# Childcare, Parental Behavior and Well-Being during Childhood

Catherine Haeck, Laetitia Lebihan, Pierre Lefebvre and Philip Merrigan

# Abstract

In this paper, we study the long-term impacts on parental health and behaviors of a low-fee universal childcare program for children aged zero to five years old. We follow families for more than 10 years after the reform. For families with preschool children, we show that the adverse effects documented in previous studies on maternal depression and parental behaviors persist over time. However, once children enter school, most negative effects of the program on parental mental health and behaviors fade away. Only the negative effect on positive interactions may have persisted over time.

JEL Classification: I31; J13; J18; J20

## 1 Introduction

In the last few decades, there has been a sharp increase in the labor market employment rate of mothers in developed countries. Although this has had a positive effect on family incomes, it has also made parenting a more demanding and stressful task given the increased time and pressure from work. Concurrently, parents growing demand for childcare has attracted the attention of policy makers toward public or subsidized childcare programs. The idea that childcare subsidies should no longer be limited to low-income families, but should instead be universal, as is the case in most European countries, is slowly emerging as a model for North American governments, particularly in Canada.

Studies estimating the effects of universal childcare policies have focused mainly on their impact on maternal employment and child development (see Baker 2011, for a review). However, as explained by Herbst and Tekin (2014), a full evaluation of childcare subsidies requires a thorough understanding of the ways in which subsidies influence both parents and their children. When mothers reallocate their time from home to the labor market, a change in the nature of time spent with the child occurs, affecting not only children’s well-being but also parents’ well-being and behaviors with their children. An extensive literature shows how maternal well-being affects by itself a child’s well-being and development (Currie and Almond 2011; NICHD 1999, 2003).

The first major study to examine the effects of a universal childcare policy on both children and parental well-being was conducted by Baker, Gruber, and Milligan (2008). Their study exploits a major childcare reform which took place in Canada in the late 1990s. In 1997, the government of the province of Quebec initiated the gradual implementation of a low-fee childcare policy. Childcare spaces were provided at a single low-fee of CA$5 per child per day, approximately US$3.50 at that time (CA$7 as of 2004). The reform was phased in to include all of Quebec’s children less than 6 years of age (not in publicly provided kindergarten) as of September 2000. This policy had the effect of moving a large proportion of children from informal care and maternal care to regulated childcare (Lefebvre and Merrigan 2008). For example, the number of regulated childcare spaces in Quebec increased from 78,864 in September of 1997 to 258,366 as of March, 2013 (Ministre de la Famille et des Ainés 2013). More importantly, as a result of the policy, the labor force participation of mothers increased by 7.7 percentage points in Quebec by 2003 (Baker, Gruber, and Milligan 2008). These authors also show that the reform had a negative effect on several parental and child outcomes during the preschool years. In particular, the policy had a negative effect on parents’ health and behaviors.[1] However, these results were obtained strictly with children aged 0–4 years old and their parents. At the time of these studies only a few post-policy years of data were available. The results thus reflect the short-term impact of the reform up to 2003. It is difficult to identify the mechanisms that explain their results, but several can be considered, as discussed in the next paragraphs. In this study, we extend the research of Baker, Gruber, and Milligan (2008) on universal childcare and parental outcomes in two ways. First, we estimate the effects during the preschool years over a longer observation period using six years of additional data, up to 2009. Second, we examine whether the effects on parents found in the preschool years persist when the child enters school. Documenting the long-term effects of universal childcare on parents is crucial to our understanding of the overall impact of such reforms, both when the child is in childcare but also once the parents are no longer directly affected by the program. Studies of long-term impacts of early childhood education and care (ECEC) on parents are very scarce and mainly focus on low-income or single-mother households (Barnett 2008; Herbst and Tekin 2014; Martin et al. 2008).[2] Regarding Quebec’s childcare policy, the long-term effects documented focused on labor supply (Lefebvre, Merrigan, and Verstraete 2009) and child outcomes (Baker, Gruber, and Milligan 2019; Haeck, Lebihan, and Merrigan 2018),[3] but not on parents and family interactions. Relative to the first objective of this study, our empirical strategy allows us to measure whether the effects on the parents of preschool children found in previous research are transitional or persist when the program matures. We use data on parents of children who were eligible for low-fee childcare for up to 5 years. In previous studies, accessible data was mainly focused on the early years of the program, when children had been eligible for low-fee childcare for only 1–2 years. At this time, the number of spaces was highly constrained (Haeck, Lefebvre, and Merrigan 2015). As a result, the estimated short-term effects in Baker, Gruber, and Milligan (2008) may be different from the longer term effects once the number of spaces has stabilized and children attend childcare for 4–5 years. During the transition, parents also had to adjust to a new social norm. This may also have taken some time and may have influenced the early measured impacts of the childcare reform. Haeck, Lefebvre, and Merrigan (2015) show that the positive effects on maternal labor force participation start in 2000 at 5 percentage points, increase to 13 percentage points by 2004 and persist up to 2009, the final year[4] of the data set. This further suggests that long term effects on parental well-being might differ. Relative to the second objective, there are several reasons to assume that the childcare policy could affect parents with children even when they are no longer in childcare. First, Lefebvre, Merrigan, and Verstraete (2009) show that the policy had long-term labor force participation effects (when their child is in school) on mothers who benefited from the program when their child was eligible at an early age. They show that the policy had a positive labor force participation impact on mothers who, without the policy, would have stayed home even when the child entered school. Negative effects on parents well-being could therefore persist because of increased difficulties balancing work and raising children even once the child enters school. Second, previous research documented negative effects on children’s developmental scores, health, and behavioral outcomes (e.g. Baker, Gruber, and Milligan 2008, Kottelenberg and Lehrer 2014). If these effects persist once they enter school, then parents’ health and behaviors could also be negatively impacted since the well-being of the child and the well-being of parents are linked. However, recent work by Haeck, Lebihan, and Merrigan (2018) rule this option out. They show that the childcare reform had limited negative effects on children’s health and behaviors once children are in primary school or even high school. More precisely, they show that for children in school, only the impact on children’s emotional disorder and anxiety persists but the magnitude of the effect is considerably smaller than the effect measured for preschoolers. Since children are doing relatively better once in school, parents may also experience a similar improvement. Third, the negative effects on parents may simply persist through some kind of auto-regressive process. Finally, a fourth possible channel could be that parent-child interactions become less stressful when the child enters school. Parents may be less stressed and anxious when their child becomes older because they are more mature and independent, and require less effort and time investment from parents. For our study, we use the National Longitudinal Survey of Children and Youth (NLSCY), cycles 1 (1994–95) to 8 (2008–09). This survey is a representative sample of the Canadian population of children and their families. To estimate the policy effects, we rely on a non-experimental evaluation framework with multiple pre- and post-treatment periods. Effectively, we compare Quebec parents before and after the reform to comparable parents in the rest of Canada (RofC). Our empirical strategy allows us to differentiate the intensity of treatment for each cohort, given the gradual implementation of the policy. Our estimates suggest that, overall, the reform had negative effects on the mental health and behaviors of parents with preschool children. However, for some outcomes, our estimates by wave suggest that the effects decrease over time and eventually become statistically insignificant by 2008. We then extend the analysis to parents with at least one child aged 5–9 years old and no children aged less than 5. We show that, once children enter school, the negative effects of the program on parental mental health and behaviors fade away, except for the negative effect on mother-child interactions. Finally, Lefebvre, Merrigan, and Verstraete (2009) show that the reform had long lasting positive labor supply effects (once the child enters school), especially for less educated mothers, yet the effects we document do not differ by maternal education. Increased labor force participation for mothers. may explain the negative effect on positive interactions, a measure of positive quality time spent with the child. The outline of the paper is as follows. Section 2 describes Quebec’s childcare reform. Section 3 describes the data set, while Section 4 presents the methodology. Empirical results are presented in Section 5, and Section 6 concludes the paper. ## 2 Quebec’s Childcare Reform In the late 1990s, the government of Quebec initiated the gradual implementation of a low-fee childcare network for children under 6 years of age. The low-fee childcare spaces could be purchased at a single price of$5 per day per child. On September 1, 1997, only the 4-year-olds were eligible for low-fee spaces. They were followed by the 3-year-olds on September 1, 1998 and the 2-year-olds on September 1, 1999. By September 1, 2000, all children aged less than 59 months, not entitled to kindergarten because their fifth birthday occurs after September 30, became eligible for subsidized childcare. While all children were eligible, the number of available spaces at the time did not meet demand. Between 2000 and 2012, the number of low-fee spaces increased from 85,000 to 217,000 spaces, thereby releasing the capacity constraint. The price of low-fee childcare increased from $5 to$7 per day per child in 2004. Overall the total number of regulated childcare places in Quebec rose from 78,864 in 1997 to 258,366 in 2013, and the total government subsidy reached 2.3 billion dollars for fiscal year 2012–2013 (Conseil du Trésor-Quebec, Budget 2012–2013). In contrast, the number of subsidized childcare spaces in the other Canadian provinces was relatively small compared with the province of Quebec and changed little between 1997 and 2009 (Haeck, Lefebvre, and Merrigan 2015). This reform drastically changed the way in which preschool children were cared for in Quebec, while no comparable changes were observed in the rest of Canada.[5]

Figure 1 presents the mean hours (conditional and not conditional on the use of childcare) per week that children aged 1 to 4 spent in their primary care arrangement, but also the labor force participation of mothers (two-parent and single-parent families) and fathers (in two-parent families) for these children in Quebec and the RofC. Haeck, Lefebvre, and Merrigan (2015) show that not only did a much larger percentage of infants and toddlers start to attend daycare in Quebec following the reform, but that the intensity of care also increased. Relative to the labor force participation, the main changes are for mothers. Indeed, for two-parent families in Quebec, mothers’ labor supply increased in most waves, starting at 55 percentage points in 1994 and reaching 76 percentage points in 2008. In contrast, no significant changes in the hours of care and maternal labor force participation has occurred in the RofC. For single mothers, there are large increases in labor force participation for both Quebec and the RofC, but the original gap, in favor of mothers in the RofC is totally closed by 2008.

### Figure 1:

Trends in childcare hours and parental employment in Quebec and the rest of Canada, 1994–2008.

Shows the evolution of the mean number of hours per week spent in the primary mode of care in the Rest of Canada (left panel) and Québec (right panel) non conditionally (hollow square) and conditionally on attending childcare (hollow circle). The sample includes NLSCY cross-sectional children aged 1 to 4. Source: Haeck, Lefebvre and Merrigan (2015).

Along with subsidised childcare, the policy implemented changes for school-age children. First, full-day kindergarten replaced half-day kindergarten for 5-year-olds in school as of September 1998.[6] Haeck, Lefebvre, and Merrigan (2015) show that the kindergarten policy itself did not have any impact on mothers’ labor force participation, but the combination of the low-fee daycare program and full-day kindergarten did. Second, before- and after-school daycare were now also offered to children aged 5 to 12 on school premises, also at the low-fee of $5 per day per child and$7 as of 2004. Lefebvre, Merrigan, and Verstraete (2009) and Haeck, Lebihan, and Merrigan (2018) show, respectively, that the before- and after-school program had no effect on labor supply and child outcomes. Indeed, these studies show that the price change was much smaller and could only marginally affect mothers’ labor supply decisions and child well-being. Clearly, this past research demonstrates that the main impacts on children and mothers are caused by the childcare policy.

## 3 Data

To estimate the long-term impacts of the reform on parents, we use the National Longitudinal Survey Children and Youth (NLSCY). The NLSCY contains both child and parental outcomes as well as extensive variables related to parental labor supply, childcare use, and other demographic characteristics. The survey started in 1994–1995 (wave 1) and ended in 2008–2009 (wave 8). This implies that we can observe families with young children in Canada four years prior to the implementation of the reform and for more than 10 years after. The first longitudinal cohort contained approximately 22,000 children aged 0 to 11 in 1994–1995. These children were originally scheduled to be surveyed every two years. In 1996, however, approximately 6000 children were dropped from the survey. The remaining children were followed until 2007–2008, the last cycle of the survey. As of cycle 2, in 1996–97, in addition to the original cohort, a new cohort of 0 to 1-year-olds was selected in every cycle. These cohorts were surveyed until the age of 5 and in some cases until the age of 7 or even 9. Therefore, we can observe families with children aged 0–9 years old throughout the length of the survey. In the NLSCY, all outcomes are reported by the person most knowledgeable about the child (almost always the mother).

