixed effects interacted with a dummy for having a child, age and age square.We use SLID longitudinal weights.Active weeks gives the number of weeks where the individual is either working or unemployed; Not employed is a binary variable indicating whether the individual had been either inactive or unemployed all year; Labour earnings gives fiscal labour earnings in constant Canadian dollars of 2002.Note: Data are from the 1993-2011 SLID panel data.The sample consists in women living in cohabitation for less than 10 years, aged between 18 and 50 years old in Canada.All regressions include controls for individual fixed effects, relationship duration fixed effects, year fixed effects, a dummy indicating having a child, year dummy interacted with a dummy for having a child, relationship duration fixed effects with an interaction for having a child, province fixed effects interacted with a dummy for having a child, age and age square.We use SLID longitudinal weights.Active weeks gives the number of weeks where the individual is either working or unemployed; Not employed is a binary variable indicating whether the individual had been either inactive or unemployed all year; Labour earnings gives fiscal labour earnings in constant Canadian dollars of 2002.
We now focus on the specific case of the dynamic of labour earnings around eligibility for the alimony regime for men and women in couples formed after the reform.Notice that the group of couples formed after the introduction of the alimony regime is the largest group of treated individuals, as most provinces passed a reform introducing the alimony regime before our period of observation.We separate provinces where couples are eligible after

A.
Figure A1.Definition of Spouse in the Family Property Act.Chapter F-6-3.(Saskatchewan, 1997) , individuals are observed for a maximum of 6 years.It means couples affected by a reform occurring in 1997 are observed 4 years before the reform maximum in panel 1993-1998 and one year maximum in panel 1996-2001.For couples eligible after the reform, time to eligibility corresponds to the duration of the couple minus the minimal duration to be eligible.Couples are eligible after 2 or 3 years of cohabitation, which means that they are observed a maximum of 2 or 3 periods before eligibility.Despite the small number of periods observed before eligibility, we think this method remains the most convincing way to test the common trends assumption.

Figure C1 .
Figure C1.Effect of the marriage-like reformNote: Data are from the 1993-2011 SLID panel data.The sample consists in individuals living in cohabitation for less than 10 years, aged between 18 and 50 years old in Canada.All regressions include controls for individual fixed effects, relationship duration fixed effects, year fixed effects, a dummy indicating having a child, year dummy interacted with a dummy for having a child, relationship duration fixed effects with an interaction for having a child, province fixed effects interacted with a dummy for having a child, age and age square.We use SLID longitudinal weights.Active weeks gives the number of weeks where the individual is either working or unemployed; Not employed is a binary variable indicating whether the individual had been either inactive or unemployed all year; Labour earnings gives fiscal labour earnings in constant Canadian dollars of 2002.

Figure C3 .
Figure C3.Effect of eligibility for the marriage-like regime on couples formed before the reform

Figure C4 .
Figure C4.Effect of eligibility for the marriage-like regime on couples formed after the reform

Figure C5 .Figure C6 .
Figure C5.Effect of eligibility for the alimony regime on labour earnings of men in couples formed after the reform Note: Data are from the 1993-2011 SLID panel data.The sample consists in men living in cohabitation for less than 10 years, aged between 18 and 50 years old in Canada.All regressions include controls for individual fixed effects, relationship duration fixed effects, year fixed effects, a dummy indicating having a child, year dummy interacted with a dummy for having a child, relationship duration fixed effects with an interaction for having a child, province fixed effects interacted with a dummy for having a child, age and age square.We use SLID longitudinal weights.Labour earnings gives fiscal labour earnings in constant Canadian dollars of 2002.

Table B2 .
Impact of the eligibility for a protective regime of cohabitation on labour supply and labour earnings.Sample restricted to individuals with child at first observation Note: Data are from the 1993-2011 SLID panel data.The sample consists in couples living in cohabitation for less than 10 years, aged between 18 and 50 years old in Canada that have a child the first time they are observed in the data.All regressions include controls for individual fixed effects, relationship duration fixed effects, year fixed effects, a dummy indicating having a child, year dummy interacted with a dummy for having a child, relationship duration fixed effects with an interaction for having a child, province fixed effects interacted with a dummy for having a child, age and age square.We use SLID longitudinal weights.Standard errors are clustered at the province level and are reported in parenthesis.Cluster p-value are reported in brackets.Wild cluster bootstrap p-values are reported in braces.Number of active weeks gives the number of weeks where the individual is either working or unemployed; Number of hours worked gives the number of hours worked during the year; Not employed is a binary variable indicating whether the individual had been either inactive or unemployed all year; Labour earnings gives fiscal labour earnings in constant Canadian dollars of 2002.

Table B3 .
Type of relationship at couple formation: probability of being in cohabitation vs. married Data are from the 1993-2011 SLID panel data.The sample consists in couples in 0 or 1 year after couple formation, aged between 18 and 50 years old in Canada.All regressions include controls for sex of the respondents, dummy for having a child, sex of the respondent interacted with dummy for having a child, year fixed effect, province fixed effect, educational attainment, age of respondent.We use SLID longitudinal weights.Standard errors are clustered at the province level and are reported in parenthesis.Cluster p-value are reported in brackets.Wild cluster bootstrap p-values are reported in braces.