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.
=

# Informality and Motherhood Penalties over the Life Cycle

Authors: Karolina Andrade, Diogo Britto, Lucas Finamor, Alexandre A. Fonseca, Fernando Mattar, Breno Sampaio, Gabriel Ulyssea  
Discussant: None  
Video: https://www.youtube.com/watch?v=0mE9GNhwdY8&t=6370s


## Talk (01:46:10 – 02:32:22)

[01:46:10] >> All right, everyone, if you could take a seat. Uh, we're going to get started with uh the second half of the morning.  
[01:46:17] So in this session we have uh one uh longer 45m minute presentation and then we'll have two egg timers for 15 minutes. So just as a reminder we have uh 10 minutes of the beginning of the presentation without questions so we can  
[01:46:32] get through some of the introduction and then you guys can ask questions. Um please go ahead. Thank you so much for joining us. We're going to hear about informality and motherhood penalties.  
[01:46:42] >> Yeah. No, thank you. Thank you all for being here. Thank you the organizers for including our paper in in the program in this very great program. I'm Lucas. I'm presenting this uh paper on informality and the motherhood panels over the life cycle. It's joint work with this great  
[01:46:56] team that's a large team like across scattered across the world. Uh won't present them. So let's jump to the paper. Uh so the good part of presenting now in the second day of the of this of  
[01:47:09] this conference is actually I'll be talking about a lot of things that we already saw at uh and hopefully in a new angle. I I hope for I hope so and and uh let's see what that gets. So we all know that uh the large share of the gender  
[01:47:24] inequality is actually explained by the motherhood penalty. So Camille here obsession as he mentioned uh with the child penalty actually contributes a lot to our understanding of that uh across the world and in particular there's a high incidence of the motherhood penalty  
[01:47:38] in in developing countries and in particular so in Latin America which will be the focus of of of this paper and there's another feature of like the Latin American labor market context that will be important for this is like  
[01:47:51] actually two features one is there's a this presence of a very rigid formal labor market regulation um and in in for formal employment but also there's huge informality and why does that intersect with uh um with the motherhood  
[01:48:06] dependency first of all we already saw it and we know from the literature this very great paper by in esperno and co-authors uh that informality is largest uh among women and especially so after uh childbearing so like women are  
[01:48:19] more likely to be working in informal uh jobs after after having kids and when we think about informality.  
[01:48:28] Uh we al the first thing that I think it comes to mind is all the bad outcomes that are associated with uh uh with informal jobs. So for example, lower earnings and lower uh on the job human capital accumulation, a lot of like lack  
[01:48:42] of protection. You're not contributing to social security benefits. You're not entitled to any leaves like maternal leave, parental leave, sick leave, anything like that. And also much more risk, right? Like there are higher volatility on earnings. you are more likely to be fired. Uh so there's all  
[01:48:57] these very bad outcomes that are usually associated with with informality. But also there's another side of informality that it may bring a lot of flexibility and lower participation costs and among  
[01:49:11] this like several dimensions that we actually saw it yesterday and today. So uh usually um informality is associated with like a higher share of workers working part-time, a higher share of workers working from home and a much  
[01:49:25] lower commuting time than their formal counterparts. So what we are interested in this paper is basically whether there is this interoral tradeoff where women after having kids looking for this more flexible uh job arrangements would tend  
[01:49:39] to go to more to the informal sector but that is compromising in the future like the career progression and security uh over the over the life cycle. So that's exactly uh the question that we are asking. So the first thing is we'll be  
[01:49:53] actually asking what are the career costs uh for women in this economy that has these two features this rigid formal labor market and high high informality and whether there is this intent temporal trade-off that I was uh  
[01:50:05] referring to and then the second is like can we think about like uh public policies that like how they shape this this career costs and how we can mitigate them uh in somehow and then to  
[01:50:18] answer this we'll be using a very high quality adomine data which would be linking 40 million births in Brazil uh to parental outcomes. I was joking with some of you yesterday that that's kind of like Scandinavian data in a  
[01:50:32] developing country. So it' be super nice. I hope you like it. uh and and then we estimate the motherhood penalty in a very granular way and in Brazil both for mothers and and fathers and then we move to develop and and estimate  
[01:50:46] a partial equilibrium model where women would be choosing both fertility and the sector of employment. Uh and then we use this model to try to first quantify the channels and the mechanisms behind this motherhood penalty and how can we think  
[01:51:00] about policies to to reduce them. And then we see this paper kind of like bridging this very two large literature's one in the career cost of children that like there are a lot of you here in the room that contribute to this literature and here we are not only  
[01:51:14] I think estimating this in using very high quality data for a developing country but we'll be very granular even in the timing of the motherhood penalty and you'll see in a couple of slides that I think that has insights for even developed countries and how we think  
[01:51:27] about the motherhood penalty there. uh we also bring in this like life cycle model where women are simultaneously choosing fertility and and labor market uh decisions and then there's other literature on labor supply and informality but usually with a uh  
[01:51:41] abstract away from from fertility decisions.  