We start by replicating earlier results on families with preschoolers, but over a longer observation period, as the program matures. In particular, we test whether the adverse effects of the program persist after 2006, when efforts around the qualification of educators and improvement in their salaries were undertaken by the government and the number of spaces in the network stabilized (Haeck, Lefebvre, and Merrigan 2015; Haeck, Lebihan, and Merrigan 2018). We then determine whether the effects on families with preschoolers persist when the policy is no longer contemporaneously effective, that is, when children are in school (ages 5–9). For children in school, we only retain in our sample families with no children of preschool age in order to avoid any confounding effects of the program.

Given the gradual implementation of the reform, children and parents were treated differently by the policy over the years. Table 1 shows the various treatment groups by presenting the eligibility of children according to their age and NLSCY wave from which they are sampled. The grey-shaded area highlights the post-reform years while the unshaded area refers to the pre-reform years. Numbers indicate the number of years of eligibility for subsidized childcare. To calculate the number of eligible years, we always use December 31 of the first year of the two-year period as a reference.[7] For example, for wave 4 (2000–01), the reference point is the child’s age on December 31, 2000. For the index 0.5, the child is eligible for a few months, not a year. Table 1 also shows that some age groups are not observed in certain waves. For example, parents with children aged 8 and 9 years old are not observed in 2004 (wave 6). So, given the structure of NLSCY, school-age children are split in two groups: the 5- to 7-year-olds and the 8- to 9-year-olds.[8] We exclude wave 3 data for children 0–7 years old in order to avoid the overlap of treated and untreated children in the same wave.[9] Moreover, the number of regulated childcare spaces did not change in the early years of the reform (before 1999). At that time, existing spaces were converted to low-fee spaces (see Figure A.1 in Haeck, Lefebvre, and Merrigan 2015). Nevertheless, we assess the robustness of our results to this restriction in the empirical section.

### Table 1:

Eligibility for low-fee childcare by age of the child and NLSCY wave.

Wave
Age Wave 1 Wave 2 Wave 3 Wave 4 Wave 5 Wave 6 Wave 7 Wave 8
(1994–95) (1996–97) (1998–99) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09)
Wave 1-5: Baker, Gruber and Milligan (2008) Additional Data
0-4 years: Baker, 0 × × × 0.5 0.5 0.5 0.5 0.5
Gruber and 1 × × × 0.5 1 1 1 1
Milligan (2008) 2 × × × 0.5 2 2 2 2
3 × × 0.5 1 2 3 3 3
4 × × 0.5 1 2 4 4 4
Additional Data 5 × × 1 2 3 4 5 5
6 × × × 1 2 (n.a) 3 (n.a) 5 5
7 × × × 1 2 (n.a) 3 (n.a) 4 5
8 × × × × 1 2 (n.a) 3 5 (n.a)
9 × × × × 1 2 (n.a) 3 4 (n.a)
1. This table shows the eligible children in Quebec to the low-fee daycare reform (grey shaded area) and non-eligible children in Quebec (indicated by a symbol ×) according to child’s age and wave. Numbers indicate the number of years of eligibility. For example, a 5-year child in wave 5 (and therefore born in 1997) was eligible for three years of low-fee child care. The index 0.5 refers to the fact that the child is eligible for a few months, not a year. We exclude wave 3 for children 0-7 years old in order to avoid overlapping of treated and untreated in the same cycle.The term n.a (not available) means that the child is eligible for low-fee child care spaces but data for this age group in this wave are not available in the NLSCY. Baker, Gruber and Milligan (2008) captured the short-term effects of the reform up to 2003. We extend the observation period to 2009.

Clearly, the number of years in low-fee childcare increased over time (Table 1). As such, families with children aged 0–4 years in Baker, Gruber, and Milligan (2008) were only treated for a few months to 2 years (waves 4 and 5 of the NLSCY).[10] Here, we add an additional six years of data and we also observe school-age children. This allows us to study the impact of the reform on parents with preschool children over a longer period. It also enables us to analyze the long-term effects of the policy on families with school-age children (aged 5–9 years old). In both groups, we can now observe families with children who were eligible for low-fee childcare since birth and were therefore highly exposed to the reform (up to 5 years of treatment).

Table A.1 summarizes the various treatments by age group while presenting the eligibility of families according to child age and NLSCY wave. We distinguish four cohorts defined by their eligibility status to low-fee childcare: (1) not eligible; (2) partially eligible, with the estimated effects for this group symbolized by β P ; (3) fully eligible but observed in the early years of the program, β FE ; and (4) fully eligible and observed in the later years of the program, β FL . Families never eligible for the childcare reform are in the first cohort. In the NLSCY, this cohort includes families with children aged 0–7 years old in waves 1 and 2 and families with children aged 8–9 years old in waves 1 to 4. The second cohort includes families partially treated by the reform; these families were all eligible for low-fee childcare but only for a maximum of 3 years. As mentioned earlier, waves including families with children who were not treated and children who were, are excluded from our main sample. The third cohort includes families with children who were treated since birth, but were still in the early years of the program. Finally, the fourth cohort includes families with children who were eligible since birth and were observed in the later years of the program. Table A.1 reveals that, for the 8- to 9-year-olds, we only observe families with children who were partially treated. For the 5- to 7-year-olds, we observe families with children who were fully treated but only in the early years of the program. Finally, as to fully treated children in the later waves of the program, we only observe families with children aged 0–5 years old. As suggested by Haeck, Lebihan, and Merrigan (2018), we consider the fact that families are exposed very differently to the program. Not separating families with different treatment intensity results in estimates that potentially mask important differences. In this study, we compare families with similar exposures to the program (partial vs. full exposure). As will become clear in the next sections, the timing and duration of eligibility should always be kept in mind when interpreting the results on the effects of the reform.

Thus, given the data availability and eligibility for subsidized childcare depending on the age of the child, we focus our analysis on the parents with children aged 1–9 years old. The evaluation is performed for three separate age groups: parents with 1- to 5-year-olds not in school; parents with 5- to 7-year-olds in school (and no children of preschool age); and parents with 8- to 9-year-olds (and no children of preschool age). We exclude from our samples children 12 months old or less who generally do not attend childcare since parental leave is 50 weeks long in Canada.[11] We also include parents with 5-year-olds not in school in the sample of preschool children as they are likely to be in subsidized childcare before being eligible for kindergarten.[12]

Following Baker, Gruber, and Milligan (2008), Kottelenberg and Lehrer (2013, 2014, and Haeck, Lebihan, and Merrigan (2018), we focus on two-parent families to avoid interference with other policies targeting low-income families (largely represented by single-parent families). Various provincial and federal reforms have been implemented since 1997 and could interact with the low-fee childcare reform. Baker, Gruber, and Milligan (2008) and Milligan and Stabile (2007) show that changes in family/child benefits have a statistically significant and relatively large impact on different outcomes for single-parent families, but little impact on two-parent families. In addition, the Government of Quebec introduced a new work incentive policy in 2005 to support and develop the work effort of low-wage workers as well as encourage people to exit welfare into work (Ministry of Finance of Quebec 2004). Therefore, since any specific policy shock in Quebec coinciding with the universal childcare reform may bias our results, we focus on two-parent families.[13]

We measure the impact of the reform on parental health and parental behavior. For parents’ health, we use the following outcomes: (1) the mother’s health status (1: excellent, 0: not excellent); (2) the father’s health status (1: excellent, 0: not excellent); and (3) the mother’s depression score (score ranging from 0 to 36). A high score indicates the presence of symptoms of depression. All questions on parents’ health are asked to households with children aged 0 to 9. For parental behavior and parenting per se, several measures are available: (1) the family dysfunction index (score ranging from 0 to 36); (2) the positive interaction score (ranging from 0 to 20); (3) the hostile/ineffective parenting score (ranging from 0 to 25); (4) the consistent parenting score (ranging from 0 to 20); and (5) the aversive parenting score (ranging from 0 to 20). A high score for (2) and (4) indicates positive parental behavior for child well-being while the opposite is true for (1), (3), and (5). The questions on parents’ behavior are asked when children are 2–9 years old, except for the family dysfunction score, which is available for all parents. The subquestions used for each measure are reported in Table A.2.[14] We construct our outcomes in the same way as Baker, Gruber, and Milligan (2008) and Kottelenberg and Lehrer (2013, 2014 in order to compare our results with these studies.

We also use the same control variables as Baker, Gruber, and Milligan (2008), Kottelenberg and Leher (2013), and Haeck, Lebihan, and Merrigan (2018) in our regression analysis to ensure that any differences between our results and theirs is not due to controls or methods. The control variables are: the sex of the child, the age group of the mother and father at the child’s birth (14–24 years old (omitted), 25–29, 30–34, 35 or more), the mother’s and father’s highest level of education (less than a high school diploma, high school diploma, some post-secondary education, with post-secondary diploma (omitted)), a dummy for whether or not the mother or father was born in Canada, the size of the area of residence (five groups from rural population to 500,000 residents or more (omitted)), the presence of older children (no older child (omitted), one older child, at least two older children), the presence of younger children (no younger child (omitted), one younger child, at least two younger children), the presence of children of the same age, and dummies for the age of the child. Summary statistics for parents with children aged 1–9 years in Quebec and the RofC pre- and post-reform are presented in Table A.3.

## 4 Empirical Strategy

To estimate the long-term effects of the daycare reform, we rely on a non-experimental evaluation framework based on multiple pre- and post-treatment periods (difference-in-differences model). The treatment group includes Quebec parents with children of a given age before and after the reform. The control group includes parents in other Canadian provinces with children of the same age observed during the same time period. Only Quebec parents are affected by the reform and periods of pre- and post-treatment depend on the age of the child (see Table 1). To account for the gradual implementation of the reform, we allow the effects of treatment to differ in each of the post-reform waves. The empirical model is as follows:

(1) Y i p t = α + t = T 8 β t W t Q i t + Φ X i t + ω t + θ p + ε i p t

where Y ipt represents a parent outcome for child i in wave t in province p. Regional and temporal differences are captured using province fixed-effects (θ p ) and survey wave[15] fixed-effects (ω t ). To capture the gradual impact of the reform due to the phase-in by age group, we estimate the coefficients on the interaction between wave dummies (W t ) and a Quebec province dummy (Q it ). The term Q it is a dummy variable taking the value 1 if child i lives in Quebec in wave t and 0 otherwise. The term W t takes the value of 1 if the wave is greater than or equal to T = 4 for families with children 1–7 years old and T = 5 for families with children 8–9 years old (see Table 1). The term X it is a vector of socioeconomic control variables and ɛ ipt is an error term.

Standard errors are estimated using the 1000 bootstrap weights provided by Statistics Canada. This procedure accounts for the complex survey design of the NLSCY. Since we estimate impacts for multiple outcomes simultaneously, we also adjust our p-values following Simes (1986). This correction assumes that our outcomes are correlated with one another and avoids the possibility of over-rejecting the null hypothesis when studying multiple correlated outcomes. As suggested by Shaffer (1995), our adjusted p-values are computed by sub-groups (health; behavior).[16]

Our empirical strategy relies on two critical assumptions. First, our approach assumes no selection based on province-specific transitory shocks. Second, in the absence of the policy, mean outcomes of Quebec families with children would have followed a similar trend as those of comparable families in the other provinces.

Concerning the first assumption, when the reform was announced, ineligible children were already born. Parents could not have delayed conception to be eligible to subsidized daycare. Also, although parents outside of Quebec could have moved to Quebec to benefit from the childcare reform, migration data do not support this idea (see Lefebvre, Merrigan, and Verstraete (2009) for more details). Another reason that could influence group composition relates to the ferility decision of parents. However, relative fertility rates did not change immediately after the reform. Figure A.1 shows that the fertility rate in Quebec since 1990 has been very similar to the Canadian fertility rate. We observe a slight change as of 2007, at the very end of our observation period, not enough to impact our results on children age 1 and over.

Concerning the second assumption, we cannot observe untreated families in Quebec post-reform, but we can observe trends in the outcome variables in the treatment and control groups prior to the reform. Figure 2 shows the evolution of a few outcome variables pre- and post-treatment. The paths of these outcomes in Quebec and the RofC are similar, especially when we have multiple years of observation pre-reform (the 8–9 years old), prior to the introduction of the policy, regardless of the age of the child. This only provides visual information for two of our eight outcomes. Following the presentation of our main results below, we formally test the common trend assumption for all outcomes and interpret our results in light of these findings.

### Figure 2:

Mean values of measures for parental outcomes by region and child age: waves 1–8 of the NLSCY.

