Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. =Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. = # Co-Parenting and Careers after Divorce Authors: Katarina Kuske Discussant: None Video: https://www.youtube.com/watch?v=1kb3a99sA-E&t=7114s ## Talk (01:58:34 – 02:44:25) [01:58:34] everyone. We are next going to have two 45minute papers. Um, same rules as before, 10 minutes, no questions, and then we can bring in questions. Um, and we're very happy to have Katina here [01:58:48] talking about co co-parenting and careers after divorce. Please take it away. Thanks a lot to the organizers for including the paper. Um so this is going to be about different parenting arrangements after divorce and how they [01:59:00] impact parents labor market outcomes. So in particular I'm interested in what happens when parents parent jointly. So I have joint physical custody rather than mothers having sole physical custody. And uh it was a nice uh paper [01:59:15] to have beforehand like Emily's great presentation because this is also going to be looking at father's involvement. [01:59:21] Yes or no? But I'm going to be looking at on average good dads. So it's a a bit of a happier picture in that sense. [01:59:30] We know that divorce negatively affects children's outcomes across a wide range of dimensions. So from emotional well-being to behavioral uh problems to education and also to later in life [01:59:43] outcomes such as teenage pregnancy and criminality. [01:59:47] And one of the pathways through which this happens is by spending less times with their fathers after divorce. And the reason for this is that on average after divorce children primarily live with their mothers and uh see their fathers less often. So a frequent [02:00:02] arrangement is for example to see the father every other weekend and perhaps for two weeks of holidays a year and recognizing this there has been a push in policy towards encouraging greater involvement by both parents and [02:00:16] the margin on which I'm going to be looking at this is joint physical custody which I also call co-parenting and here I would just like to to fix ideas physical custody is different from legal custody so legal custody refers to [02:00:30] the right of parents to make important decisions about the child's life and these involve which school to go to. So, educational decisions, medical care, religion and so on. Instead, physical custody simply is with whom the child [02:00:44] lives. So, this affects a lot the day-to-day arrangements of parents. And there is evidence that suggests that joint physical custody can benefit kids, especially on the education margin. What [02:00:58] is less clear though is what it does to parents and in particular there's the risk that co-parenting may have unintended consequences for parents in particular for the labor market outcomes. [02:01:11] So what I'll be focusing on is really the parent side of of outcomes and how greater parental involvement as proxied by co-parenting affects these. [02:01:22] I will start the the presentation with a conceptual framework in which I contrast the constraints that parents face under two different types of regimes. So there is one regime that's the sort of more traditional one sole physical custody [02:01:36] which I assume is going to be by mothers throughout because that's the case in more than 90% of of the data empirically and I contrast this to having joint physical custody. [02:01:48] So there are two key constraints. The first one is the time constraint. If parents share child care more equally then the both parents uh would see a difference in their time constraint relative to sole custody by mothers. So [02:02:02] mothers if they generally have their kid would see a relaxation in their time constraint. So one could expect to either see them work more hours in the market or consume more leisure or both. [02:02:13] And of course you would expect the opposite to happen for fathers who with greater childare involvement would experience a tightening in their time constraints. [02:02:21] But co-arenting also introduces a new constraint and that's a location constraint. The the reason for this is um that for for parents for this co-arenting to work out parents have to live in relatively close proximity to [02:02:35] each other. Uh so this quite explicitly restricts their geographical mobility and in particular if you think about sort of how couples choose locations it may often be for the career of the male partner and then co-parenting can in [02:02:50] effect turn divorced mothers into tight stairs if they're not in a labor market that was originally also chosen for their own careers but rather only for those of their at this point ex-husband and I'm I'm going to present this in a [02:03:05] bit more detail and derive some predictions what we can expect for mothers and fathers wages and hours um when contrasting these two custody regimes. [02:03:14] I bring this to the data and test it empirically using a reform in the Netherlands that increased the share of parents who co-parent after divorce. I believe this is quite a relevant setting. So about one in three divorces in the Netherlands involve children [02:03:29] under the age of 10 or age 10 or younger. And uh the reason why this is sort of the most relevant group uh in my setting is that teenage children they really dislike being co-parented because I haven't said it yet but generally [02:03:42] co-arenting in involves sort of having one week with a mother one week with a father. So their frequent moves between different homes and teenagers want to have their room and that's that's it. Uh so um these younger children they they [02:03:56] uh are a bit more willing to do that and crucially older children are also asked for their preferences. So from uh 12 years on in the Netherlands the preference of the child is quite strongly taken into account. I then combine this reform with administrative [02:04:10] data and also