[01:51:46] Okay, let me describe this almost Scandinavian data in a developing country context. So we start with like this in this data in this slide I will show actually four different data sets that has some versions of links between  
[01:51:59] the children and their parents and they will vary in terms of like which links are available and which part of the population they are available to but then after we I I fill this slide in like with the four data sets I think I will give you a picture of like how  
[01:52:13] comprehensive that that is. So the first one is this tax authorities registry.  
[01:52:17] This is a universe of adult population and then there's like name, gender, date of birth. But importantly uh there is the mother's name there uh for every uh person in this data set. So for example, if I'm in this data set, I would have my  
[01:52:31] my my the child my me as a child like my ID, my social security number, but also the name of my mother being written there. And then if we also use a second data set with this tax authoris tax  
[01:52:44] return data, we have tax returns between 2006 and 2019. And that's good because like whenever a parent uh files the tax return and claim one of the children as a dependent uh we have both uh of their  
[01:52:59] social security numbers. So both like the parents ID but also the children's ID uh there. What is the issue? It's usually only one of the of the parents that would claim uh their child as as a dependent and that only covers like the  
[01:53:13] upper part of the income distribution because not everyone uh uh needs to file a tax return in Brazil. And then the third data set to kind of like covers for the the gap that the second one generates is this kadastro is a low  
[01:53:27] registry um it's a it's a registry uh for welfare program. So like whenever you need to receive a welfare program uh you need to be registered in this data set. uh so basically it covers more like the bottom of like the income uh income  
[01:53:40] distribution and there again we also have child ID and then not only mother's names but then both parents' names in this data set and lastly there's this school census which is administrative records of all the students in Brazil  
[01:53:54] between 2008 and 2017 and there we we don't have IDs but we have then the children's name and their parents' names so with all the four of them we have like a sequential uh way of trying to  
[01:54:06] get for each uh child who is the mother and who is the father we do it separately for mothers and father so for example if here in the I don't know the father claim the number two like we have like a high quality uh knowing that this  
[01:54:20] is the link because of the ideas but then we also look for the mother or vice versa so we do separately for mothers and and and fathers uh and then I think this gives kind of like a quite uh a comprehensive view of like why we can we  
[01:54:34] can link so many births using uh these four data sets together and then these four data sets will be used to construct the family links and then we use uh additional data sets for like the outcomes and for the analysis  
[01:54:47] uh and mainly two data sets. So we have highs which the matched employer employee data in Brazil that covers like the universe of formal labor contracts and also household surveys u at the  
[01:54:59] quarterly level which is uh pinage. Okay, so I' I've been talking about like informality a lot and I think usually it's good to define because not every paper use defines informality or uses  
[01:55:13] informality in the same way. So I'll be using uh three definitions uh three different categories and then that's the definition. So when I'm saying formal employment, formal job or formal sector, I'll be using these three words almost  
[01:55:27] as the same meaning. I mean an employee. So there's the person has to be an employee with a formal contract. That means like in Brazilian uh language that's like there's a registered booklet that means like there's a formal contract the government knows about you.  
[01:55:41] You're paying taxes you're contributing to social security you're entitled to unemployment insurance to leaves to pensions all the packets that we usually think are associated with uh with a job.  
[01:55:52] When I'll be talking about informal uh employee, I will be talking only about an employee, but the flipped of like everything that's in the first one. So, it's a person that's working for a firm as an employee, but without a formal contract. So, this person is not paying  
[01:56:06] taxes, is not contributing to to social security, is not entitled to UI leaves and and so forth. So, it's kind of like the flipped uh version of the first, but there's also self-employment, right?  
[01:56:16] Like so there's three uh sectors would be separate both in the analysis and and also later later in the model. So you're self-employed if you in the household survey for example you said uh you are self-employed and then you are typically also not paying taxes and not  
[01:56:31] contributing to social security and you're not entitled to UI to leaves and benefits. So in some sense that's like some papers usually will bunch the informal and self-employment together in a nonformal sector. Uh we will keep them separately. Uh but you should like bear  
[01:56:46] in mind that like the two of them are kind of like informal in this sense of uh uh not having like security, not having insurance associated with this with this type of of of employment.  
[01:56:59] Yeah. >> Um can you Okay. There's a growing recognition that um I think in a lot of industrializing economies that employees  
[01:57:13] aren't either formal or informal, but in many cases partially informal.  
[01:57:18] >> Um and I'm just wondering if that's going to matter here. And partial informal is like someone is paid on the books formally but only a fraction of their wages.  