Shows the trajectories for the mean of two outcomes (standardized) by age of the child for Quebec and the Rest of Canada (RofC). For families with children in school, figures are obtained for families with the youngest child aged 5-7 years (in school) or 8-9 years.

Data: NLSCY.

We also formally test the common trend assumption for all of our outcomes and age groups using an event-study analysis (Tables A.4 and A.5). The reference category is always wave 1. Results suggest that for preschoolers, the common trend assumption holds for all outcomes: the null hypothesis cannot be rejected for β 2 and β 3. For children aged 5 to 7 the common trend assumption generally holds as well, except for positive interaction and hostile parenting in wave 3 (β 3). It also does not hold for positive interaction for children aged 8 to 9 in wave 4 (β 4). This suggests that our control groups are generally reliable, except when studying the impact of the reform on positive interaction.

## 5 Econometric Results

We start by providing evidence on the effects of the program on the well-being of families with preschool children. These include the families with 1- to 4-year-olds and also the 5-year-olds not yet in school. Since the policy was implemented progressively, relative to former studies, we take into account the differences in intensity of treatment. Some families were treated partially while others fully. We also assess whether the initial impact on families with preschool children persisted as the program matured, especially as of 2006.

We then present the results once these children enter school. We estimate the effects of the program on the families with children aged 5–7 years old and 8–9 years old (without children of preschool ages).

This allows us to estimate the contemporary effect of the reform on parents with preschool children, and the spillover effects into the school years while taking into account the intensity of treatment. We present the average effect over the entire post-reform period in the tables, along with the estimates per wave. We also report whether families with children were treated partially (β P), full early (β FE), or full late (β FL), based on Tables 1 and A.1. We report the coefficients, standard errors, and results from several statistical tests. Estimated coefficients that are statistically significant based on adjusted p-values are presented in bold.[17] We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial to families. To ease the interpretation of our results, all non-binary outcomes were standardized for all respondents to have a mean of 0 and a standard deviation (SD) of 1. The coefficients can thus be interpreted in terms of changes in SDs. We focus on the estimates in boldface.[18]

Table 2 presents the estimated effects of the subsidized childcare policy on parents’ health and behavior when their child is 1–5 years old. The results for parents with children in school, aged 5 to 7 and 8–9 years old, are presented in Table 3.

### Table 2:

Estimated effects of the policy on the health and behavior of parents with children aged 1 to 5 not in school.

Variable β 4−8 β 4 P β 5 F E β 6 F E β 7 F L β 8 F L β 78 N
(2000–09) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09) vs β 4−8
Parent health
Mother in 0.005 0.021 −0.006 −0.011 0.005 0.015 0.008 40,865
Excellent health (+) (0.024) (0.030) (0.031) (0.033) (0.034) (0.035) (0.024)
Father in 0.005 −0.005 −0.020 −0.018 0.038 0.028 0.046** 40,639
Excellent health (+) (0.024) (0.029) (0.032) (0.032) (0.032) (0.034) (0.024)
Mother’s depression 0.151*** 0.132* 0.125** 0.187** 0.270*** 0.051 0.008 39,889
Score (−) (0.049) (0.068) (0.061) (0.075) (0.079) (0.063) (0.053)
Parent behavior
Family dysfunction 0.034 0.084 −0.080 0.024 −0.001 0.124* 0.051 40,336
Index (−) (0.050) (0.059) (0.064) (0.067) (0.070) (0.072) (0.049)
Positive interaction −0.217*** −0.278*** −0.219*** −0.325*** −0.151** −0.121* 0.166*** 30,124
(from 2 years) (+) (0.048) (0.061) (0.060) (0.067) (0.064) (0.064) (0.051)
Hostile parenting 0.201*** 0.187*** 0.202*** 0.251*** 0.261*** 0.118 −0.027 29,654
(from 2 years) (−) (0.064) (0.072) (0.077) (0.090) (0.085) (0.084) (0.055)
Consistent parenting −0.076 −0.177*** −0.149** −0.053 −0.015 0.022 0.133*** 29,272
(from 2 years) (+) (0.054) (0.066) (0.071) (0.078) (0.074) (0.071) (0.050)
Aversive parenting 0.170*** 0.092 0.130* 0.250*** 0.208*** 0.179** 0.038 29,982
(from 2 years) (−) (0.055) (0.067) (0.071) (0.076) (0.077) (0.081) (0.051)
1. This table displays the estimated policy effects and standard errors (in parentheses) for parents of children aged 1 to 5 not yet in school. This table shows the average effect for the full post-treatment period (β 4−8, column 1), and the effects by wave ( β 4 P to β 8 F L , columns 2 to 6). We also test for fade-out effect in waves 7 and 8 using a model including β 4−8 and β 7−8 instead of the wave specific effects (only β 7−8 is shown here in column 7). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table 3:

Estimated effects of the policy on the health and behavior of parents with children aged 5 to 9 in school (youngest child).

Variable Panel A: Children aged 5 to 7 Panel B: Children aged 8 and 9
β 4−8 β 4 P β 7 F E β 8 F E Test β 4 Test β 4 N β 5−7 β 5 P β 7 P Test β 5 Test β 5 N
(2000–09) (2000–01) (2006–07) (2008–09) =β 7=β 8 =β 7=β 8=0 (2002–07) (2002–03) (2006–07) =β 7 =β 7=0
Parent health
Mother in 0.003 0.034 −0.017 −0.024 0.580 0.772 8716 0.050 0.041 0.060 0.769 0.491 5295
Excellent health (+) (0.045) (0.052) (0.063) (0.063) (0.050) (0.065) (0.051)
Father in −0.079* −0.063 −0.103* −0.077 0.802 0.341 8666 −0.037 −0.074 0.009 0.204 0.432 5264
Excellent health (+) (0.045) (0.053) (0.060) (0.060) (0.049) (0.066) (0.047)
Mother’s depression 0.055 −0.008 0.067 0.135 0.568 0.708 8566 −0.128 −0.277* 0.060 0.015 0.049 5223
Score (−) (0.103) (0.120) (0.133) (0.133) (0.121) (0.152) (0.120)
Parent behavior
Family dysfunction 0.004 −0.180 0.108 0.170 0.012 0.031 8622 −0.183* −0.320** −0.014 0.040 0.072 5221
Index (−) (0.097) (0.113) (0.131) (0.128) (0.107) (0.146) (0.109)
Positive interaction −0.307*** −0.345*** −0.303*** −0.257** 0.708 0.005 8733 −0.215** −0.242** −0.181* 0.647 0.049 5304
(from 2 years) (+) (0.091) (0.102) (0.105) (0.120) (0.087) (0.116) (0.100)
Hostile parenting 0.018 0.015 −0.062 0.105 0.359 0.557 8586 0.019 0.025 0.011 0.919 0.983 5215
(from 2 years) (−) (0.137) (0.145) (0.155) (0.156) (0.109) (0.138) (0.114)
Consistent parenting 0.066 −0.010 0.154 0.090 0.305 0.412 8478 0.156 0.156 0.156 0.999 0.301 5142
(from 2 years) (+) (0.094) (0.108) (0.118) (0.113) (0.104) (0.132) (0.112)
Aversive parenting −0.151* −0.232** −0.077 −0.108 0.356 0.192 8702 −0.162* −0.316** 0.030 0.007 0.017 5296
(from 2 years) (−) (0.091) (0.108) (0.115) (0.109) (0.096) (0.125) (0.103)
1. This table displays the estimated policy effects and standard errors (in parentheses). Estimates for parents of children aged 5 to 7 in school are reported in panel A, those for parents of children aged 8 and 9 are reported in panel B. This table shows the average effect for the full post-treatment period (β 4−8, columns 1 and 8), and the effects by wave ( β 4 P to β 8 F E , columns 2 to 4, and columns 9 and 10), based on Table 1. Statistical tests are reported in columns 5, 6, 11 and 12. All the tests show the p-values derived from a Chi-square distribution. Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### 5.1 Estimated Effects for Parents of Preschool Children

For parents with preschool children, we start with a model where the policy effects are not allowed to vary by cycle. Estimated effects are reported in the first column, labeled β 4−8 (2000–2009). The results show that the reform is associated with a significant increase in the mother’s depression score (0.15 SD). No significant associations are found on mothers’ or fathers’ excellent health status. For parental behavior, we find a reduction in positive interactions (0.22 SD) between the child and his/her parents as well as an increase in hostile and aversive parenting behaviors (0.20 SD and 0.17 SD, respectively). We do not find any effects on the family dysfunction index or the consistent parenting score. These results are in line with those of Baker, Gruber, and Milligan (2008).

When we allow the policy effects to vary by wave, β 4 to β 8, we find that the impact on the depression score and aversive parenting persists over time. However, the impact on positive interactions tends to become smaller over time, resulting in a fade-out pattern. For consistent parenting, although we do not find an effect on average, we find a negative effect in the short term (2000–2003) that fades out completely over time (2004–2009). To validate this statement, we run a specification with a dummy for the entire post-period (β 4−8) and another dummy for the full late period (β 7−8). For positive interactions and consistent parenting, we indeed observe coefficients β 7−8 that are consistently positive (indicating a fading out), large in magnitude, and statistically significant. This result, however, appears to be driven mainly by the wave 8 estimates.

We performed several statistical tests to gauge the stability of the effects over time (Table A.6). First, we test the equality of policy effects, where H0: β 4 = β 5 = … = β 8. Except for parental health, hostile and aversive parenting, the null hypothesis of equality of effects between the waves is rejected. This implies that, for half of the outcomes, the effects differ over time. Second, we test whether all effects are equal to zero. We fail to reject the null only for mothers’ health and fathers’ health. This suggests that the policy generally has an impact on families with young children. Third, we test whether the impact is different from zero for both 2006 and 2008 (H0: β 7 = β 8 = 0). These families were fully treated in the later years of the program. For half of the outcomes (health, family dysfunction, and consistent parenting), the null hypothesis is not rejected. Fourth, we test whether the effects observed in the later waves (2006 and 2008) are different from those of the earlier waves (H0: β 456 =β 78). For fathers’ health, positive interaction, and consistent parenting, the null hypothesis of equality is rejected. This suggests that the size of the effects is different between the two periods. Finally, we perform a one-tailed test where we report p-values for the test β 7−8>β 4−6 (excellent health, positive interaction, and consistent parenting) and β 7−8<β 4−6 (depression, family dysfunction, hostile and aversive parenting). Again, for fathers’ health, positive interaction, and consistent parenting, the null hypothesis is rejected. Overall, for families with preschool children, we find that previously documented negative effects are persistent. Only for positive interaction scores and consistent parenting scores, we find less persistence, but this seems to be mainly driven by the much smaller effects found in wave 8 (2008–09).

In light of these findings, our results therefore suggest that for families of preschool children the adverse effects documented early in the reform generally persist in later waves of the reform. There are however two exceptions: for families of preschoolers, we provide evidence that the adverse behavioral effects on positive interaction and consistent parenting were smaller in the later stages of the reform.

### 5.2 Estimated Effects for Parents of Primary School Children

Families of preschool children in waves 1 to 6 can be observed later in life when the children are in school (see Table 1). We can therefore analyze if childcare subsidies can cause changes in parental health and behavior when the policy is no longer contemporaneously effective, that is, when parents no longer have preschoolers in the household. Table 3 presents the estimated impacts of the reform on parents with school age children and no preschoolers. These are the first empirical results on the effects of Quebec’s childcare policy for parents on these age groups.[19]

For parents with children aged 5–7 (Table 3, panel A), the average effect, β 4−8, is significant at the 1 percent level for only one outcome out of eight: positive interaction at −0.31 SD. This is also true for parents of children aged 8 or 9 years old (Table 3, panel B). The estimate of β 5−7, when the effect is constrained to be identical across periods, equals −0.22 SD and is significant at the 5 percent level. We also find that this adverse effect on positive interaction is persistent. The estimates by wave suggest a permanent reduction of positive interaction of −0.35 to −0.26 (β 4 and β 8) for children aged 5–7 years old, and a comparable reduction of −0.24 to −0.18 (β 5 and β 7) for children aged 8 or 9 years old. These effects on positive interaction are relatively similar in magnitude for parents with preschool children (0.22 SD). We also observe a significant reduction in aversive parenting behavior in the early year of the reform for both age groups ( β 4 P = −0.23 and β 5 P = −0.32). These favorable effects do not hold in the long run.

Our findings suggest that the negative effects documented for families with preschoolers generally do not persist once children enter school. More specifically, we find that, over time, only the impact on positive interaction seems to persist. However, given the fact that the common trend assumption does not hold in wave 3 for positive interaction for older children, we are not able to conclude whether the effect on positive interaction documented in younger children persists later in life or not.