supplemented with survey data and then run an event study difference and difference is a design in which I compare parents who have young children and divorced before the reform so have a lower propensity of [02:04:23] co-parenting to parents who divorced after the reform and also have young children. So the only difference between these two groups of parents should really be their exposure to the reform and thereby their propensity to co-arent. [02:04:37] Now what do I find? The main finding really is that mothers who are exposed to the reform they see persistent decreases in their hourly wages. So I'm going to be talking about hourly wages when I say wages otherwise I refer to [02:04:51] earnings. And in an intention to treat setting these are relatively small about 0.8%. [02:04:57] 8%. This is because the share of compliance is not huge. Once you scale that up to a local average treatment effect, you find a much more sizable relative decrease in wages for the treated mothers by about 10%. [02:05:11] Important to say is it's not an absolute decrease in wages for mothers who were treated. Instead, it's wage increase for those mothers who um have a lower propensity to go parents. So, who divorced divorced prior to the reform. [02:05:24] um and you see wage stagnation for the mothers in the treated group. So it's simp it's not that these mothers saw an absolute wage decline rather they missed out on wage gains after divorce. [02:05:34] The um the groups demographically that are most uh susceptible to this are mothers who have the highest potential for wage growth. So these include mothers who are younger at the time of [02:05:46] divorce. So um below median age of 38, it's of course those who also have a steeper life cycle profile still and would expect to see more wage growth. [02:05:55] It's also those mothers who are more highly educated. So who could have high a higher benefits from having a very specific job match. Think of a marine biologist for example. Um and it's also those mothers who cut back most on their [02:06:09] own careers while they were married. So the effect is strongest when I condition on the sample of secondary earners who are defined as mothers of or fathers earning 40% or less of household income. [02:06:23] I do not find an increase in mother's hours worked and this I attribute to the setting that is perhaps quite specific. [02:06:30] So in the Netherlands already at the time of the reform which happened in 2008 uh child care was widely available. [02:06:37] It was of high quality and it is a setting in which the vast majority of mothers work part-time. So 87% of the mothers in my sample with kids under the age of 10, they're part-time working. So it doesn't seem to be that childare is the hugest constraint. This of course [02:06:51] you may expect to look rather different in a setting such as the United States where part-time work is not as widely available. Child care neither is very costly. Um, so this I would just like to say it's the Netherlands are rather specific setting here. [02:07:05] As a lot of family policy, this does not seem to have meaningfully affected men's outcome, at least as it refers to wages or hours worked. [02:07:16] The mechanism that I suggest that is driving this is this geographical mobility restriction arising from co-parenting because that effectively anchors parents in the marital location most likely. And while this effect uh [02:07:29] this this constraint is there for both parents. So both mothers and fathers who co-arent they find it harder to move away. It's binding in terms of labor market outcomes only for mothers because of the way in which uh couples made [02:07:42] decisions about labor supply and sort of market work non-market work while they were married. So this is not a a gendered reform per se but rather the effect shows up as a gendered story because of household specialization. [02:07:57] Now I contribute to I would say two broad areas in the literature. The first is this vast one on uh post- divorce legal accommodations and labor supply and a lot of that of course has looked sort of at the introduction of no fault [02:08:11] divorce and how that affects labor supply of still married couples. Um there are two papers that I would like to highlight that are the closest. Um these are I don't know if you can see here those two uh by Adams, Allesandra [02:08:24] Buena and co-authors. um and also Stella Kanessa and Susan Niknami who look at Sweden. So the first paper looks at the United States and overwhelmingly looks at the introduction or becoming available of the option of having joint [02:08:38] legal custody and uh the second paper looks at Sweden uh and I should say the first paper finds also interestingly a mobility effect but rather on men's side and over longer distance mobility so interstate migration um whereas the second paper looks at a wide out range [02:08:53] of outcomes for both parents and children in Sweden using a judge IV design for contested court cases. So just to preface these also looking at legal custody but these uh complier groups in that setting are going to be [02:09:07] very different. So these are more negatively selected as you will see my parents or the parents who are choosing to co-arent they are overwhelmingly positively selected. So that's why the good dads also the good moms it's uh it's those who can can choose it and can [02:09:20] afford it. And instead I isolate this physical custody component from the legal one. um because at the time of the reform divorcing parents automatically both had joint legal custody over the children. So that was not up to up to [02:09:35] debate. Uh so it's really purely about this physical custody. And then I show evidence on wages, hours and also workplace mobility. And the second part of the literature that I think it relates to nicely is the location [02:09:49] decisions and the labor market outcomes of dual earner couples. And there's this like very cool