[01:57:26] >> Yeah. So that for us they would be in the first category because probably they would like answer like the I mean we would have their their administrative record.  
[01:57:33] >> We would understate their compensation. we would understand there I mean yes in the administrative data but like for earnings in particular would be mostly relying on the household survey and then I think it's likely that this person would be reporting their full like gross wage even though they are just paying  
[01:57:47] taxes on a fraction of it >> uh that still has consequence that for example for UI like if this person is fired probably their UI compensation would be lower because uh that's actually what was was registered but for  
[01:58:00] this talk I think like you can think of them as formal Yes.  
[01:58:05] >> So, how will you distinguish between informal and self-employed? Like, think of somebody in the gig economy in the US, I think we would call them selfmployed.  
[01:58:15] Uh, but if you were with a cart selling mangoes in the corner of a street in uh Brazil, in Rio, would you think of them as informal or self-employed?  
[01:58:27] >> So, that like basically depends on what they answered in the household survey.  
[01:58:30] If they answered that they are self-employed, they would be here. If they answer that they are working for a firm like they are an employee then we would be in the second but as I mentioned like that's in this the sense like that both of them are like unprotected from the point of view that  
[01:58:44] are not contributed to social security >> I mean we will like there will be separate like I think there are important dynamics that are different uh but in the sense like of unprotected they are the same yeah >> but so we might also think about the  
[01:58:58] distinction between informal and self-employed in terms of like who's the you know who has a first claim on profits who bears the risk and in the friction in how smoothly you can adjust your hours and so are those features >> those will be present and it'll be very  
[01:59:13] important like self-employment like they'll be much more likely to work part-time for example than formal employees I I'll be showing why not >> okay >> yeah that will be present yeah yeah yeah >> and I think like actually I'm actually  
[01:59:25] showing this now so I think we saw it uh a couple of like uh numbers from like uh from developed economies of like for example the fraction of women working part-time. Uh I hope I don't shock you  
[01:59:38] with the numbers but this is the fraction of both men and women uh split by education. So at most high school education or uh high school education more uh working part-time and here I'm being very generous of like saying part-time is less than 30 hours. We saw  
[01:59:52] it like in some countries like full-time was actually close to 34 or 37. So I'm defining part-time here as 30 which maybe is a little bit high even uh for example women with at most high school education which will be the focus of our  
[02:00:05] uh analysis later on only two and a half% of them uh are working part-time in the formal sector. So basically there is no part-time opportunities in the formal sector. I just think like this graph makes that very clear while there  
[02:00:20] are a lot more if they are informal employees or even more if they are self-employed. So for example for women that goes from two and a half for women with at most high school education to like 10 times higher for informal for  
[02:00:33] informal and then I don't know I I won't do the math but like it goes up to the 38% uh for for self-employment. So this is in the dimension of part-time we have the same as in the dimension of working from home. So this is the fraction of  
[02:00:48] them working from home. Uh so that doesn't go a lot higher with informal but is is uh a little bit higher. And then almost half of self-employed women are working from home. And then the same is about commuting time and even  
[02:01:03] conditional on any. So like the number that I'll be about to show you is even like we know that 50% of self-employed women are working from home. So like get the other half that are still commuting uh even so they are commuting much less  
[02:01:15] than the formal uh uh the the ones that are formally employed. So as long as these three dimensions that we've saw it in the other papers are very important right like the fraction working from home the commuting time and uh the fraction working part-time these three  
[02:01:29] sectors these three types of job arrangements offer very different distinct uh um uh opportunities for for women.  
[02:01:38] Okay. Now let's go to the the motherhood penets in the in the informal in the formal sector in Brazil. So what we do we have that family link that I showed you in the four data sets but to be precise and then to be able to compute that like in a very granular level as I  
[02:01:53] mentioned we'll be using as outcome for now only the formal sector employment.  
[02:01:57] So everything that I will show you in the next couple of slides will be all the family links but as an outcome whether you are working formally or not.  