Clearly, the reform did not contribute positively to parental health and behaviors on average in Quebec. However, we provide evidence that some of the negative effects of the reform have become smaller over time.

### 5.3 Robustness Checks

Thus far, the sample was divided into several age groups and years accounting for the intensity of treatment. As a robustness test, we take a sample of parents with children aged 5–9 in waves 1 and 2 (pre-period) and compare them to parents with children aged 5–9 in wave 7 (post-period) (youngest child) (column 1 of Table A.7).[20] The results remain similar. We observe no impact of the policy on health and behavior of parents with children aged 5–9 (youngest child), except for the detrimental impact on the positive interaction score.

Tables A.8Tables A.13 present several other robustness checks. Tables A.8 and A.9 present results in which families living on Ontario are used as the control[21] instead of all families from the RofC. Tables A.10 and A.11 include parents of children observed in wave 3, these were excluded from our previous estimates. Parents of children observed in wave 3 are classified pre or post-reform as follows: parents of children aged 0–2 years old and children aged 6–9 years old are classified pre-reform since their children were not eligible to low-fee childcare (see wave 3, Table 1). Parents of children aged 3–5 years old are considered post-reform since their children were eligible to low-fee childcare for a few months to one year.[22] Finally, Tables A.12 and A.13 present estimates of our base model using the RofC as a control group, but this time adding a linear time trend. None of these alternatives modify our main conclusions.[23]

We also estimated the models for single-parent families (Tables A.14 and A.15). Despite the fact that single parents were possibly affected by other policies during this time period, it is interesting to observe that the estimated effects are qualitatively very similar to effects on two-parent families. The effects for single parents are generally larger in magnitude than those of two-parent families, but given that the standard errors are also larger, they are generally not significant. For single parents of children age 5–9 years old, we find that the policy had limited impacts, except for an increase of the maternal depression score of about 0.45 SD for mothers of 5- to 7-year-olds. Therefore, for single parents, we must conclude that the large standard errors limit our ability to draw strong conclusions.

### 5.4 Estimated Effects by Maternal Education

In this section, we analyze whether the estimated effects differ according to maternal education. To capture the differential effects by maternal education, we interact a maternal education dummy with the our post-treatment dummies (W t Q it ). Our maternal education dummy equals one if the mother has no more than a high-school degree (low education) and zero if the mother has some post-secondary education or more (high education). Tables 4 and 5 present the estimated effects, for parents with preschool children and for parents with children aged 5–9 years old in school (youngest child). In each table, the first two columns present estimates from our aggregated model in which the effect of the reform is aggregated over all post-treatment waves. The other columns relate to the model in which we allow the reform effect to vary over survey waves.

### Table 4:

Estimated effects of the policy on the health and behavior of parents with children aged 1 to 5 (not in school) by maternal education.

Variable Model 1: Aggregated model Model 2: Model with wave specific effects interacted with maternal education
β 4−8 β 4−8 β 4 P β 4 P β 5 F E β 5 F E β 6 F E β 6 F E β 7 F L β 7 F L β 8 F L β 8 F L N
(2000–10) × LowEduc (2000–01) × LowEduc (2002–03) × LowEduc (2004–05) × LowEduc (2006–07) × LowEduc (2008–09) × LowEduc
Parent health
Mother in 0.022 −0.059 0.037 −0.059 0.021 −0.096 0.017 −0.099 0.011 0.015 0.026 −0.020 40,865
Excellent health (+) (0.029) (0.050) (0.034) (0.062) (0.037) (0.062) (0.039) (0.062) (0.039) (0.071) (0.039) (0.076)
Father in 0.033 −0.102** 0.013 −0.069 −0.016 −0.016 0.002 −0.077 0.073** −0.154** 0.071* −0.200*** 40,639
Excellent health (+) (0.027) (0.051) (0.033) (0.064) (0.038) (0.066) (0.038) (0.068) (0.037) (0.074) (0.038) (0.069)
Mother’s depression 0.134** 0.073 0.096 0.130 0.100 0.097 0.140 0.161 0.302*** −0.214 0.030 0.118 39,889
Score (−) (0.056) (0.105) (0.064) (0.167) (0.067) (0.134) (0.087) (0.153) (0.090) (0.153) (0.069) (0.145)
Parent behavior
Family dysfunction 0.059 −0.086 0.113* −0.108 −0.057 −0.083 0.062 −0.136 0.008 0.029 0.145* −0.048 40,336
Index (−) (0.058) (0.108) (0.065) (0.141) (0.073) (0.142) (0.078) (0.135) (0.079) (0.154) (0.081) (0.157)
Positive interaction −0.210*** −0.022 −0.271*** −0.028 −0.201*** −0.063 −0.332*** 0.017 −0.168** 0.150 −0.119* 0.015 30,124
(from 2 years) (+) (0.055) (0.098) (0.071) (0.122) (0.066) (0.121) (0.078) (0.138) (0.072) (0.137) (0.070) (0.150)
Hostile parenting 0.189*** 0.051 0.166** 0.070 0.235*** −0.105 0.158 0.278 0.266*** −0.023 0.125 −0.033 29,654
(from 2 years) (−) (0.073) (0.140) (0.081) (0.158) (0.088) (0.159) (0.097) (0.184) (0.094) (0.200) (0.096) (0.184)
Consistent parenting −0.021 −0.205* −0.156** −0.081 −0.094 −0.185 0.032 −0.275 0.007 −0.052 0.088 −0.334* 29,272
(from 2 years) (+) (0.064) (0.121) (0.075) (0.154) (0.081) (0.148) (0.090) (0.168) (0.084) (0.169) (0.080) (0.175)
Aversive parenting 0.157** 0.047 0.105 −0.042 0.134 −0.013 0.200** 0.153 0.192** 0.082 0.156* 0.124 29,982
(from 2 years) (−) (0.064) (0.113) (0.074) (0.134) (0.082) (0.140) (0.089) (0.146) (0.086) (0.190) (0.089) (0.180)
1. This table displays the estimated policy effects and standard errors (in parentheses) by maternal education for parents of children aged 1 to 5 not yet in school. The first two columns are estimates from one model where waves 4 to 8 are aggregated. The next 10 columns are estimates from another model where the policy effect is allowed to vary by wave and maternal education. Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table 5:

Estimated effects of the policy on the health and behavior of parents with children aged 5 to 9 in school (youngest child) by maternal education.

Panel A: Children aged 5 to 7
Variable Model 1: Aggregated model Model 2: Model with wave specific effects interacted with maternal education
β 4−8 β 4−8 β 4 P β 4 P β 7 F E β 7 F E β 8 F E β 8 F E N
(2000–09) × LowEduc (2000–01) × LowEduc (2006–07) × LowEduc (2008–09) × LowEduc
Parent health
Mother in −0.030 0.100 0.037 −0.002 −0.057 0.132 −0.086 0.261** 8716
Excellent health (+) (0.059) (0.092) (0.065) (0.104) (0.074) (0.121) (0.076) (0.130)
Father in −0.116** 0.122 −0.117* 0.156 −0.127* 0.079 −0.102 0.090 8666
Excellent health (+) (0.056) (0.094) (0.065) (0.108) (0.070) (0.119) (0.071) (0.124)
Mother’s depression 0.152 −0.317 −0.003 −0.049 0.213 −0.504* 0.288* −0.625** 8566
Score (−) (0.119) (0.207) (0.135) (0.236) (0.146) (0.273) (0.159) (0.266)
Parent behavior
Family dysfunction 0.097 −0.274 −0.071 −0.300 0.177 −0.191 0.223 −0.109 8622
Index (−) (0.124) (0.192) (0.142) (0.212) (0.164) (0.260) (0.155) (0.256)
Positive interaction −0.241** −0.188 −0.273** −0.207 −0.237* −0.188 −0.208 −0.091 8733
(from 2 years) (+) (0.111) (0.185) (0.123) (0.201) (0.124) (0.238) (0.141) (0.221)
Hostile parenting 0.060 −0.134 0.035 −0.067 −0.004 −0.197 0.153 −0.179 8586
(from 2 years) (−) (0.200) (0.256) (0.210) (0.264) (0.214) (0.299) (0.216) (0.331)
Consistent parenting 0.027 0.122 −0.077 0.193 0.177 −0.105 0.013 0.324 8478
(from 2 years) (+) (0.128) (0.205) (0.148) (0.226) (0.147) (0.273) (0.141) (0.291)
Aversive parenting −0.083 −0.234 −0.152 −0.210 −0.025 −0.199 −0.057 −0.275 8702
(from 2 years) (−) (0.115) (0.188) (0.130) (0.205) (0.140) (0.217) (0.133) (0.244)
Panel B: Children aged 8 and 9
Variable Model 1: Aggregated model Model 2: Model with wave specific effects interacted with maternal education
β 5−7 β 5−7 β 5 P β 5 P β 7 F E β 7 F E N
(2002–07) × LowEduc (2002–03) × LowEduc (2006–07) × LowEduc
Parent health
Mother in 0.057 −0.020 0.084 −0.100 0.026 0.110 5295
Excellent health (+) (0.066) (0.098) (0.092) (0.119) (0.064) (0.108)
Father in −0.033 −0.015 −0.087 0.018 0.025 −0.043 5264
Excellent health (+) (0.061) (0.097) (0.083) (0.116) (0.061) (0.105)
Mother’s depression −0.181 0.135 −0.423*** 0.324 0.081 −0.071 5223
Score (−) (0.134) (0.255) (0.155) (0.281) (0.148) (0.255)
Parent behavior
Family dysfunction −0.102 −0.201 −0.164 −0.352 −0.042 0.104 5221
Index (−) (0.118) (0.227) (0.163) (0.273) (0.118) (0.233)
Positive interaction −0.052 −0.443** 0.006 −0.605*** −0.118 −0.176 5304
(from 2 years) (+) (0.114) (0.172) (0.161) (0.202) (0.112) (0.187)
Hostile parenting 0.065 −0.116 0.049 −0.063 0.083 −0.209 5215
(from 2 years) (−) (0.135) (0.205) (0.176) (0.241) (0.134) (0.217)
Consistent parenting 0.200 −0.123 0.230 −0.186 0.165 −0.017 5142
(from 2 years) (+) (0.139) (0.214) (0.182) (0.250) (0.137) (0.234)
Aversive parenting −0.194* 0.098 −0.378*** 0.169 0.002 0.082 5296
(from 2 years) (−) (0.117) (0.190) (0.144) (0.227) (0.126) (0.202)
1. This table displays the estimated policy effects and standard errors (in parentheses) by maternal education. Estimates for parents of children aged 5 to 7 in school are reported in panel A, those for parents of children aged 8 and 9 are reported in panel B. The first two columns are estimates from one model where post-reform waves are aggregated. The next 4 to 6 columns are estimates from another model where the policy effect is allowed to vary by wave and maternal education. Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

For families with preschool children, the results are similar between low-educated and high-educated mothers: the coefficients on the interaction with maternal education are generally not statistically significant. There is one exception. Paternal health appears to be lower post-reform in families in which the mothers have a lower level of education.

For families with children in school, the policy generally has no impact on parental health and behaviors, and maternal education does not appear to change these patterns. Results on positive interaction need to be interpreted with caution since the common trend assumption did not hold for this outcome.

### 5.5 Discussion

The Quebec daycare reform appears to have had a negative effect on the well-being and behavior of families with young children (1–5 years old not in school). In line with previous studies, we show that the policy was followed by an increase in the maternal depression score and a deterioration of parental behaviors. Our results suggest that the effects on parents of preschool children generally persist over time up to 2009. However, we also note some improvements for positive interaction and consistent parenting in the later years of the reform. This is related to the results by Herbst and Tekin (2014) who discuss how receiving childcare subsidies affects parental well-being.

We propose some mechanisms that could explain these findings. First, the allocation of time between work and leisure for the mother, and home and childcare for the child, changed following the Quebec policy. Haeck et al.(2015) document a substantial increase in hours spent in childcare, combined with an increase in labor force participation and number of weeks worked for the mother. The time spent by the mother with the child is thereby reduced, which may have implications for child and maternal well-being. Going back to the labor market implies a busier schedule leading to more stress, especially if work, family, and parenting must be reconciled harmoniously. Higher stress levels may worsen health outcomes and reduce the quality and frequency of child-parent interactions. Habits and types of activities between the child and the parent may change, at least in the short run, until the mother is physically and psychologically fit to work again or work more intensively (Herbst and Tekin 2014). Roggman et al. (1994) show a link between increased leisure time and social support of Head Start parents and positive parenting and parental psychological well-being.