literature on co- collocation constraints and how um frequently mothers or women in general [02:10:03] in couples are more likely to be either tight movers or tight stayers and how that negatively impacts their earnings. [02:10:10] What I can show here is that there's a collocation friction that persists even after relationship dissolution if the parents parent their ch children or child jointly. [02:10:22] Now let me start by showing you sort of a theory part of of the paper in this picture. Uh this is to fix ideas and to think about what we would expect to see in terms of effects uh from the [02:10:36] empirical results. I'm going to be quite stereotypical in my language here and that's because I'm going to be talking about statistical averages. So to the left of this uh dashed line you have the [02:10:48] intact family the the couple with one or or several children aged 10 or younger who make a decision about where to live and that location decision is taken jointly and I assume that it maximizes [02:11:01] household welfare frequently because these like over 85% of couples uh have mothers who are working part-time. It makes sense from a household perspective to choose a location that maximizes the [02:11:14] father's hourly wage and thereby also the earnings. Um but this location importantly must not necessarily reflect good opportunities for the mother's wage. Yes. So are you able to see data [02:11:27] on like the um child support regime in either of these two situations and also like even realized income because I can kind of imagine two different cases. So one, I know when there's joint physical custody, sometimes there's a lower child [02:11:41] support obligation from the higher earner because the child is also spending time both places. But also, I think there's a lot of research showing that like there's less sherking in child support obligations when the dad is more involved in the kid's life because they see it instead of giving money to the [02:11:56] mom who they don't like, it's giving money to the kids who they they do like. [02:11:59] Um, so I just wonder about the income effects and it's sort of not on here as a mechanism. [02:12:05] >> No, you're right. It's not on here. And there is um so the idea is that if both parents co-parent so roughly share child care equally and each parent pays for the child on their own time. So generally there's no child child support payment in co-parenting families unless [02:12:20] the differences are huge between the two parents. There's no income effect because well in this picture at least uh because the child support payments they are not judge awarded. There's a table by a Dutch social ministry and it's really dependent on the child's age and [02:12:34] the part uh partner's income who has to pay. Um and the amounts are not huge. So on average for children that I look at is about €250 per month. Um which is really not that much of of money if you [02:12:47] think it's probably enough to cover half of the expenses of the child but not going to be a huge income transfer either way. [02:12:54] >> What about alimony? >> Alimony, yes, good point. Alimony is not affected by co-parenting. So alimony is purely for the uh former spouse's um originally was to keep up the living standard and that's not affected. [02:13:06] Alimony is not very prevalent in the Netherlands. So um in the survey data I can see it. It's less than 20% of of mothers who receive alimony virtually basically no fathers. It's very very rare. Um h but that is not [02:13:20] affected by co-arenting status at all. >> Yes. very much very related to that point. Um if could this be actually a positive result if suddenly the dad is contributing more to you know he provides just more meals and that's less [02:13:34] expense for me and he provides more sports events or whatever and therefore I don't actually have to work as much anymore. I can pull back and spend more time with my kid which you know I love my job but if I had a crappy job that might be preferable. No it's true. Um, however, then I would still expect the [02:13:49] person to rather work fewer hours at a higher wage rather than working the same amount of hours at a lower wage. But you're right. So, this could still in some way there could still be sort of smaller income effects or benefits. And I I can show you some suggestive [02:14:02] evidence that these co-parenting mothers take more free time, they sleep more, they feel less stressed. So, there seem to be some perhaps beneficial margins. [02:14:11] Um, but it doesn't really show up in the labor market data. [02:14:15] All right. Um now these par these families are in a location and now they're hit by a divorce truck rather that one of them chooses to divorce and they can choose between the two regimes when it comes to custody for their children. So in the top the pink part [02:14:30] you see this is sole custody which I assume to be by mothers and under that both parents are in theory free to move. [02:14:36] Uh and you would expect that if mothers can move away they should perhaps also at least some of them move towards locations where they get paid more. So you would expect to see wage growth over time. However, um for fathers, given this perhaps relatively strong [02:14:51] assumption that they're already pretty well matched in the location of marriage, you would not expect to see an awful lot of changes to their wages. [02:14:59] Now, when parents are location constrainted in the green part at the bottom, you would expect the opposite in relative terms. Of course, that mother's wages in relative terms decrease. And again, no change to fathers. In terms of mother's hours worked, I think it's a bit more nuanced because you have this [02:15:14] relaxation of the time constraint that mothers could translate into higher hours worked. But at the same time, if you're getting paid less than you could earn, you also have this negative substitution effect of low wages and hours