[02:02:06] So like the the zero there could be being out of the labor force or uh working informally right. Uh so what we do is we're basically restricting the sample to individuals with at most uh high school education. We do so just to  
[02:02:19] be better aligned with the model later on. Of course you could run everything for like the the the fraction uh that has also some college uh education. And then we do a different with matching where the treated parents the treated uh  
[02:02:34] individuals will be parents that had their first child in 2012 or 2013. And the potential controls will be all the parents that had their their kids four years later in 2016 or 17. And then we'll be matching them as I mentioned  
[02:02:47] like we have 40 million birth. So we can be very specific here. uh the the potent the we'll be matching the controls on in gender, the birth cohort, the municipality, their education and uh their status in the previous year in the  
[02:03:01] formal uh data set which is high or in the their uh presence in the the catastrophic which is the lower uh the the welfare uh uh data set for like uh  
[02:03:14] the low-inccome uh families. Then we'll be implementing the traditional event sets approach. I won't beliber too much on it and before even doing anything fancy on econometrics I would just show the raw data between this treated and the match edited control and I think I  
[02:03:29] could even stop it here because I think that shows everything. So like in the top there we have fathers. So the gray one uh is the control fathers, the fathers that will be employed will be will become fathers uh four years later  
[02:03:41] and then uh in yellow the the treated fathers and basically nothing or a very small uh decline in their probability of working uh formally uh a little bit later than one and a half years after the birth. And here notice that we are  
[02:03:56] doing everything at the monthly level. And then when we look at women, we see the traditional uh like child penalty uh graph. But notice there's something here that like is a little bit interesting that like when we do it at the monthly  
[02:04:10] level is like a little bit flat actually around the the birth of the child. Uh and there's something happening even before the child is born. And the first time I stared at this, I thought that was very very interesting. And then if  
[02:04:24] you count, I'm calling pregnancy here just like nine months before the the kid is born. Of course, we never know when the woman got pregnant. I'm just counting nine months. We see that like this diversion seems to be happen starting as soon as uh women get  
[02:04:38] pregnant. Uh I will get back to it in two slides if you have questions of why we are seeing it. But let yeah let me hear your question and then I tell you >> no no no go ahead please  
[02:04:52] give us a little bit of a um of a sense of what is the legal protection for pregnant women in the formal sector and how well it is enforced.  
[02:05:02] >> Yeah. Uh so I will defer like when I talk because then I will show that that that works uh but for formally employed women. Uh if you were worried about like the the control group that we selected that was like we matched on people that were uh getting pregnant four years  
[02:05:16] later. Uh oh sorry this is the event studies version of this graph that I think it shows the same thing. So just I skip it. So like nothing oh almost nothing for for fathers and then this of course huge penalty for women and then  
[02:05:30] that's something uh arising and then and notice that like there's something arising before it's actually quite substantial here. uh it's almost like 40% of the entire penalty it's actually arising before. Uh so keep that in mind.  
[02:05:43] Okay. If you were worried about like okay we are using this four years maybe they're very different uh we show that basically they're not. So if we use control groups of two or three we would have the same of course they start like not being the same as they are closing  
[02:05:57] to the control group uh getting getting pregnant. Uh so basically even like before one year one year and a half they are all the same the same pattern. So that's not a concern. Okay, let's go back to that  
[02:06:10] somewhat puzzling thing that like this uh motherhood penalty is actually uh materializing before the the the child is is uh born. What we did, we went back to the data and we split this sample in  
[02:06:24] in uh according to the baseline year which is 12 months before the kid is born or one year before the kid's born uh according to your formality status and that's the graph. So the top part is the women both in the treatment and  
[02:06:39] control that were formally employed 12 months before the kid's born or 12 months the placebo uh when the kid is born for the control group. So we don't see any difference uh until the kid's born like almost no difference in during  
[02:06:54] pregnancy and then you can count one two three and four dots after the kids born four months of maternity leave in Brazil and then that declines. So basically uh answering your question now so four months of maternity leave actually you  
[02:07:08] are entitled to six months of job protection. So that's why I don't think it it drops exactly at the four. Uh but you can see like as soon as the maternity leave ends, women start uh to not be employed in the in the formal  
[02:07:20] sector. So basically this is for women that were formally employed. Now the the bottom part of this graph for women that were not formally employed. So what happens is like of course if you are conditioning on them not being employed  
[02:07:33] in minus 12 and then in minus 11 some of women will get formal jobs and then the same for the next month the same for the next month as soon as they get pregnant the country women still like keep  
[02:07:46] accessing formal jobs keep flowing into the formal sector however that's completely flat for pregnant women. uh of course with this data and with unfortunately with all this approach we cannot say why whether that's discrimination whether that's preference  
[02:08:01] for women that's like whether women are anticipating future discrimination and stop searching well together uh I would love to be able to to answer that we can't with this data yeah >> yeah um  
[02:08:14] >> um what explains that big uh shift into formality but more um more importantly and I'm now forgetting the authors was I have terrible memory. There's a paper that looked at um people going in and  
[02:08:29] out of for formal to un informal and what they showed was people and they had quarterly data for two years. There it was a panel data set and what they showed is people really shuffle in and out of these sectors over you know  
[02:08:42] basically after three quarters they're very likely to go out because they're getting really bad jobs in the formal sector. These are not good jobs necessarily in the formal sector.  
