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. = # Child Penalties in Time Use: A Global Perspective Authors: Gabriela Deschamps, Amory Gethin, Camille Landais, Gabriel Leite-Mariante Discussant: None Video: https://www.youtube.com/watch?v=0mE9GNhwdY8&t=0s ## Talk (00:00:00 – 00:46:04) [00:00:00] extremely simple. Um there's a large literature documenting the presence of of child penalties that is uh massive and systematic gender divergence in labor market outcomes when kids uh [00:00:14] arrive. Uh but labor market outcomes are by construction what happens when you uh invest time in the labor market. And when it comes to the time that we spend in the labor market, take even the most [00:00:26] active of us uh in between age 20 and 60, that's going to be at most 15% of the entire time that we spend in our lives. So that bears the question uh what happens with the rest of that time. [00:00:39] What about all the time and task allocation at the moment kids arrive uh that affect the rest of that time and create uh these effects in the time that we spent in the labor market and the labor market outcomes. So for this [00:00:53] basically what you need is time use data and what we're going to do today is just this. We're going to use on a global scale in a very descriptive way time use data to better document characterize and understand the gender specialization [00:01:06] that takes place around the arrival of children. We're going to have essentially three main questions in mind. Uh first uh when kids arrive those children what type of time investments do they require from parents? Second [00:01:21] question is also to think about how these time investments are allocated within the household in particular between mothers and fathers. And then we're going to also think about how this [00:01:34] uh allocation of tasks at the moment kids arrive vary across countries and over the course of of development. I think the goal through this is really to offer a better understanding of the mechanisms uh behind gender inequality [00:01:47] in the uh labor market. in some sense to microound through the concrete allocation of of time the mechanisms that lie behind the child penalties in the labor market and through this I think it's important as well because it [00:02:00] might offer a better understanding of broader concepts of gender inequality uh that take into account all the dimensions of well-being through the tasks uh that we're uh taking on every [00:02:12] every day. Good. So just to give you a brief preview of the of the results, uh maybe not surprisingly, we uh document that the arrival of of children corresponds to really a very large time [00:02:25] investments for for parents. Basically, when your kids arrive, uh you're going to have an increase in your total active time. I'll define total active time in a minute. Uh of roughly 25% and that's extremely persistent over the next 15 [00:02:40] years. Okay. Um what we also document is the massive change in both the extent and the nature of parental investment over the course of development. I'm going to show you that actually child [00:02:52] care time uh is extremely small at very low level of development and is essentially multiplied by three over the course of development. And with this what's interesting is uh you also see a drastic change in the extent and the [00:03:06] nature of gender specialization. Uh what we find is that indeed children drive much less gender spe special specialization at low level of development which is reminiscent of the findings that we had in the child [00:03:18] penalty atlas. Uh but what is in some sense really uh novel is to show how important domestic work is in driving the uh allocation of time spent in the [00:03:31] labor market much more uh than um the time that you invest in in childare. So at the end I'll show you how we can use these moments uh in uh disciplining a very simple model of household task [00:03:46] allocation to rationalize these patterns. And what they're going to teach us is that well that we kind of knew as well but it's extremely complicated to explain the patterns of gender specialization that you're seeing without extremely strong gender norms [00:04:01] with respect to the roles that different gender have to take in all the tasks uh in uh your daily time youth. Uh and the second thing is that it's going to be also very hard through the lens of these models to rationalize what we're seeing [00:04:15] over the past 50 years or over the course of development between middle inome country and high income country in this increase in mother's paid work. Uh if you don't have quite some action on the domestic work side of things through [00:04:29] marketization and increase in productivity in domestic work. Okay, good. So clearly uh this very descriptive paper fits uh into very large literature. I'd say that there are [00:04:44] at least four literature'ses that we're trying to branch out to. Most obviously given my obsession with child penalties, the child penalty literature is something we wanted to contribute to. Uh again, what's the contribution here? uh [00:04:58] child penalties because of measurement issues in some sense are essentially uh uh looked at through the lens of of paid work and labor market outcomes. Uh with uh opening the full 24-hour day at a global scale, we can really microfund uh [00:05:13] the the mechanism and in particular showing just how important domestic work is in driving gender specialization in the labor market. Okay. The second literature uh we're are trying to branch out to is the large literature on [00:05:28] parental investment in human capital. This state bags at the very least to to baker and there is a very large literature just trying to document how parental investment changes across level of income and so on and so forth. I [00:05:41] think here what we're doing is providing really the first global evidence on both the quantity but also the nature of parental investment over the full course of of development. And I think this is uh quite quite novel. [00:05:55] The third thing uh that we are trying to contribute to is that broad literature on time use over development with a gender length. uh there are a bunch of uh papers that have used either [00:06:09] historical data or some cross-country data to document evolution of time use but given most of that data was uh using uh paid work and time in paid work uh it hasn't add I would say the granularity [00:06:23] that we're