worked. So even extant anti- [02:15:28] theoretically, it's not quite clear which way you would expect the the hours to go. [02:15:35] Now, a little bit about Oh, sorry. Yes. wondered if you can look at this by the [02:15:52] age of the child at the time of the custody arrangement being made. I imagine it's U-shaped where if the child's very young or quite grown up, then the mother is more willing to relocate. But if the child is in the [02:16:05] middle and might have more engagement with the dad and more conscious attachment, they might be more likely. [02:16:11] So I have some results by sort of cut off date. I choose 10 based on my reading of the literature when it's sort of a a good age to stop assuming that. [02:16:20] Um but if I look at the treatment effects, they're actually strongest for sort of early primary school age children. So when the child is aged between five and eight at the time of divorce. [02:16:31] Yeah, I was thinking if they're like age one or age 18 that it's less likely to >> both are very unlikely to have for age one because of breastfeeding at age 18 because the child really doesn't want to [02:16:43] have to move between or 17 at least I think it's the the latest possible. [02:16:48] >> Yeah. But yes. Yes. >> Since a lot of this depends on the location choice, can you also look at the job market and you know how suitable the local market is in that area for the mother's education [02:17:02] and her occupation? That would be one way to verify. [02:17:06] >> So I tried sadly the data does not include occupation. So I can't do a very nice outside option index. I try to look at sort of concentration in the labor market by industry of the firms and I do find that it's really driven by those uh [02:17:20] mothers who during marriage are located in the more concentrated industry. So are fewer outside options locally and would have to move if they wanted to to move to a better employer. [02:17:32] All right. Now I'd like to tell you a bit about the reform. So importantly, even before the reform, it was uh possible or it was the default to have joint legal custody for divorcing parents. This was not the case until [02:17:46] 2023 for cohabiting parents, which is why I restrict this analysis to divorcing parents. Otherwise, you would probably expect some to see something very similar between cohabiting and divorcing parents. Um also because cohabitation is is much has been on the [02:18:00] rise a lot for the past 20 years in the Netherlands. The key part of the reform really um so I do a literal translation from from Dutch. It's called law to promote continued parenthood and careful divorce. So the title already says that [02:18:13] really the key focus and and aim of this reform was to minimize the disruption to the child's life from moving from a intact nuclear family to living with divorced parents. [02:18:25] And the key part of the legal text is that now a child has the right to equal care and upbringing by both parents. [02:18:31] This of course is a bit vague. Um how it was implemented in practice is in the following way. So it was not a legal default and I think that's very important. Um it was rather that courts were instructed to uh prioritize this [02:18:46] child's right, this newly established right and to encourage co-parenting after divorce. [02:18:52] H it more or less just made co-arenting the legal starting point for talks between the parents. It really didn't. [02:18:59] Um, parents could easily opt out if they could argue in front of the judge, this may not be be the right thing for us. If one of the parents is working a 60-hour job per week, this may not be be feasible in practice. Uh, and also these [02:19:11] very negative cases where there's for good reason a separation of the father from that family. Um, in case of violence or or otherwise, this would also not be the case. [02:19:22] And there's a second component through which this possibility of co-parenting really was driven home to parents especially to fathers and that was because um now after the reform at the time of filing for divorce they had to [02:19:35] submit a parenting plan and um in this parenting plan they of course had to include whether they wanted to have co-arenting so joint physical custody or sole custody by one of the other parents and then they could include a number of [02:19:49] other um sort of teachers how they want to parent their child and this was very customable. [02:19:56] They so some of them uh can also specify a maximum mutual living distance between parents. This can be either in in kilometers distance in commuting time or whatever mode of transportation. uh but this of course makes it very salient [02:20:09] that there is a re uh a restriction in the unilateral relocation opportunities and um yeah just to say the parents who choose into co-aring or open to co-arenting they are very positively [02:20:23] selected so both mothers and fathers are more highly educated they they are more likely to be employed they work more hours and importantly they also have less of a distance a gap in hours between them and they share child care [02:20:36] more equally pre- divorce. Uh they also have less conservative gender norms. So really these are what you can think of as like good good moms and good dads. H what's important though is that um the it doesn't change dramatically the type [02:20:50] of parent that opts into co-aring pre and post divorce uh reform they become a tiny bit less positively selected the compliers. So those who originally didn't uh choose co-arenting but then post reform this increased share that chose co-arenting they're still [02:21:05] positively selected but a little bit less so than the sort of always takers. [02:21:11] And now the data that I'm using come from statistics Netherlands. So the Dutch administrative data the main data sets are fairly standard administrative data sets sort of household composition that includes the