[02:08:52] >> Yeah. Yeah. Absolutely. You you can see that there's a lot of like churn and movement because like for example if you go like one year before like only 60% of them were for example formally employed and then that goes up to 100. So there's a lot of churn actually in the model we  
[02:09:06] be we I won't I was not like bringing this in this presentation but we actually have two types of of of women and like the model will be only about women. uh that there's like the high churn type and then the low churn type  
[02:09:20] which exactly to to match this pattern. It's like with only one type we couldn't match like there's all these transition.  
[02:09:25] So you kind of need some heterogenity in order to to me to to mimic this this transitions. How long?  
[02:09:41] >> How long do you need to be in the formal um sector such that so that you qualify >> actually for no time requirement? So like if you are employed in like the last month of pregnancy, you are entitled to maternity leave. Oh really?  
[02:09:55] and to job protection. Of course, then it's >> my hypothesis would have been that it's 12 months, but [laughter] maybe not.  
[02:10:02] >> Then it's very difficult for you to get a job uh if you're lucky. Yeah.  
[02:10:10] >> Okay. So, I'm doing good on time. Oh, I'm sorry.  
[02:10:14] >> Yeah. >> Are there bonuses built into formal employment contracts like 13th month or something that would depend on the duration?  
[02:10:24] >> Yeah, they are. Yeah. Okay. So, so, so Martha's timing intuition I think maybe is born out there.  
[02:10:36] >> Yeah. Yeah. So here like for maternal living like you don't but for all other benefits including for example UI you need like some time in the uh in informal employment and then I think like just before I move into the model as I have a bit of time not much but I  
[02:10:51] think like that actually being able to do that at like the monthly resolution here like in in in time it's actually very very important and you can see that like like 40% of the the motherhood penalty actually as I mentioned arise  
[02:11:03] before the kid is born and that's I think That at least says something like when we run the the m the the motherhood penalties like at the annual level maybe we should stop using minus one as the reference and we should use minus two as  
[02:11:17] the reference because there are some effects already in minus one for like a large part of the uh of women at least in in our context that that would be true. Uh so that's why I said like I think this paper even has some methods  
[02:11:29] more broadly on how to think about the the motherhood penalty. Yeah.  
[02:11:35] Do you interpret it as >> I mean I I won't make any [laughter] claim about it. Uh I think it's like a mix of everything. It could be like in  
[02:11:49] the beginning that women are feeling sick and then they are like they put less effort in search because they are feeling sick because of pregnant. It could be they are already anticipating that it would be very hard for them to find a job. Uh it could be discrimination. I mean I don't know. Uh  
[02:12:02] I think it's all mixing together here. I think it would be a great topic for future research to kind of like disentangle all these mechanisms. Uh we won't be able to say anything unfortunately about it in the model that would be all lumped into one factor that  
[02:12:16] depress the arrival rates for women uh that could have all this this uh mechanisms behind it.  
[02:12:23] >> Do you see occupation? we see maybe you can look at whether there is any heterogeneity by um you know prevalence of u women in a specific sector or in a specific occupation  
[02:12:38] assuming that maybe they're going to be less discriminatory against a pregnant woman.  
[02:12:43] >> Yeah, I mean we could uh to be honest like as this data is very large we're thinking okay there's like a lot of very interesting heterogenities like of this graph that we couldn't do. uh we just yeah we were thinking about maybe doing a separate paper just focusing on that  
[02:12:58] because this would be become too much for this one but I think that's certainly interesting like for occupations of sectors for like your history and and on that on that sector and and in all of that I think it's very very very interesting  
[02:13:10] okay let me jump into the model uh I will put myself at a challenge and then just present the model in words but if anytime anyone wants I click a button and then all the formulas and equations will appear  
[02:13:23] But let's see if how I go with words. Uh if it's you're not satisfied, I'm happy to to click. So the the model will be a life cycle model where in every semester woman woman um will choose whether to try to conceive and importantly there's  
[02:13:38] a try there like so we will be built in that the that is an uncertain event right like you can try but maybe you won't be successful in in getting pregnant and then there are like a bunch of labor market decisions that actually will be like three discrete decisions.  
[02:13:52] So first decision it's whether you participate or not in the labor market in that semester. If you do participate whether you work formally, informally or self-employed and then conditioning on working um you can uh work full-time or  
[02:14:05] part-time. Yes, a question.  
[02:14:08] >> Does that does that sequencing matter? So from the data you showed us earlier, it seems like actually what people are doing is picking part-time or full-time and then conditional on that selection either  
[02:14:21] formal work I you know they can either search for formal work or or not.  
[02:14:27] >> So we have actually search frictions here. So it's not that they can literally choose the three sectors depend on having an offer and then what we actually have is like some offers will offer you the ability of work part-time and others won't. that is just the ability of working part-time. It's  
[02:14:42] still a choice. If if your job offers the the possibility of working part-time, you can still choose full or part-time. But if your job doesn't have this option, then you just you can only then sequence is important. Yeah.  