Clearly, increased labor force participation necessarily reduces the time available to parents and many of the measures of parental behavior used in this study are based on questions that relate to the frequency of the interactions between the parent and the child. However, while the frequency of parental interactions may have decreased following the reform, the overall quality of the interactions for a given quantity of time spent with the child is not directly captured by our measures. It is therefore possible that quality has remained stable.

Second, long hours in daycare, in particular if care is of insufficient quality, may affect the child’s behavior and temperament at home, increasing tensions within the household and affecting parental health and behavior. Japel, Tremblay, and Côté (2005) show that the quality of care, a few years after the new policy was implemented, was low in many situations, and even more so for low income families and for family-based care. Combined with the increased hours of care mentioned above, it is likely that this contributed in part to the reduction in the well-being of parents with preschoolers.

Third, as pointed out by Herbst and Tekin (2014), the effect of childcare subsidies may depend on the characteristics of the job mothers obtain in the labor market. While these results must be interpreted with caution, the persistent negative effects we uncover on positive interaction once children are in school may be due to jobs that are of lower quality that offer, for example, less flexible working hours. It is possible that subsidized mothers endure substantial job-related stress because of the inaccessibility of reliable transportation, presence of hazardous work conditions, or unpredictable and non-standard work schedules. Alternatively, mothers may move into family-friendly work environments with access to high-quality health insurance options, paid family and medical leave, and on-site childcare arrangements. The existing empirical evidence, however, suggests that parents receiving subsidies work disproportionately in low-wage occupations that offer few training opportunities and benefits (Berger and Black 1992; Danziger, Ananat, and Browning 2004; Davis and Jefferys 2007; Ha 2009; Okuyama and Weber 2001). In Quebec, the childcare program is universal and not income based. However, Lefebvre, Merrigan, and Verstraete (2009) document that over the long run, once the child is in school, the labor supply effect was mainly concentrated among less educated mothers who may also have lower quality jobs.

Fourth, childcare subsidies may also affect family well-being through changes in income and consumption (Herbst and Tekin 2014). For example, the policy could impact work-related expenses and also expenditures related to children, family goods and services with a collective dimension (Lundberg, Pollak, and Terence 1997; Ward-Batts 2008).

Finally, changes in the well-being of the child may directly influence that of the parents, or vice versa. Haeck, Lebihan, and Merrigan (2018), following up on Baker, Gruber, and Milligan (2008),[24] show that the policy had negative effects on several aspects of the well-being of preschooler that are still present 10 years after the policy was implemented. For certain behavioral outcomes at ages 4 and 5, they document some fade-out over the years. Haeck, Lebihan, and Merrigan (2018) suggest that this could be explained by the efforts of the Quebec government to improve the quality of the staff and the educational program. Interestingly, this fade-out pattern is observed in our study on positive interaction and consistent parenting for mothers with preschoolers. Therefore, it is possible that the negative effects of the policy on mothers and parenting measures could be the result of the interaction between the increased labor force participation of mothers and the increase in health and behavioral problems of preschoolers.

However, there is little evidence that these effects persist later in the child’s life. For families with children aged 5–9 years old, most coefficients are not different from zero. This is consistent with Haeck, Lebihan, and Merrigan (2018) showing that the policy had much smaller adverse impacts on children aged 5–9 years old. We conjecture that if there are adverse effects on parental interaction for school age children, they are mainly due to a time constraint resulting from the increased labor force participation of mothers.

In summary, the impact of the reform on parents, except for positive interaction, is essentially contemporaneous or direct, that is, when children who are not yet in school. However, even among families with preschoolers, we find that some of the effects documented in the first years of the program appear to fade in the later years of the policy.

## 6 Conclusion

Our paper shows that the Quebec childcare policy had detrimental effects on parental mental health and behavior for parents of preschoolers. However, the effects documented early on in the reform on parents of children not yet in school appear to diminish over survey cycles for positive interaction and consistent parenting, but less so for positive interaction. Effects remain persistent for hostile parenting and aversive parenting. Also, our results suggest that once children enter school and the parents no longer have preschool children in the household, most detrimental effects fade in later survey cycles.

We are unable to identify the exact drivers of these results, but evidence presented in previous studies of this reform suggests a few possibilities. First, as mentioned earlier, previous research documented that the childcare reform increased maternal labor force participation, work hours and children’s time spent in childcare. The effects on maternal work were documented both in the preschool years and once children were in school. These effects mechanically lower the number of hours available for parents to interact with their child. Second, childcare quality was considered fairly low, which may have had an impact on parental behavior. Third, negative effects on preschoolers also appear to fade once children enter school. Parental behavior and mental health therefore seem to mirror what has been previously documented in children.

One of the limits of our study is that while the frequency of parental interactions may have decreased following the reform, the overall quality of the interactions for a given quantity of time spent with the child is not directly captured by our measures. Another limit relates to the lack of data after 2008 which limits our ability to follow children in more recent years of the program.

Our results, in conjunction with those of previous studies, suggest that the negative effects of the policy on preschool children and their parents should be of concern for policy makers seeking to provide universal care to children. However, concerns about the long run effects on parents and children once children enter school should be limited given the evidence presented here and in Haeck, Lebihan, and Merrigan (2018). Any major policy that seeks to radically increase the labor force participation of mothers with young children through highly subsidized childcare must thoroughly consider all family dimensions, in particular physical and behavioral, before its implementation. Finally, changes in childcare subsidies may affect child and parental well-being through increased household income, enlarging consumption possibilities. The impact of the Quebec childcare reform on household finances and consumption need to be further studied in the future.

Corresponding author: Laetitia Lebihan, Part-time Professor, Department of Economics, University of Ottawa, Faculty of Social Sciences, 120 University Private, Ottawa, ON, K1N 6N5, Canada, E-mail:

# Acknowledgement

The analysis is based on Statistics Canada’s National Longitudinal Survey of Children and Youth (NLSCY) restricted-access Micro Data Files, which contain anonymous data. All computations on these micro-data were prepared by the authors who assume responsibility for the use and interpretation of these data. Declarations of interest: none.

1. Conflict of interest: The authors declare that they have no conflict of interest.

Appendix

### Figure A.1:

Total fertility rate over time.

Data comes from Statistics Canada, Canadian Vital Statistics, Birth Database (CVSB) and Centre for Demography (population estimates), Table: 13-10-0418-01 (formerly CANSIM 102-4505). Total fertility rate is an estimate of the average number of live births a female can be expected to have in her lifetime, based on the age-specific fertility rates (ASFR) of a given year. The total fertility rate (TFR) = (SUM of single year of age-specific fertility rates)/1000.

### Table A.1:

Eligibility for low-fee childcare by child’s age and NLSCY wave.

Child’s age Eligibility
No Partially β P Full early β FE Full late β FL
(1–3 years) (4–5 years) (4–5 years)
1–5 years not in school Wave 1-2 Wave 4 Wave 5-6 Wave 7-8
5–7 years in school Wave 1-2 Wave 4-6 Wave 7-8 n.a
8–9 years Wave 1-4 Wave 5-7 n.a n.a
1. This table shows eligibility for the Quebec childcare program for families with a child according to his age and NLSCY wave. We distinguish four eligibility cohorts: (1) completely ineligible families; (2) partially eligible families i.e. with 1–3 years of program eligibility; (3) fully eligible families from birth of the child and during the first years of the program; and (4) fully eligible families with children from birth since 2006. The term n.a (not available) means that the child is eligible for low-fee childcare spaces but data for this age group in this wave are not available. For families with children aged 5–9, it is the youngest child in the family.

### Table A.2:

Parent outcomes index component (Appendix).

Parent outcome index Questions Types of questions
Family dysfunction index Planning family activities is difficult because we misunderstand each other Strongly agree (1) to
In times of crisis we can turn to each other for support strongly disagree (4)
Individuals, in the family, are accepted for what they are
We avoid discussing our fears or concerns
We express feelings to each other
There are lots of bad feelings in our family
We feel accepted for what we are
Making decisions is a problem for our family
We are able to make decisions about how to solve problems
We don’t get along well together
We confide in each other
Positive interaction How often do you praise this child, by saying something like ‘Good for you!’ Never (1) to many
Or ‘What a nice thing you did!’ or ‘That’s good going!’? times each day (5)
How often do you and this child talk or play with each other, focusing
Attention on each other for 5 min or more, just for fun?
How often do you and this child laugh together?
How often do you do something special with this child that he enjoys?
How often do you play sports, hobbies or games with this child?
Hostile/ineffective parenting How often do you get annoyed with this child for saying or doing something he Never (1) to many
is not supposed to? Of all the times that you talk to this child about his behaviour, times each day (5)
what proportion is praise? Of all the times that you talk to this child about his behaviour,
what proportion is disapproval? How often do you get angry when you punish this child?
How often do you think that the kind of punishment you give this child depends on your mood?
How often do you feel you are having problems managing this child in general?
How often do you have to discipline this child repeatedly for the same thing?
Consistency parenting When you give this child a command, what proportion of the time do you make Never (1) to all the
sure that he does it? If you tell this child he will get punished if he doesn’t stop time (5)
doing something And he keeps doing it, how often will you punish him?
How often does this child get away with things that you feel should have been punished?
How often is this child able to get out of a punishment when he really sets his mind to it?
How often when you discipline this child, does he ignore the punishment?
Aversive parenting How often do you raise your voice, scold or yell at him, when the child breaks the rules? Never (1) to always (5)
How often do you calmly discuss the problem, when the child breaks the rules?
How often do you use physical punishment, when the child breaks the rules?
How often do you describe alternative ways of behaving that are acceptable,
When the child breaks the rules?

### Table A.3:

Summary statistics for two-parent families with children aged 1–9.

Pre-policy Post-policy Pre-policy Post-policy
Child is a boy 0.51 0.51 0.51 0.51
Mother
Less than high school 0.17 0.11 0.10 0.08
High school diploma 0.17 0.15 0.20 0.19
Some post-secondary 0.24 0.16 0.28 0.14
Post-secondary degree 0.42 0.57 0.42 0.59
Age 14–24 at birth 0.20 0.20 0.17 0.16
Age 25–29 at birth 0.42 0.38 0.37 0.32
Age 30–34 at birth 0.29 0.30 0.32 0.35
Age 35 or more at birth 0.09 0.12 0.14 0.18
Immigrant 0.08 0.10 0.19 0.21
Father
Less than high school 0.19 0.15 0.14 0.10
High school diploma 0.17 0.19 0.19 0.21
Some post-secondary 0.21 0.16 0.23 0.13
Post-secondary degree 0.43 0.51 0.44 0.56
Age 14–24 at birth 0.08 0.09 0.07 0.07
Age 25–29 at birth 0.32 0.29 0.27 0.24
Age 30–34 at birth 0.39 0.36 0.38 0.37
Age 35 or more at birth 0.21 0.26 0.27 0.32
Immigrant 0.09 0.13 0.18 0.20
Family
Rural region 0.19 0.15 0.16 0.12
Region < 30 K 0.12 0.12 0.15 0.16
Region 30–99,999 K 0.09 0.09 0.07 0.09
Region 100–499 K 0.08 0.06 0.22 0.19
Region > 499 K 0.52 0.58 0.39 0.44
None older sibling 0.47 0.47 0.41 0.43
One older sibling 0.36 0.38 0.37 0.39
At least two older siblings 0.17 0.16 0.22 0.19
None younger sibling 0.58 0.66 0.57 0.65
One younger sibling 0.34 0.29 0.34 0.30
At least two younger siblings 0.08 0.05 0.09 0.06
Same age siblings 0.03 0.02 0.02 0.03
N 4387 8,577 19,367 47,128
1. This table displays the weighted summary statistics for children, mothers, fathers, and families (we use the sample weights from Statistics Canada). The statistics are presented by region: Quebec and the Rest of Canada, for the pre-reform and post-reform periods as described in Table 1. Wave 3 of the NLSCY is excluded for children 1–7 years old. All statistics appearing in the table are proportions.

### Table A.4:

Event-study on the health and behavior of parents with children aged 1 to 5 not in school (Appendix).