age, education, [02:21:25] gender, number and age of children and then of course the cohabitation and marital status of parents and the IDs of each so that I can link them over time. [02:21:34] I combine that with the employment data that have monthly observations including hours uh earnings the the sector of of firms and location of the employer. So they don't don't include occupation. I mentioned that. But one nice thing is [02:21:49] that they do include hours and I think in the Dutch context that's quite relevant because the hours distribution especially of mothers is pretty smooth. [02:21:56] So there's no one part-time spike, one full-time spike. There's a lot of freedom and flexibility in how individuals choose their hours worked. [02:22:03] So, it matters a lot to be able to to control for that. Yes. [02:22:10] >> Sorry if you've already uh mentioned this and maybe I missed it, but so you talked about uh selection into uh co-arenting. Uh but what about selection into divorcing? Because if now all of a sudden I might have to co parent or [02:22:24] that's an option maybe people who decide to divorce. So I'm just trying to understand what is the counterfactual. [02:22:30] >> This was this was a key concern and it does not seem to be the case. So I can show that parents um I get to that in a bit but who divorce pre and post reform. [02:22:38] They don't differ on observables. Uh there's no big jump in in the number of divorces. Then there's also an official review by the Dutch Ministry of Justice that checked if this reform imposed any additional barriers to divorce that would lead uh couples that would have [02:22:53] wanted to divorce prior to not be able to do so after. [02:22:58] Um I I have the location of the employer at the four-digit postcode level and this is what I'm going to use as my measure for mobility later on. One thing that's important to say here is that I have the location of the firm of the [02:23:11] head office. So this means that I don't have the plant level and for the mobility analysis I'm going to restrict the sample to single establishment firms to a to be able to pinpoint exactly uh where where uh firms are located. Um [02:23:25] I'll show you before though that the effect does not differ by single establishment firms or not. [02:23:31] In this admin data there's no information on who exactly co-arents or not. Uh this is why I complement the uh the admin data with the new families in the Netherlands survey. It's uh a few thousand individuals. It's a sociologist [02:23:46] from Utret University who has collected uh there's a very nice and detailed information about post-abos parenting arrangements a lot of detail um and importantly for me co-arenting status and also some measure of sort of distance from ex partner that is [02:24:00] different from the one that I have in the admin data. Yes. Um relatedly, [clears throat] do you have any information about time use in that my sense at least for the US is that joint custody is very far from a 50/50 sharing [02:24:13] of the actual time with the child. So some of the labor supply effects will be constrained by the actual division of >> so in the in this survey it's very nice because they ask for each uh for month period which day of the month the child [02:24:27] was with which parent. Um, and so I treat anybody as co-arenting was between 40 and 60% of the time. So close to 50/50, but it doesn't have to be exactly 50/50. I could do some sort of like [02:24:40] extending that to 30 70 or not. Um, but yeah, it it's at least by day. So that uh that is there. [02:24:47] Now I'm pretty light on the sample selection. I choose any individual that's aged between 18 and 60 at the time of divorce and has a child aged 10 or younger. So this means that the age of 60 cutoff is not binding for for [02:25:00] mothers but it is for fathers. And then I also choose the sample to be individuals divorcing between uh sort of in 18 months uh cutoffs on each side of the reform date. So those who are [02:25:13] divorcing between September 2007 and uh uh August 2010. So the first group is my group of control divorcees and the the darker one the ones who are treated so who have the greater chance of [02:25:26] co-arenting. Now I'm going to show you two pieces of evidence that come from the survey data and this is sort of my first stage. It shows that the reform indeed had an effect but it was relatively mild and uh [02:25:41] that I attribute of course to the fact that the reform by no means made co-arenting mandatory and it also helps explain why I don't find different selection into divorce or a change in the number of divorces uh after after the reform because it was just basically [02:25:56] a salience thing. So you see that um the share of divorce parents who co-parent after divorce increased by about 8 percentage points. So it was even after far from becoming the the main uh [02:26:09] parenting arrangement, it's still mostly sole physical custody by mothers. [02:26:14] Now and the second piece of evidence that goes towards the mechanism side is the distance between parents who are co-parenting or not. And I think this is very nice because the survey has the exact distance from ex partners in minutes. So these are self-reported [02:26:28] minutes in whatever means of transportation is relevant to the parents. It can be cycling, it's the Netherlands, can be driving if these parents live in more rural areas. But what you see is that if they're not co-arenting, there's quite a wide range of distance between these parents. Um, [02:26:43] however, if you look at the subset of parents who co-arent, it's not only that the median distance goes down. It's the whole distribution that's much more compressed. So, these parents really live very close to each other. [02:26:56] Now, let me talk to you a bit about the design. It's a fairly straightforward uh event study differences and difference differences design and I have an uh intention to treat framework. [02:27:07] So