[02:14:54] >> Yeah. >> I mean the first one is almost logical, right? Like if you're not participating then like you cannot be working. But then the second maybe Yeah. There's a a lot more things built in. Then women will be different uh in a lot of dimensions in the model. So first of  
[02:15:09] course in demographics so there will be different uh it's a life cycle model so there's age uh and marital status they can be single uh or married uh in the number of children they have and then the age of the youngest uh kid um in experience they can be different uh in  
[02:15:23] terms of formal and nonformal experience and this this is something that like I think we also are um bringing something new here so we are actually building in the model that this the ways the sectors accumulate experience are different so like if you're working in the formal sector you are accumulating what we are  
[02:15:38] calling formal experience. Uh while when you are working either informally or self-employed, you are accumulating some experience that could be important or not. But it's a different uh type of experience and then these two experience may be rewarded differently in this two  
[02:15:52] in these two sectors. Yes. >> Can I ask is the distinction with self-employment and jobs coming to you  
[02:16:06] that you can always do self-employment, you don't have to wait for an offer.  
[02:16:09] Like what are the benefits of self-employment relative to informality?  
[02:16:13] >> Yeah. So all these like we have like this three sector should be totally different in every dimension that I will present to you in two slides. Um it will still be the case that there are some frictions on self-employment. Uh so like you cannot always be self-employed. So  
[02:16:28] we interpret this as kind of like arrival of business opportunities or something like that. uh but there will still be a a smaller friction but there will be still a friction in self-employment mainly because if we don't uh include that it's very difficult to fit the self-employment  
[02:16:42] pattern and to be transparent. Uh and then of course there's another dimension that we may we defer which is your uh lab your labor market status in the last period just because of course if you're employed you can continue always continue to be employed if you  
[02:16:55] are not fired. Uh and as already said to you um um there are uh labor mark there are search frictions in in the labor market. So you cannot always choose these three options only when the offers um arrive.  
[02:17:08] Okay. What are the career cause of children in this model? So like there will be three channels through which children will affect um mothers and like in in this model and especially when they are their relationship with work.  
[02:17:20] The first is what we call like this participation cost. So this will be measured like in utility. So there's a disutility for you if you choose to participate. It doesn't matter if you even have a job, you have unemployment because of the the frictions. Uh if you  
[02:17:34] just choose okay, I want to participate and that depends on the age of the youngest child. So like for example, it's very costly for women to participate uh with young uh babies and that cost likely declines uh when the the kids uh get older. So that that's  
[02:17:49] measured in utils and that only depends on your decision on to participate or not and that's it. Then the second is more like childare child child care cost that in the model will be measured in hours. So they will be subtracting  
[02:18:01] leisure of mothers uh in the in the child care that will also depend on the age of the youngest child. But importantly that also depend on the sector of employment. And what we want to capture here with this child care cost depending on the sector of  
[02:18:15] employment is that like okay the sectors will be different in part-time opportunities will be different in maybe the commuting time. But there are other ways that like these sectors maybe are more flexible than than uh the formal job for example in your ability to  
[02:18:29] change your schedule, your ability to not work in the evening, your ability to not choose not to participate in a late like a meeting in the day. So this will be all built in into this heterogenity here uh depending on the sector and it  
[02:18:44] turns out like we didn't know a priority what would that be and that turns out to be very very important. And then the third is I uh told you when we were seeing that graph like we we're building the search costs which is the search cost is basically saying that the their  
[02:18:58] arrival rates for either pregnant women or women with young uh babies are smaller uh that can be again discrimination search effort preference we don't know that we will all come to this that I'm labeling uh search search  
[02:19:12] costs and uh we'll be in a model respecting the Brazilian legislation of four months of maternity leave and job protection uh for six months. As the mo as the model is in a semester, we'll be just saying that the mo the mother is  
[02:19:26] enjoying full maternal leave and job protection for one period in the model.  