Variable β 2 β 3 β 4 P β 5 F E β 6 F E β 7 F L β 8 F L N
(1996–97) (1998–99) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09)
Parent health
Mother in 0.007 0.016 0.023 −0.003 −0.008 0.008 0.018 50,292
Excellent health (+) (0.040) (0.035) (0.037) (0.038) (0.038) (0.040) (0.041)
Father in 0.011 0.020 −0.000 −0.015 −0.014 0.042 0.033 50,054
Excellent health (+) (0.042) (0.038) (0.038) (0.040) (0.040) (0.039) (0.042)
Mother’s depression 0.077 0.056 0.174** 0.168** 0.232*** 0.312*** 0.093 49,217
Score (−) (0.083) (0.067) (0.078) (0.071) (0.084) (0.087) (0.073)
Parent behavior
Family dysfunction 0.117 −0.006 0.145* −0.017 0.087 0.060 0.187** 49,681
Index (−) (0.085) (0.074) (0.075) (0.079) (0.080) (0.083) (0.084)
Positive interaction 0.121 −0.038 −0.214*** −0.157** −0.260*** −0.084 −0.056 34,703
(from 2 years) (+) (0.085) (0.075) (0.073) (0.072) (0.077) (0.074) (0.075)
Hostile parenting 0.007 0.101 0.190** 0.206** 0.253** 0.264** 0.125 34,205
(from 2 years) (−) (0.111) (0.096) (0.092) (0.097) (0.108) (0.104) (0.104)
Consistent parenting 0.178* 0.034 −0.077 −0.051 0.046 0.084 0.117 33,815
(from 2 years) (+) (0.103) (0.091) (0.087) (0.089) (0.098) (0.093) (0.090)
Aversive parenting −0.084 0.096 0.048 0.087 0.205** 0.166* 0.137 34,541
(from 2 years) (−) (0.090) (0.087) (0.083) (0.084) (0.089) (0.091) (0.093)
1. Each set of two lines is a separate regression. All available cycles are included and cycle 1 (1994–95) is the reference. Standard errors are in parentheses and statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. Estimates are obtained for two-parent families. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.5:

Event-study on the health and behavior of parents with children aged 5 to 9 in school (Appendix).

Panel A: Children aged 5 to 7
β 2 β 3 β 4 P β 7 F E β 8 F E N
Variable (1996–97) (1998–99) (2000–01) (2006–07) (2008–09)
Parent health
Mother in −0.012 −0.004 0.024 −0.028 −0.033 11,553
Excellent health (+) (0.070) (0.073) (0.059) (0.068) (0.070)
Father in 0.062 0.021 −0.031 −0.071 −0.045 11,497
Excellent health (+) (0.070) (0.070) (0.062) (0.070) (0.067)
Mother’s depression −0.143 0.008 −0.093 −0.016 0.051 11,372
Score (−) (0.173) (0.158) (0.163) (0.171) (0.172)
Parent behavior
Family dysfunction −0.103 −0.125 −0.238* 0.052 0.107 11,426
Index (−) (0.165) (0.153) (0.136) (0.150) (0.150)
Positive interaction −0.060 −0.383*** −0.378*** −0.339*** −0.289** 11,571
(from 2 years) (+) (0.158) (0.143) (0.118) (0.118) (0.127)
Hostile parenting 0.269 0.274** 0.166 0.091 0.259* 11,407
(from 2 years) (−) (0.237) (0.137) (0.120) (0.136) (0.137)
Consistent parenting 0.003 −0.148 −0.008 0.150 0.089 11,300
(from 2 years) (+) (0.168) (0.136) (0.134) (0.139) (0.142)
Aversive parenting −0.155 0.058 −0.319*** −0.162 −0.188 11,531
(from 2 years) (−) (0.150) (0.124) (0.118) (0.130) (0.121)
Parent health
Mother in −0.049 −0.060 −0.034 0.006 0.025 5295
Excellent health (+) (0.093) (0.096) (0.100) (0.092) (0.081)
Father in −0.043 −0.041 0.033 −0.088 −0.005 5264
Excellent health (+) (0.095) (0.099) (0.100) (0.087) (0.072)
Mother’s depression 0.438* 0.259 −0.137 −0.129 0.208 5223
Score (−) (0.230) (0.249) (0.189) (0.181) (0.155)
Parent behavior
Family dysfunction 0.311 0.095 0.013 −0.209 0.098 5221
Index (−) (0.213) (0.202) (0.219) (0.202) (0.177)
Positive interaction 0.060 −0.124 −0.397** −0.346** −0.285** 5304
(from 2 years) (+) (0.172) (0.151) (0.158) (0.144) (0.136)
Hostile parenting 0.506** 0.188 0.302 0.281 0.267 5215
(from 2 years) (−) (0.216) (0.237) (0.234) (0.204) (0.181)
Consistent parenting −0.260 −0.030 −0.115 0.049 0.049 5142
(from 2 years) (+) (0.185) (0.210) (0.193) (0.172) (0.159)
Aversive parenting 0.035 −0.149 −0.154 −0.377** −0.030 5296
(from 2 years) (−) (0.198) (0.188) (0.181) (0.180) (0.162)
1. Each set of two lines is a separate regression. All available cycles are included and cycle 1 (1994–95) is the reference. Standard errors are in parentheses and statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. Estimates are obtained for two-parent families. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.6:

Statistical tests and estimated effects of the policy on the health and behavior of parents with children aged 1 to 5 not in school (Appendix).

Variable β 4−8 β 4 P β 5 F E β 6 F E β 7 F L β 8 F L Test β 4 Test β 4 Test β 7 Test β 456 One-tail N
(2000–09) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09) = … = β 8 = … = β 8=0 = β 8 = 0 = β 78 Tests
Parent health
Mother in 0.005 0.021 −0.006 −0.011 0.005 0.015 0.851 0.922 0.908 0.742 0.371 40,865
Excellent health (+) (0.024) (0.030) (0.031) (0.033) (0.034) (0.035)
Father in 0.005 −0.005 −0.020 −0.018 0.038 0.028 0.350 0.487 0.483 0.049 0.024 40,639
Excellent health (+) (0.024) (0.029) (0.032) (0.032) (0.032) (0.034)
Mother’s depression 0.151*** 0.132* 0.125** 0.187** 0.270*** 0.051 0.047 0.004 0.002 0.886 0.557 39,889
Score (−) (0.049) (0.068) (0.061) (0.075) (0.079) (0.063)
Parent behavior
Family dysfunction 0.034 0.084 −0.080 0.024 −0.001 0.124* 0.020 0.035 0.140 0.301 0.850 40,336
Index (−) (0.050) (0.059) (0.064) (0.067) (0.070) (0.072)
Positive interaction −0.217*** −0.278*** −0.219*** −0.325*** −0.151** −0.121* 0.015 0.000 0.042 0.001 0.001 30,124
(from 2 years) (+) (0.048) (0.061) (0.060) (0.067) (0.064) (0.064)
Hostile parenting 0.201*** 0.187*** 0.202*** 0.251*** 0.261*** 0.118 0.508 0.026 0.009 0.627 0.313 29,654
(from 2 years) (−) (0.064) (0.072) (0.077) (0.090) (0.085) (0.084)
Consistent parenting −0.076 −0.177*** −0.149** −0.053 −0.015 0.022 0.035 0.028 0.879 0.008 0.004 29,272
(from 2 years) (+) (0.054) (0.066) (0.071) (0.078) (0.074) (0.071)
Aversive parenting 0.170*** 0.092 0.130* 0.250*** 0.208*** 0.179** 0.354 0.025 0.014 0.464 0.768 29,982
(from 2 years) (−) (0.055) (0.067) (0.071) (0.076) (0.077) (0.081)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F L ), based on Table 1. All the tests show the p-values derived from a Chi-square distribution. For one-tail tests, we report p-values for test β 78 > β 456 (excellent health; positive interaction and consistent parenting) and β 78 < β 456 (depression; family dysfunction index; hostile and aversive parenting). We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for two-parent families with children aged 1–5 not in school. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.7:

Estimated effects of the policy on the health and behavior of parents with children aged 5 to 9 in school (wave 7 only) (Appendix).

Variable Two-parent families Education Only Ontario Trend One-parent families All families
β 7 N β 7 β 7 × LowEduc N β 7 N β 7 N β 7 N β 7 N
(2006–07) (2006–07) (2006–07) (2006–07) (2006–07) (2006–07) (2006–07)
Parent health
Mother in 0.013 7443 −0.033 0.144* 7443 0.030 3061 0.013 7443 0.033 1959 0.006 10,107
Excellent health (+) (0.046) (0.054) (0.086) (0.051) (0.046) (0.078) (0.038)
Father in −0.053 7420 −0.073 0.074 7420 −0.058 3053 −0.053 7420 −0.069* 8071
Excellent health (+) (0.044) (0.054) (0.085) (0.049) (0.044) (0.042)
Mother’s depression 0.017 7345 0.106 −0.269 7345 0.050 3025 0.017 7345 0.284 1950 0.089 9990
Score (−) (0.099) (0.106) (0.203) (0.111) (0.099) (0.190) (0.089)
Parent behavior
Family dysfunction 0.034 7378 0.087 −0.117 7378 0.043 3043 0.034 7378 −0.249 1918 −0.034 9992
Index (−) (0.092) (0.116) (0.191) (0.101) (0.092) (0.205) (0.081)
Positive interaction −0.307*** 7459 −0.234** −0.220 7459 −0.344*** 3071 −0.307*** 7459 −0.134 1973 −0.253*** 10,156
(from 2 years) (+) (0.081) (0.100) (0.174) (0.094) (0.081) (0.159) (0.071)
Hostile parenting −0.038 7324 0.032 −0.213 7324 0.057 3024 −0.038 7324 0.255 1936 0.064 9968
(from 2 years) (−) (0.111) (0.146) (0.209) (0.129) (0.111) (0.176) (0.091)
Consistent parenting 0.169* 7241 0.220** −0.132 7241 0.160 2985 0.169* 7241 −0.239 1915 0.052 9854
(from 2 years) (+) (0.089) (0.110) (0.194) (0.101) (0.089) (0.162) (0.080)
Aversive parenting −0.065 7446 −0.074 0.017 7446 −0.051 3066 −0.065 7446 0.040 1966 0.004 10,131
(from 2 years) (−) (0.088) (0.112) (0.169) (0.102) (0.088) (0.175) (0.079)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. Wave 7 is the only post-reform wave. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for families with children aged 5–9 in school (youngest child). Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.8:

Estimated effects of the policy on the health and behavior of parents with children aged 1 to 5 not in school (Quebec and Ontario) (Appendix).

Variable β 4−8 β 4 P β 5 F E β 6 F E β 7 F L β 8 F L β 78 N
(2000–09) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09) vs β 4−8
Parent health
Mother in 0.026 0.035 0.004 0.018 0.026 0.044 0.016 16,915
Excellent health (+) (0.027) (0.033) (0.036) (0.037) (0.037) (0.039) (0.027)
Father in 0.024 0.002 −0.004 −0.008 0.058 0.065* 0.065** 16,823
Excellent health (+) (0.026) (0.032) (0.036) (0.036) (0.036) (0.038) (0.027)
Mother’s depression 0.173*** 0.140* 0.196*** 0.191** 0.307*** 0.045 −0.003 16,491
Score (−) (0.057) (0.076) (0.069) (0.086) (0.087) (0.072) (0.058)
Parent behavior
Family dysfunction −0.003 0.024 −0.156** −0.001 0.010 0.087 0.091* 16,708
Index (−) (0.057) (0.067) (0.072) (0.078) (0.078) (0.081) (0.055)
Positive interaction −0.221*** −0.274*** −0.240*** −0.322*** −0.164** −0.114 0.166*** 12,244
(from 2 years) (+) (0.056) (0.069) (0.070) (0.076) (0.074) (0.075) (0.058)
Hostile parenting 0.199*** 0.200** 0.205** 0.226** 0.274*** 0.102 −0.027 12,065
(from 2 years) (−) (0.074) (0.084) (0.089) (0.101) (0.099) (0.097) (0.063)
Consistent parenting −0.054 −0.145* −0.137* −0.051 −0.009 0.075 0.148** 11,913
(from 2 years) (+) (0.063) (0.077) (0.081) (0.090) (0.087) (0.084) (0.058)
Aversive parenting 0.164** 0.089 0.110 0.217** 0.272*** 0.146 0.066 12,187
(from 2 years) (−) (0.068) (0.080) (0.086) (0.090) (0.092) (0.098) (0.060)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F L ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for two-parent families with children aged 1–5 not in school (only Quebec and Ontario). Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.9:

Estimated effects of the policy on the health and behavior of parents with children aged 5 to 9 in school (Quebec and Ontario; youngest child) (Appendix).