I'm going to be so my coefficient of interest is going to be this gamma j which compares the difference in post- divorce outcomes of the treated code. So the post-preform cohort relative to the [02:27:20] different uh the post- divorce outcomes of the control group. [02:27:25] And I'm going to be controlling for individual fixed effects of course and in addition to just normal time fix effects as in month fix effects. I think in this setting uh given that the reform happened very close to the great recession it's important to control for [02:27:39] a bit more. So I don't only control for month fix effects in my main specification but rather month time times labor market region and sometimes also uh extending this to uh month times [02:27:51] region time sector fixed effects. Uh standard errors are clustered at the individual level. [02:27:58] Now um a bit to to one of the questions earlier there does not seem to be any selection into treatment. I can show you that the characteristics really of parents who divorce pre and postreform are very very similar. Here I highlighted those. So I'm only showing [02:28:12] you a subset and including those who are not balanced. And those who are not balanced are really the ages because age at divorce is increasing in line with age at marriage increasing and the nominal labor market outcomes such as ages earnings and wages which we would [02:28:26] hope to increase uh to see increase over time. And also if I look at the ones that are not balanced, I can see that really there's no break around the time of the reform, but rather there's really uh it's a it's a smooth upward trend. [02:28:40] There doesn't also seem to be strategic timing in the ball so that people bunch right before after the the reform date. [02:28:47] You'll see in my event studies that parallel trends do seem to be supported and there are no other time bearing confounders in terms of family policy. [02:28:56] But of course I mentioned already at no twice it's that there is the great recession. So I do quite a bit throughout the paper to to address this. [02:29:02] Um but perhaps just one thing we all know that the great recession was a key session. So that men on average were more negatively affected than than women. Um but of course that is not uh uh the the main reason why we can't see anything. But I think the main way to address this really is these uh [02:29:16] monthtimes region fix effects which net out uh all the perhaps local labor market issues that might be happening at the same time. [02:29:25] Now what do I find this uh main result is that mothers who are treated by this reform relative to those who were not see a persistent decline in their hourly wages. So on the x-axis you see six month periods around the time of divorce [02:29:39] and I normalized to the six months uh prior to divorce. On the y-axis you have log wages. This again is the intention to treat setting. So the effects do seem relatively small and you see that this um negative wage effect opens up [02:29:54] gradually over time and this is consistent with sort of not moving um towards these better jobs. [02:30:02] Um if I split this into treated and control groups, you really see it much more clearly. So the light color is those mothers who were in the control group who see earn um wage increases and earnings look pretty much the same uh [02:30:17] post divorce. Whereas this darker group, the treated mothers, they have flat wage profiles. They do not see this wage growth. [02:30:25] Now I'm only going to show you very little about fathers. One of them is uh this wage graph. And you see that pretty much nothing happens to father's wages in either direction. [02:30:36] Now I already mentioned that children who are teenagers, so who are um 11, 12 to 17, they really don't like being co-parented. So I'm going to be using that those as a placebo group to uh [02:30:50] further reinforce that the effects that I do observe are driven by parents who co-arent and not by something else that may be happening at the same time. Um we know there's a literature on graduating into a recession. It could be that divorcing in a recession per se is bad. [02:31:04] But it seems that really so on the left you see these younger children that are just for the different fixed effect specification the effect is very stable. [02:31:12] When you compare them to older children you find really null results for older children if anything minimally positive uh point estimates. [02:31:23] Now I wanted to show you this um because I'm going to later be looking at the single establishment firms only. So um I do a number of robustness here as well looking for example at a donut hole [02:31:36] design that uh excludes the immediate months around the reform date. Facts are similar. I can interact uh linearly a trend with the um unemployment rate in the region at the time of divorce. it doesn't seem to be driving the effect. [02:31:50] But importantly, single establishment firms, even though the sample is much smaller, it does not seem to be affecting differentially whether um mothers who are in single establishment uh firms um whether they see this uh negative [02:32:04] wage effect or not. Now to the hour side, if you look at ours again, we have these six months periods around the balls on the x-axis. there doesn't seem to be anything after a year and a half to two [02:32:19] years. However, you see this immediate dip which is quite I I found this very puzzling at first um that mothers who now spend less time on childare also work a little bit less. Um this is entirely driven by more overtime work in [02:32:33] the control group. So mothers who move to these new jobs perhaps want to prove that despite having a young child and being a divorced mom, they're a good worker and put in more overtime. The overtime vanishes though after a year or [02:32:46] so. So really this entire bump in the control group is driven by initially more overtime. [02:32:52] And again