[02:19:31] And then the in terms of like how these sectors are different as uh someone already asked they are different in all dimensions that we characterized uh job opportunities. So they are different in the earnings process in the returns. Uh  
[02:19:45] um uh the or that means like in how do you actually um like uh how the the your your experience vector the form and informal experience translates to earnings in the human capital  
[02:19:59] accumulation whether you are work you're accumulating formal nonformal experience in the arrival rates destructions crucially in part-time opportunities in the fixed cost of working in trying to capture that that part of like it's uh those two sectors have like lower  
[02:20:13] commuting times than the formal sector and in childare care costs. And importantly, I think like these three last three bullet points here could interact with like women choices uh as they have kids, right? Like if if if  
[02:20:26] part-time opportunities is the the key thing, we are allowing the three sectors to have different part-time opportunities uh if the fixed cost of working is important and the same and or if it's all built in this child care costs. So a priority I think like again  
[02:20:41] the model could actually say that it's everything part-time like actually what matters is just part time in these three sectors happen to have like different part-time uh opportunities uh in a very romantic way marital status  
[02:20:55] in the model is exogenous. So you randomly get married and randomly get divorced. [laughter] And uh what marital status it's uh fits into the model. It fits in the dutertility of participation just to  
[02:21:08] capture the idea that um married women tend a higher fraction of married women are out of the labor force than uh of single women and that they have more kids. So that these are the only dimensions that uh marriage um matters  
[02:21:22] in women's decision here. And then consumption will be a typical CRA over consumption in leisure. Uh we don't have savings. So basically whatever you consume is whatever you have of earnings of that period plus transfers. Um um  
[02:21:36] that's it. Yes.  
[02:21:41] So I know you said that you're going to look at formal versus informal sector experience, but how about just the skill level of women? Because, you know, if you're worried that some employers are just laying off women because they're  
[02:21:56] now thinking, "Okay, I have to pay this woman four months of pay after she gives child birth. She's too costly." You're going to fire them if they're low skill because you can just hire them again as soon you know there's a very big pool of  
[02:22:11] them. It's no problem. But if it's a high skill woman as talent matters, then in those sectors, you're less likely to do that. So I'm just wondering if you build that do you get more insight?  
[02:22:21] >> Yeah. So important like so this paper it's actually focusing a lot on like the labor supply. So like we're so like basically the firms here are almost absent like they're just generating uh the offers and so like there's no like  
[02:22:34] indogenous decisions of firms. Uh in this model we have something like along those lines of like my answer to Raquel which is like there's two types of like the churn the high churn and low churn.  
[02:22:45] So like if you think that it's also correlated with skew uh then the lowest skill will have like a higher probability of being fired for example.  
[02:22:52] So there's a dimension of this that will be present in the model but not really like a I don't know a formal feature of the model.  
[02:23:03] Okay. Uh so now I will summarize two and a half years of work in one sentence. We estimated model which [laughter] very very difficult very difficult like we're doing SMM uh but it works and I  
[02:23:17] the only thing that that I will say is like of course we are using the event study estimas as moments so in the sense of like trying to get like this nice feature that we have from the high qual data into the model by making the model  
[02:23:30] replicate uh those event states. Okay, two and a half years later we have the estimates and [laughter] um what okay with standard errors [laughter] so I'm not a macroeconomist so there  
[02:23:44] will be standard I mean I'm not showing any parameters so like I won't show the the stand errors but they are in the paper [laughter] uh I will just mention some of the of the parameters that I think they're that are interesting and then I'll move to  
[02:23:58] the fit and then the contrafactuals in the last seven minutes so in terms of there's three channels that uh we built in in the model of child care cost. I think that it's u useful to to get a sense of their dimensions. So if you like the participation cost the utility  
[02:24:12] of participation um for women if you decide to participate in the labor force and you will have a young baby a baby uh between zero and six months. This is equivalent for you of like losing 13% of  
[02:24:25] consumption and leisure at that uh at that time. So that's like a huge huge cost uh for women if they decide and that of course declines very very quickly like or not very quickly but okay quickly uh as soon for example for  
[02:24:40] a kids age three years that drops to 2.1%. Of course that does a continuous function and in the model I'm just showing you two two data points here for child care costs in in in the in time.  
[02:24:53] Um I yeah so one important thing I don't think that's like literally child care cost time like as for example presented by Camille again like this is building what like women are losing in leisure and how that's differential by this  
[02:25:07] sector. So not necessarily all the time that they are spending uh in child care costs but nevertheless I think one thing that's interesting that uh declines much more slowly than the participation cost.  
[02:25:19] So the participation cost is actually much for like young young babies and then that really declines quickly while this child care costs they are higher and then they are much slow to to decline. So that's equivalent about 21 hours of week if you have a baby and  
[02:25:32] that declines by half for example when the the kid is three years but crucially and this is much much smaller if you are working informally. So this is 82% is smaller if you're working informally and  
[02:25:45] almost like 90% is smaller uh if you are working self-employment which you already preview some of the results that actually that's the main thing. So part-time is not super important. Part time is again part of all this. Commuting time is part of all  
[02:25:59] this. But all these other dimensions of flexibility that makes this jobs more flexible are maybe more important than just the part-time and the commuting uh that we showed. And then there are also builtin. And then in terms of the third channel, the search costs uh for  
[02:26:14] pregnant women uh they're almost like totally cut out of the market. So they see uh their opportunities their their arrival rates dep are depressed by 94%.  