Variable Panel A: Children aged 5 to 7 Panel B: Children aged 8 and 9
β 4−8 β 4 P β 7 F E β 8 F E Test β 4 Test β 4 N β 5−7 β 5 P β 7 P Test β 5 Test β 5 N
(2000–09) (2000–01) (2006–07) (2008–09) =β 7 = β 8 =β 7 = β 8 = 0 (2002–07) (2002–03) (2006–07) = β 7 =β 7 = 0
Parent health
Mother in 0.036 0.075 0.013 0.003 0.535 0.602 3301 0.046 0.020 0.076 0.477 0.422 2265
Excellent health (+) (0.052) (0.060) (0.069) (0.070) (0.060) (0.081) (0.059)
Father in −0.057 −0.033 −0.094 −0.051 0.663 0.572 3292 −0.015 −0.051 0.026 0.311 0.594 2251
Excellent health (+) (0.051) (0.062) (0.067) (0.067) (0.057) (0.077) (0.056)
Mother’s depression 0.090 0.013 0.100 0.191 0.513 0.570 3257 −0.167 −0.353* 0.046 0.035 0.106 2223
Score (−) (0.117) (0.142) (0.147) (0.145) (0.146) (0.203) (0.134)
Parent behavior
Family dysfunction 0.015 −0.180 0.102 0.209 0.020 0.050 3279 −0.244** −0.422*** −0.043 0.024 0.028 2230
Index (−) (0.109) (0.131) (0.143) (0.141) (0.115) (0.160) (0.119)
Positive interaction −0.337*** −0.368*** −0.369*** −0.259* 0.580 0.006 3313 −0.199* −0.221 −0.174 0.763 0.158 2276
(from 2 years) (+) (0.104) (0.117) (0.122) (0.135) (0.104) (0.141) (0.115)
Hostile parenting 0.121 0.102 0.052 0.222 0.394 0.453 3262 0.113 0.154 0.065 0.578 0.646 2238
(from 2 years) (−) (0.149) (0.160) (0.171) (0.168) (0.131) (0.164) (0.139)
Consistent parenting 0.023 −0.061 0.152 0.007 0.204 0.359 3222 0.078 0.042 0.122 0.593 0.617 2206
(from 2 years) (+) (0.110) (0.129) (0.133) (0.132) (0.115) (0.148) (0.125)
Aversive parenting −0.179* −0.310** −0.092 −0.083 0.166 0.108 3306 −0.147 −0.309** 0.038 0.027 0.617 2271
(from 2 years) (−) (0.106) (0.130) (0.132) (0.127) (0.117) (0.156) (0.120)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F E ), based on Table 1. All the tests show the p-values derived from a Chi-square distribution. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for two-parent families with children aged 5–9 in school (youngest child) (only Quebec and Ontario). Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.10:

Estimated effects of the policy on the health and behavior of parents with children aged 1 to 5 not in school (Wave 3 included) (Appendix).

β 3−8 β 3 P β 4 P β 5 F E β 6 F E β 7 F L β 8 F L N
(1998–2009) (1998–99) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09)
Parent health
Mother in −0.010 −0.035 0.010 −0.017 −0.022 −0.006 0.004 50,292
Excellent health (+) (0.019) (0.028) (0.026) (0.029) (0.031) (0.031) (0.033)
Father in 0.000 0.002 −0.011 −0.025 −0.025 0.032 0.023 50,054
Excellent health (+) (0.019) (0.029) (0.026) (0.030) (0.030) (0.030) (0.032)
Mother’s depression 0.153*** 0.092 0.143** 0.138** 0.202*** 0.282*** 0.062 49,217
Score (−) (0.040) (0.056) (0.064) (0.055) (0.071) (0.074) (0.058)
Parent behavior
Family dysfunction 0.047 −0.004 0.104* −0.058 0.046 0.019 0.147** 49,681
Index (−) (0.041) (0.060) (0.053) (0.059) (0.062) (0.065) (0.068)
Positive interaction −0.178*** −0.073 −0.262*** −0.205*** −0.309*** −0.133** −0.105* 34,703
(from 2 years) (+) (0.042) (0.061) (0.057) (0.057) (0.065) (0.061) (0.060)
Hostile parenting 0.139** 0.001 0.154** 0.170** 0.217** 0.228*** 0.089 34,205
(from 2 years) (−) (0.055) (0.066) (0.066) (0.072) (0.086) (0.080) (0.080)
Consistent parenting −0.054 −0.028 −0.160** −0.134** −0.036 0.001 0.034 33,815
(from 2 years) (+) (0.048) (0.065) (0.062) (0.068) (0.075) (0.071) (0.068)
Aversive parenting 0.089* −0.038 0.037 0.076 0.195*** 0.155** 0.126 34,541
(from 2 years) (−) (0.048) (0.066) (0.064) (0.067) (0.073) (0.074) (0.078)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. Wave 3 is included as post-reform period for 3–5 years old and pre-reform period for 0–2 years old. The table also shows the average effect for the full post-treatment period (β 3−8) and the effects by wave ( β 3 P to β 8 F L ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial.Estimates are obtained for two-parent families with children aged 1–5 not in school. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.11:

Estimated effects of the policy on the health and behavior of parents with children aged 5 to 7 in school (Wave 3 included) (Appendix).

β 3−8 β 3 P β 4 P β 7 F E β 8 F E N
(1998–2009) (1998–99) (2000–01) (2006–07) (2008–09)
Parent health
Mother in −0.010 −0.035 0.025 −0.026 −0.031 11,553
Excellent health (+) (0.040) (0.043) (0.050) (0.060) (0.061)
Father in −0.065 0.008 −0.060 −0.100* −0.074 11,497
Excellent health (+) (0.040) (0.044) (0.052) (0.057) (0.057)
Mother’s depression 0.012 −0.033 −0.045 0.031 0.099 11,372
Score (−) (0.086) (0.087) (0.109) (0.123) (0.125)
Parent behavior
Family dysfunction 0.024 0.018 −0.157 0.132 0.187 11,426
Index (−) (0.088) (0.097) (0.109) (0.124) (0.126)
Positive interaction −0.182** −0.060 −0.237** −0.199* −0.149 11,571
(from 2 years) (+) (0.088) (0.103) (0.101) (0.104) (0.119)
Hostile parenting −0.051 −0.149 −0.041 −0.115 0.053 11,407
(from 2 years) (−) (0.123) (0.132) (0.133) (0.144) (0.146)
Consistent parenting 0.123 0.111 0.051 0.210* 0.148 11,300
(from 2 years) (+) (0.083) (0.096) (0.103) (0.110) (0.107)
Aversive parenting −0.198** −0.124 −0.293*** −0.136 −0.162 11,531
(from 2 years) (−) (0.080) (0.082) (0.104) (0.109) (0.104)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. Wave 3 is included as post-reform for 5–7 years old. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F E ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for two-parent families with children aged 5–7 in school (youngest child). Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.12:

Estimated effects of the policy on the health and behavior of parents with children aged 1 to 5 not in school (Trend included) (Appendix).

Variable β 4−8 β 4 P β 5 F E β 6 F E β 7 F L β 8 F L N
(2000–09) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09)
Parent health
Mother in 0.005 0.006 0.016 −0.013 0.011 0.003 40,865
Excellent health (+) (0.024) (0.029) (0.030) (0.031) (0.033) (0.034)
Father in 0.005 −0.024 −0.002 −0.014 0.049 0.017 40,639
Excellent health (+) (0.024) (0.029) (0.031) (0.031) (0.031) (0.033)
Mother’s depression 0.151*** 0.135** 0.098* 0.211*** 0.265*** 0.053 39,889
Score (−) −0.048 (0.067) (0.059) −0.07 (0.077) (0.062)
Parent behavior
Family dysfunction 0.042 0.105* −0.046 −0.017 0.036 0.109 40,336
Index (−) (0.050) (0.059) (0.062) (0.064) (0.069) (0.071)
Positive interaction −0.216*** −0.313*** −0.142** −0.328*** −0.175*** −0.121* 30,124
(from 2 years) (+) (0.048) (0.060) (0.058) (0.064) (0.062) (0.063)
Hostile parenting 0.199*** 0.187*** 0.225*** 0.223** 0.223*** 0.147* 29,654
(from 2 years) (−) (0.064) (0.071) (0.075) (0.087) (0.081) (0.084)
Consistent parenting −0.08 −0.169*** −0.181*** −0.050 −0.004 0.008 29,272
(from 2 years) (+) (0.054) (0.065) (0.068) (0.074) (0.071) (0.071)
Aversive parenting 0.169*** 0.113* 0.117* 0.259*** 0.168** 0.203** 29,982
(from 2 years) (−) (0.055) (0.067) (0.067) (0.073) (0.075) (0.079)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F E ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for two-parent families with children aged 1–5 not in school. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.13:

Estimated effects of the policy on the health and behavior of parents with children aged 5 to 9 in school (youngest child) (Trend included) (Appendix).

Variable Panel A: Children aged 5 to 7 Panel B: Children aged 8 and 9
β 4−8 β 4 P β 7 F E β 8 F E N β 5−8 β 5 P β 7 P N
(2000–09) (2000–01) (2006–07) (2008–09) (2002–07) (2002–03) (2006–07)
Parent health
Mother in 0.001 0.032 −0.017 −0.027 8716 0.041 0.017 0.076 5295
Excellent health (+) (0.045) (0.051) (0.062) (0.062) (0.050) (0.063) (0.051)
Father in −0.081* −0.069 −0.092 −0.086 8666 −0.042 −0.084 0.014 5264
Excellent health (+) (0.045) (0.052) (0.058) (0.058) (0.049) (0.064) (0.047)
Mother’s depression 0.044 −0.034 0.036 0.172 8566 −0.110 −0.187 −0.003 5223
Score (−) (0.104) (0.120) (0.131) (0.133) (0.119) (0.139) (0.122)
Parent behavior
Family dysfunction 0.016 −0.151 0.077 0.201 8622 −0.172 −0.257* −0.058 5221
Index (−) (0.097) (0.112) (0.127) (0.127) (0.109) (0.144) (0.106)
Positive interaction −0.309*** −0.345*** −0.313*** −0.245** 8733 −0.196** −0.191* −0.203** 5304
(from 2 years) (+) (0.090) (0.100) (0.102) (0.118) (0.088) (0.111) (0.099)
Hostile parenting 0.023 0.041 −0.093 0.118 8586 0.002 −0.002 0.008 5215
(from 2 years) (−) (0.135) (0.139) (0.151) (0.157) (0.108) (0.135) (0.114)
Consistent parenting 0.062 −0.004 0.108 0.115 8478 0.142 0.121 0.173 5142
(from 2 years) (+) (0.094) (0.108) (0.116) (0.111) (0.102) (0.124) (0.113)
Aversive parenting −0.146 −0.210** −0.095 −0.101 8702 −0.171* −0.302** 0.009 5296
(from 2 years) (−) (0.090) (0.106) (0.111) (0.108) (0.096) (0.117) (0.104)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F E ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for two-parent families with children aged 5–9 in school (youngest child). Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.14:

Estimated Effects of the Policy on Children 1–5 not in school (one parent families) (Appendix).

Children aged 1 to 5 not school
Variable β 4−8 β 4 P β 5 F E β 6 F E β 7 F L β 8 F L β 78 N
(2000–09) (2000–01) (2002–03) (2004–05) (2006–07) (2008–09) vs β 4−8
Parent health
Mother in −0.003 0.020 0.016 −0.073 0.004 0.015 0.021 6778
Excellent health (+) (0.053) (0.065) (0.076) (0.071) (0.077) (0.082) (0.053)
Mother’s depression 0.239** 0.175 0.166 0.335** 0.277 0.255* 0.042 6700
Score (−) (0.116) (0.139) (0.194) (0.152) (0.179) (0.148) (0.108)
Parent behavior
Family dysfunction 0.138 0.051 0.128 0.106 0.184 0.250 0.127 6423
Index (−) (0.118) (0.148) (0.159) (0.159) (0.161) (0.180) (0.117)
Positive interaction −0.259** −0.275** −0.434*** −0.307 −0.080 −0.205 0.211 5174
(from 2 years) (+) (0.114) (0.136) (0.145) (0.192) (0.148) (0.187) (0.141)
Hostile parenting 0.275** 0.238 0.076 0.348 0.274 0.427** 0.127 5076
(from 2 years) (−) (0.138) (0.155) (0.181) (0.230) (0.216) (0.201) (0.155)
Consistent parenting −0.122 −0.178 −0.008 −0.343* 0.102 −0.119 0.164 5022
(from 2 years) (+) (0.122) (0.145) (0.142) (0.199) (0.159) (0.202) (0.143)
Aversive parenting 0.382*** 0.333** 0.203 0.386** 0.380** 0.599*** 0.175 5139
(from 2 years) (−) (0.122) (0.149) (0.180) (0.187) (0.181) (0.194) (0.137)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F L ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for one-parent families with children aged 1–5 not in school. Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### Table A.15:

Estimated Effects of the Policy on Children 5–7 and 8–9 in school (one-parent families) (youngest child) (Appendix).