for fathers you see very little if if you look very closely you see kind of a decrease in father's hours worked but it's very slight um and uh and not not super large. Part of the reason Oh yes. [02:33:06] >> Yeah. I do find it very plausible that they're trying to uh prove something to their employer because they're a newly divorced mother. Um it just seems probably their income fall is greater because the father might be putting less [02:33:21] effort and she needs also more child care hours now that there is no father that's involved. [02:33:30] It's so the father should still be putting of course um it may not be through the divorce is relatively fast in the Netherlands. So if there's a child support settlement it should come pretty soon about three months. [02:33:43] >> I'm talking about your your single your control group. [02:33:46] >> The control group. Yes. So you think it's it's other reasons that why they work more overtime >> work more because they're trying to prove something to their employer. It just >> No, this is >> okay. Thank you. I'll think about it [02:33:58] more. To me they seem like a for >> uh the hours they're administrative. [02:34:06] >> Yes, administrative. >> Is there overtime in pay? [02:34:12] >> Yes, there is overtime in pay. >> Okay. [02:34:14] >> So this is uh the total >> and overtime is after 35. [02:34:19] >> Overtime is as per the contract. So if somebody has a 20-hour contract, overtime is going to be sooner. [02:34:26] Yeah, I forgot it is the Netherlands at >> 20 hours. That's right. [02:34:30] >> But they're separately reported. So there's base hours and overtime hours that you can clearly distinguish. [02:34:35] >> Okay. >> Can I ask you a last question? Maybe not relevant but so can you can we think about what part of the so of the growth in earnings for for those who have saus [02:34:50] comes from because I I think initially I thought it was a story of tide stays tied mover right that happened way before divorce right but for those do do you see that those who don't have the [02:35:02] joint custody are more likely maybe to take a job that is further away they're are They is that the mechanism that they are reoptimizing or maybe based on personal experience [02:35:15] the one who took coarent then they have this huge burden of coordinating everything and blah blah bl >> I would love to be able to do that but in order in the survey data I see the co-arenting status in the admin data I [02:35:29] see where they move so I cannot combine these two um there is a big spike in mobility rates in general after divorce so parents with young kids they move really across mun municipalities across regions but it's the one time when they [02:35:42] have this chance to reoptimize or this and and really they go up by about 50%. [02:35:48] Uh but of course I can't I can't link that on the individual level. [02:35:53] All right a little bit of heterogeneity here when I split the samples u by different margins that I find relevant um I find that mothers who are below median aged divorce of 38 in the sample have larger effects than those who are [02:36:07] older. Similarly, those who were secondary earners during marriage and I should say here if I look at secondary earners who are fathers the effect is pretty similar in magnitude if not slightly larger in absolute terms but [02:36:20] it's 800 fathers so I have very little power to to see anything but it it doesn't seem to be really like that it's only mothers it seems for for fathers who sort of have the more female role during marriage um these these penalties [02:36:33] also show up and then as I mentioned earlier it's more highly educated mothers who could have potentially gained more from moving. [02:36:41] Now, um there was a question earlier about the age cut off uh and this is cumulative. So, it's always a child age six or younger, a child age seven or younger. The point estimate I don't have enough uh power in the sample to do this [02:36:54] one by one, but you do see that the greatest negative effects are really in the sort of early primary school age. [02:37:02] Now, okay, let me skip this. I I do a number of robustness checks. If you would like to see anything, please let me know. Um uh one yes also regarding the great recession, but this hasn't come up too much uh [02:37:16] today. I wanted to show you a bit more on the mechanism that I think is at play and I already said I think it's a mobility constraint issue that doesn't allow parents to move further away and that especially mothers are the ones who would have otherwise like to move or [02:37:30] gained from moving. Here, I'm going to be looking at the location of workplaces uh that are single establishment firms. So, I'm going to be looking at how close or how far the ex-husband and the ex-wife are [02:37:43] are working from each other. And you see that prior to divorce, there is virtually no difference in the treated and control group. However, the picture looks, if you don't look at the labels on the axis, looks pretty much the same as the the wage picture. That it's [02:37:58] gradually over time that these individuals are working in greater proximity to each other. And again, it's not that they're moving closer together. [02:38:06] It's rather that the control group move further apart. So again, there's a there's a flat line here and an increasing control group. And you may think, okay, this is about 2 km. 2 km should be easy to overcome by commuting. [02:38:18] Again, this is intention to tree estimate. If you scale this up, it's about 13.2 kilometers distance in century. So travel time can be more substantial. And I think Elaria is later going to uh say a bit about how even [02:38:31] smaller changes in in commuting or congestion can affect uh labor market outcomes of parents by a lot. So uh this is not this is not small. [02:38:43] Now I can look at an alternative outcome. for example, the probability of divorcing parents to remain in the same