[02:26:26] Uh which basically is the flat thing that we saw at uh in in that graph. Uh but for for women with young babies that's a little bit better but that's still 25% uh depressed  
[02:26:40] and and in part-time we we have like the the first one is set. So like you can always be part-time. So going back to that question, you cannot always be self-employed, but once you're self-employed, you can always choose to work part-time. You're self-employed. Uh but then the the later two are  
[02:26:54] estimates. So like uh when you receive an offer from the informal sector, there's a 30% chance that this that that job will allow you to work part-time and for the formal sector, that's less than 1% as we saw it in the uh in that in the  
[02:27:09] figures uh in the beginning. Okay. Uh the model does a good job on fitting several features of the data. So for example, here's the life cycle uh u um splitting women over the life cycle by the number of children like nonchild  
[02:27:22] one, two or three. We can see that like the model fits that very well. Also uh the stat the labor market status by the age of the youngest child where we have none here and then the age of the youngest child in years for some beings uh being out of the labor force formal  
[02:27:37] informals of employment uh oh I thought I put the part time here but also part-time conditioning on these sectors by age of the youngest child and the event studies here we do a little bit worse job than uh on the other moment  
[02:27:51] but still I think we we we can fit the event studies almost uh uh in a in a nice way given the complexity of of this model. So in the last three minutes I would just talk about the contrafactuals. So one of the nice  
[02:28:05] things about this model is like in the event studies we are constrained by the control group like we can only see effects up to three and and and two months period because like in uh nine months before the four years the control  
[02:28:19] women will already be pregnant and then we cannot uh estimate anymore uh any of the effects. While in the model like we can extend that over the entire life cycle and you can see that like very much persistent from the model uh from the model perspective. Of course this  
[02:28:34] building that like women have more than one children. So like at some point here like they have the second and the third child that that accumulates uh I have the one child version here that I won't click but it's a little bit smaller. And here is I would just show this graph because I will use it again. So this is  
[02:28:48] the participation margin and then this the two others are the intensive margin.  
[02:28:53] So like condition on on women staying in the labor force whether they are the formal or nonformal occupation lumping informal and self-employment just to show this graph of course we could do it separately. So then we can see the two margins of the cost right like the  
[02:29:07] participation margin and the the sector of employment margin over there. Uh we decompose this I will skip just because I don't have time but we de composed this in all these three channels that we have in the model. uh what I want to say  
[02:29:20] is like first of all because of the uh on the part-time so what happens if we in the contrafactual we say that the formal opportunities will now be like the self-employment and you can always work part-time you will always have the  
[02:29:33] option of working part-time uh so that helps a little bit so like we we reduce a bit at least in the intensive margin of this but that's not the bulk of the effect and in particular not in in participation  
[02:29:46] uh if you do the same for the analysis the blue line would be a little bit confusing. Uh but that's the blue line here. We do the same for the commuting time. Again, it helps >> say a question. If people are working part-time, their pay is para.  
[02:30:01] >> So what we do uh we just we we do exactly. So we we just computed like in the med like in the median of the two groups. So like they're working part-time, they work 57%  
[02:30:16] um of the uh of hours. uh compared to full-time women and and then in the model they will earn 57%. Why we do it that linearly? Because like actually if we try to do that in the data then there's a problem of selection of  
[02:30:30] part-time actually there is a part-time like premium actually like the women that work part-time earn a little bit more per hour than the women working full-time.  
[02:30:39] >> That's generally not the case and it is generally a huge part-time penalty.  
[02:30:44] >> Yeah. Yeah. Yeah. Absolutely. So that's why we were like trying not to build in this here. Uh but I think like we could make like at least some sensibility of like how that is important of like going to 57% a little bit lower or a little  
[02:30:58] bit higher as in the data. And then what really kills the total uh this intensive margin is actually saying this child care cost is the same. So like this all the other dimensions that the the informal jobs are more flexible. And  
[02:31:12] then in the last 30 seconds I was [laughter] I will say about this which I think it's kind of like connects with this intertemporal trade-off that we said we did an experiment in this paper of like saying let's create like a formalization program like that that  
[02:31:25] that some countries think of like let's I don't know increase uh the the cost of like uh being being formal like increasing monitoring or whatever and make let's make this this formalization program such as that there is an increase in formal employment by five  
[02:31:40] percentage points and that there's a decrease in nonformal employment by five percentage points uh and there's no uh no change in participation. So we chose this particular design. What happens with women when they have kids? So we  
[02:31:53] actually increase uh the the childhood penalty in participation and that's somewhat expected right like we are actually making the jobs for women worse uh because in all the dimensions they are worse. So in this case like this  
[02:32:08] welfare goes up. So like in the long term that's better like the welfare goes up by 1%. But actually that decrease the participation of mothers with young babies. So that's exactly the intertemporal trade off that we are trying to analyze. So my time is over. I