Variable Panel A: Children aged 5 to 7 Panel B: Children aged 8 and 9
β 4−8 β 4 P β 7 F E β 8 F E Test β 4= Test β 4 N β 5−7 β 5 P β 7 P Test β 5 Test β 5 N
(2000–09) (2000–01) (2006–07) (2008–09) = β 7 =β 8 = β 7 =β 8=0 (2002–07) (2002–03) (2006–07) =β 7 =β 7=0
Parent health
Mother in 0.056 0.113 −0.039 0.080 0.326 0.435 2383 0.087 0.069 0.103 0.806 0.550 1415
Excellent health (+) (0.089) (0.107) (0.111) (0.104) (0.096) (0.142) (0.094)
Mother’s depression 0.452** 0.484** 0.431 0.428 0.977 0.161 2361 −0.073 −0.129 −0.023 0.700 0.900 1411
Score (−) (0.202) (0.238) (0.276) (0.271) (0.185) (0.280) (0.177)
Parent behavior
Family dysfunction −0.030 0.023 −0.073 −0.060 0.920 0.983 2290 −0.182 0.029 −0.366** 0.079 0.092 1387
Index (−) (0.203) (0.209) (0.280) (0.278) (0.163) (0.210) (0.184)
Positive interaction −0.162 −0.070 −0.148 −0.323* 0.356 0.305 2397 −0.157 −0.224 −0.100 0.602 0.601 1416
(from 2 years) (+) (0.165) (0.195) (0.245) (0.181) (0.159) (0.232) (0.166)
Hostile parenting 0.305* 0.124 0.396 0.485** 0.326 0.216 2348 0.117 0.293 −0.020 0.35 0.632 1390
(from 2 years) (−) (0.183) (0.206) (0.247) (0.236) (0.218) (0.336) (0.216)
Consistent parenting −0.262 −0.303 −0.335 −0.112 0.494 0.440 2316 0.079 0.071 0.087 0.932 0.847 1368
(from 2 years) (+) (0.191) (0.225) (0.245) (0.205) (0.142) (0.186) (0.159)
Aversive parenting 0.173 0.281 0.065 0.135 0.680 0.674 2395 0.058 0.146 −0.016 0.545 0.824 1410
(from 2 years) (−) (0.198) (0.239) (0.245) (0.246) (0.183) (0.267) (0.188)
1. This table displays the estimated policy effects and standard errors (in parentheses). Statistically significant estimates according to the adjusted p-values are presented in bold. These p-values make use of a Simes p-value adjustment procedure to account for testing effects on multiple related outcomes. The table also shows the average effect for the full post-treatment period (β 4−8) and the effects by wave ( β 4 P to β 8 F E ), based on Table 1. We also report a plus or minus sign for each outcome showing the direction the effect must take for the policy to be beneficial. Estimates are obtained for one-parent families with children aged 5–9 in school (youngest child). Bootstrap weights from Statistics Canada are used for inference. Significance is denoted using asterisks: *** is p < 0.01, ** is p < 0.05, and * is p < 0.1.

### References

Baker, M. 2011. “Innis Lecture: Universal Early Childhood Interventions: What is the Evidence Base?” Canadian Journal of Economics 44 (4): 1069–105. https://doi.org/10.1111/j.1540-5982.2011.01668.x.Search in Google Scholar

Baker, M., J. Gruber, and K. Milligan. 2008. “Universal Child Care, Maternal Labor Supply, and Family Well–Being.” Journal of Political Economy 116 (4): 709–45. https://doi.org/10.1086/591908.Search in Google Scholar

Baker, M., J. Gruber, and K. Milligan. 2019. “The Long-Run Impacts of a Universal Child Care Program.” Applied Economic Journal: Economic Policy 3 (11): 1–26. https://doi.org/10.1257/pol.20170603.Search in Google Scholar

Barnett, S. W. 2008. “Preschool Education and its Lasting Effects: Research and Policy Implications.” Technical Report, Education Policy Research Unit.Search in Google Scholar

Berger, M. C., and D. A. Black. 1992. “Child Care Subsidies, Quality of Care, and the Labor Supply of Low-Income, Single Mothers.” The Review of Economics and Statistics, 635–42. https://doi.org/10.2307/2109377.Search in Google Scholar

Brodeur, A., and M. Connolly. 2013. “Do higher Child Care Subsidies Improve Parental Well-Being? Evidence from Quebec’s Family Policies.” Journal of Economic Behavior & Organization 93: 1–16. https://doi.org/10.1016/j.jebo.2013.07.001.Search in Google Scholar

Chatterji, P., S. Markowitz, and J. Brooks-Gunn. 2013. “Effects of Early Maternal Employment on Maternal Health and Well-Being.” Journal of Population Economics 26 (1): 285–301. https://doi.org/10.1007/s00148-012-0437-5.Search in Google Scholar

Connolly, M., and C. Haeck. 2015. “Are Child Care Subsidies Good for Parental Well-Being? Empiral Evidence from Three Countries.” CESifo Dice Report 13 (1): 9–15.Search in Google Scholar

Conseil du Trésor Québec 2012. Budget de dépenses 2012-2013, volume II: plans annuels de gestion des dépenses des ministères et organismes, pour l’année financière se terminant le 31 mars 2013. Technical Report.Search in Google Scholar

Currie, J., and D. Almond. 2011. “Human Capital Development before Age Five.” In Handbook of Labor Economics, Vol. 4, edited by D. Card, and O. Ashenfelter, 1315–486. Elsevier.Search in Google Scholar

Danziger, S. K., E. O. Ananat, and K. G. Browning. 2004. “Childcare Subsidies and the Transition from Welfare to Work.” Family Relations 53 (2): 219–28.Search in Google Scholar

Davis, E. E., and M. Jefferys. 2007. “Child Care Subsidies, Low-Wage Work and Economic Development.” International Journal of Economic Development 9 (3): 122.Search in Google Scholar

Ha, Y. 2009. “Stability of Child-Care Subsidy Use and Earnings of Low-Income Families.” Social Service Review 83 (4): 495–523. https://doi.org/10.1086/650352.Search in Google Scholar

Haeck, C., L. Lebihan, and P. Merrigan. 2018. “Universal Child Care and Long-Term Effects on Child Well-Being: Evidence from Canada.” Journal of Human Capital 12 (1): 38–98. https://doi.org/10.1086/696702.Search in Google Scholar

Haeck, C., P. Lefebvre, and P. Merrigan. 2015. “Canadian Evidence on Ten Years of Universal Preschool Policies: The Good and the Bad.” Labour Economics 36: 137–57. https://doi.org/10.1016/j.labeco.2015.05.002.Search in Google Scholar

He, A., and N. Sayour. 2020. “After-School Care, Child Care Arrangements, and Child Development.” Journal of Human Capital 14 (4): 617–52. https://doi.org/10.1086/711950.Search in Google Scholar

Herbst, C., and E. Tekin. 2014. “Child Care Subsidies, Maternal Health and Child-Parent Interactions: Evidence from Three Nationally Representative Datasets.” Health Economics 23 (8): 894–916. https://doi.org/10.1002/hec.2964.Search in Google Scholar

Japel, C., R. E. Tremblay, and S. Côté. 2005. “Quality Counts!.” Choice 11: 5.Search in Google Scholar

Kottelenberg, M. J., and S. F. Lehrer. 2013. “New Evidence on the Impacts of Access to and Attending Universal Child-Care in Canada.” Canadian Public Policy 39 (2): 263–86. https://doi.org/10.3138/cpp.39.2.263.Search in Google Scholar

Kottelenberg, M. J., and S. F. Lehrer. 2014. “Do the Perils of Universal Childcare Depend on the Child’s Age?” CESifo Economic Studies 60 (2): 338–65. https://doi.org/10.1093/cesifo/ifu006.Search in Google Scholar

Kottelenberg, M. J., and S. F. Lehrer. 2017. “Targeted or Universal Coverage? Assessing Heterogeneity in the Effects of Universal Child Care.” Journal of Labor Economics 35 (3): 609–53. https://doi.org/10.1086/690652.Search in Google Scholar

Kottelenberg, M. J., and S. F. Lehrer. 2018. “Does Quebec’s Subsidized Child Care Policy Give Boys and Girls an Equal Start?” Canadian Journal of Economics 51 (2): 627–59. https://doi.org/10.1111/caje.12333.Search in Google Scholar

Kröll, A., and R. Borck. 2013. “The Influence of Child Care on Maternal Health and Mother-Child Interaction.” CESifo Working Paper Series. Technical Report 4289.Search in Google Scholar

Lefebvre, P., and P. Merrigan. 2008. “Child-care Policy and the Labor Supply of Mothers with Young Children: A Natural Experiment from Canada.” Journal of Labor Economics 26 (3): 519–48. https://doi.org/10.1086/587760.Search in Google Scholar

Lefebvre, P., P. Merrigan, and M. Verstraete. 2009. “Dynamic Labour Supply Effects of Childcare Subsidies: Evidence from a Canadian Natural Experiment on Low-Fee Universal Child Care.” Labour Economics 16 (5): 490–502. https://doi.org/10.1016/j.labeco.2009.03.003.Search in Google Scholar

Lundberg, S. J., R. A. Pollak, and J. W. Terence. 1997. “Do husbands and Wives Pool Their Resources? Evidence from the United Kingdom Child Benefit.” Journal of Human Resources: 463–80. https://doi.org/10.2307/146179.Search in Google Scholar

Martin, A., J. Brooks-Gunn, P. Klebanov, S. L. Buka, and M. C. McCormick. 2008. “Long-term Maternal Effects of Early Childhood Intervention: Findings from the Infant Health and Development Program (IHDP).” Journal of Applied Developmental Psychology 29 (2): 101–17. https://doi.org/10.1016/j.appdev.2007.12.007.Search in Google Scholar

Milligan, K., and M. Stabile. 2007. “The Integration of Child Tax Credits and Welfare: Evidence from the Canadian National Child Benefit Program.” Journal of Public Economics 91 (1): 305–26. https://doi.org/10.1016/j.jpubeco.2006.05.008.Search in Google Scholar

Ministère de la Famille et des Ainés (MFA) Québec 2013. “Situation des centres de la petite enfance, des garderies et la garde en milieu familial au Québec.” Technical Report.Search in Google Scholar

Ministry of Finance of Québec 2004. “Plan Budgétaire 2004-2005.” Technical Report.Search in Google Scholar

NICHD 1999. “Child Care and Mother-Child Interaction in the First 3 Years of Life.” Developmental Psychology 35 (6): 1399–413.Search in Google Scholar

NICHD 2003. “Early Child Care and Mother–Child Interaction from 36 Months through First Grade.” Infant Behavior and Development 26 (3): 345–70.Search in Google Scholar

Okuyama, K., and R. Weber. 2001. Parents Receiving Child Care Subsidies: Where Do They Work? Oregon Child Care Research Partnership.Search in Google Scholar

Roggman, L. A., T. SondraMoe, A. D. Hart, and L. F. Forthun. 1994. “Family Leisure and Social Support: Relations with Parenting Stress and Psychological Well-Being in Head Start Parents.” Early Childhood Research Quarterly 9 (3-4): 463–80. https://doi.org/10.1016/0885-2006(94)90020-5.Search in Google Scholar

Shaffer, J. P. 1995. “Multiple Hypothesis Testing.” Annual Review of Psychology 46 (1): 561–84. https://doi.org/10.1146/annurev.ps.46.020195.003021.Search in Google Scholar

Simes, R. J. 1986. “An Improved Bonferroni Procedure for Multiple Tests of Significance.” Biometrika 73 (3): 751–4. https://doi.org/10.1093/biomet/73.3.751.Search in Google Scholar

Statistics, C. 2007. “Microdata User Guide National Longitudinal Survey of Children and Youth Cycle 7- September 2006 to July 2007.” Technical Report.Search in Google Scholar

Ward-Batts, J. 2008. “Out of the Wallet and into the Purse Using Micro Data to Test Income Pooling.” Journal of Human Resources 43 (2): 325–51. https://doi.org/10.1353/jhr.2008.0005.Search in Google Scholar