municipality post divorce. And here you see also that um prior to divorce there's not much of a difference but [02:38:57] after these like uh ex partners about three percentage points more likely to work in the same municipality. [02:39:06] Uh because I have a few minutes I can show uh show you this. No, sorry. Um yes. Oh yes. [02:39:14] >> Have you done uh some sort of back of the envelope calculation as to how much it benefits you to be living next to your ex- spouse and sharing the custody this way which meaning one time you know [02:39:28] one week per every two you're not paying for the kids food you're not paying for the kids child care you're not doing the transportation arrangements and how much they're giving up in terms of future wages because it'd be interesting to see [02:39:41] if on net I mean you care about money but you care about how you're spending spending your money on that. They may not be better off in terms of what that purchasing power gets them. [02:39:52] >> It's true. It's it's a little bit tricky for me to do this because I don't know exactly who is co-parenting. So, I there's a wide range of income distributions in the sample and in the survey data I see hour I see hours worked and employment status, but I [02:40:07] don't have their wages even self-reported. So it's hard most likely it's those parents who are higher earning who can afford to have a bedroom for the child in two homes for example. [02:40:16] So there's an additional cost margin as well. Um but it's and and but those people those parents are probably also the ones who would spend more on their child. So in that sense you would have what you say that if somebody pays more [02:40:28] on on that uh that you don't have to pay you would also see it's a bit of a more net benefit but at the same time I would still uh want them to be in in jobs that pay them more per hour and rather reduce [02:40:42] their hours rather than them continuing to work the hours. Uh I sadly cannot do this in a credible way I think what you suggest but it would be really nice to see. [02:40:52] >> Yes. Oh, sorry. I was just wondering >> if you have any sense of um extending the time period longer into the future. [02:41:01] Does this peter out? Um >> so um the reason why I kept it at two years is that there's evidence that custody arrangements are pretty stable in close to 90% for up to two years. [02:41:11] After that there can be there can be renegotiations and so on. So it's less clear who which share would still be in in um shared custody, shared uh physical custody. I have extended it by a little bit by like 3 to four years and the [02:41:25] results they don't increase. So the wage part after up to two years it goes down. [02:41:28] It doesn't go down any further but it stays roughly at that level. So it stays at about um like yeah z minus 1.1 in the sort of later later years. Um but I haven't done the very long long run analysis. So two good questions. One as [02:41:43] soon as a child turns 18 then presumably I mean not exactly then because they don't leave always but maybe you would see a bigger jump in uh the mom moving. [02:41:50] So it' be interesting if there's any slight discontinuity in the job at age 18. And then second um for this debate about whether this is good or bad, another question as a policy maker I would have is can you look at any child outcomes? So is it better for the child, you know, despite the earnings penalty [02:42:04] for the mom, as a mom, I might be still okay with that if it's better for my child to have good involvement for both parents. [02:42:10] >> Totally. I think so. The moms internalize this and it's I would assume that so they they're not like stupid in in making this decision. Um and I think most parents would rather their child has better outcomes uh than than this. [02:42:22] Uh I have not I don't have great data for the child outcomes. So there's school information but it's more or less sort of completion of which level of schooling. Um and also I can't really look at many kids who are 18 yet because it's too early. Uh but also it would [02:42:37] come at a time in life for parents when they're much less likely to move in general and also when they would see less of an earnings boost on average from from moving. [02:42:47] All right. Um, yes, I just mentioned like I also did this by labor market concentration and you really see it's those parents in the more concentrated labor markets during marriage who would have moved away but then didn't and also it's those who who see the much larger [02:43:02] negative wage effects. Um, yeah, this is oh somebody asked an alimony earlier. I wanted to show this at the end. It's just really that there's no correlation between alimony and uh co-arenting in the survey data. Uh but let me conclude [02:43:16] here. I'm almost out of time. So I showed you that this reform that increased co-parenting also slowed uh wage growth for mothers after divorce and that ex partners remained working in [02:43:29] closer proximity which is consistent with a story of a post-dorce collocation problem that in terms of labor market outcomes binds on average only for mothers. And there of course you have a policy trade-off if there's evidence [02:43:42] that children benefit and of course uh mothers know this. Uh but you also have this clear economic cost that mothers are exposed to. Uh and also they are inefficiently allocated across basic. If somebody could really like contribute more economically that's why they paid a [02:43:56] higher wage elsewhere. Um this is sort of a loss at a macro level. And one policy consideration could be irrespective of the uh potential uh unclear effects for now on of remote work on longer run career outcomes. I [02:44:11] think this is a key demographic that could benefit from greater remote work because it would allow these mothers to overcome this uh location constraint and access higher paying labor markets further away. Thank you.