going to that we're going to add here I think a very close paper is a paper by gut and and co-authors here what we're going to add is quite a lot of granularity uh and a focus on the on on parenthood. [00:06:37] And finally, I think uh we contribute or at least we aim to contribute to the vast literature on structural transformation and own production that dates back to uh work of economic [00:06:49] historians like Ruth Cohen. uh here I think um the uh important contribution is really to show uh and I'm going to show you a very simple statistical decomposition that proves that it's very [00:07:02] hard to explain the rise in female labor force participation if you don't understand uh the domestic work productivity revolution that takes place uh over the course of development okay good so without uh further ado uh let me [00:07:17] explain why uh I believe that the data that we've assembled here uh is really the key [snorts] asset of this project and why it has the uh ability to teach us uh really important new lessons on on [00:07:32] gender specialization. So what we've done basically is to partner up with sociologists from the center for time use research at UCL. Uh and uh the the goal is really to create a unique global [00:07:47] repository of all time use micro data that exists. Uh and right now we are covering roughly 80 plus uh countries. [00:07:56] uh and compared to uh previous exercises that have tried to uh use micro data on a on a large scale uh I think the advantages of our infrastructure are at [00:08:08] least three-fold. Uh the first is uh because of this partnership we put a lot of hands on deck uh and that has enabled us in the past year to expand the coverage quite massively and quite [00:08:20] quickly uh both uh across space uh and also across time. So maybe uh let me show you across space. Basically this is our [00:08:32] current map. Uh the idea is to use as much as possible a diary based data. Uh but sometimes this is not available and we have to use what we call recall uh [00:08:45] data where people are asked in surveys about specific time that they spent typically in the last day or in the last week on specific activities. But as you can see uh the the beauty is not only that we are covering large set of [00:08:59] geographies but also vastly different levels of of uh development uh with countries like Sudan uh so very very low uh income countries uh that gives us quite um [00:09:12] a nice range in terms of understanding what's happening to gender spatialization over the course of development. A really nice advantage as well compared to existing infrastructure is the fact that we cover China. I should be very clear, unfortunately [00:09:25] today we're not been able to get the authorization to release the data for China and so everything that I'm going to show you does not include China but uh soon enough. Yes. [00:09:46] >> Yes. uh in the American tu survey a lot of it is diary so this is why we're uh classifying it as as diary but yeah [00:09:57] okay uh so uh the other thing is we have for some countries now quite a lot of uh um depth in terms of historical data uh some countries going all the way back to the 1960s okay so that's the first [00:10:12] advantage the the large coverage the second advantage is that every Everything we're doing is going to keep all the data at the episode level. Okay, that's going to offer us a lot of [00:10:24] versatility, modularity, and granularity in the sense that at the episode level in any diary, a lot of extra information is going to be collected such as uh where were you taking this uh activity? [00:10:38] Uh who are you with? Uh so core presence is going to be really important as well. [00:10:42] uh all this type of information we can keep when we're harmonizing all the data at the episode level rather than at the uh task level as most of the uh recent uh exercises of harmonization have have [00:10:56] done. So basically the u harmonization efforts that have been taken so far are essentially saying oh let's define a concrete set of uh activities such as child care and then basically they just [00:11:10] tell you at uh the level of an individual how much childare has been done in a day. Okay, whereas we're keeping all the micro data at an episode level so that if you want to recreate your own set of activities, you are absolutely free to uh to do so. And on [00:11:24] top of that, you are keeping a lot of extra information that is particularly uh important if you want to understand more granularly what's happening with with your tenures. And the final and really important dimension of this effort. Yes, you have a question. [00:11:39] >> Yes. for you as a secondary activity through [00:11:51] my child care labor market. [00:12:12] >> Absolutely. That's a very important question. So, I'll try to show you some uh things on this. Uh in uh some countries, but not all countries, we're going to have very precise definition of co-presence. And so, here we're going to [00:12:24] be able to uh define supervision versus active child care. Okay? Uh in a lot of countries, we're going to have uh not only primary activity, but secondary activities as well. And when we have secondary activities, we're going to [00:12:38] count them meaningfully when we want to account for the childare that is being done. I'll I'll I'll be quite precise on this when when we go to this but I agree with you. This is uh this is not assumption free. Uh the the way you want [00:12:51] to uh allocate time that is spent on specific activities has to come with specific assumption that you put on the diary data that is given to you. Uh and the nice thing about this uh infrastructure is we're going to leave a [00:13:04] lot of freedom for people to uh recreate basically their own assumptions if they want rather than doing them uh for uh for them and then ending out the data that is already insanely pre-processed because that was our frustration with [00:13:19] the existing data. And so what we wanted is an infrastructure that let everyone basically uh did uh their own exploration. And so that's uh coming nicely to my third point which is the all uh goal of this exercise as well is [00:13:32] to be open source and collaborative with all the micro data being made publicly available to uh all researchers and in that respect we're going to uh put together a release of uh this new uh [00:13:45] data infrastructure mus plus uh with the center for tu research uh in the beginning of the fall for all of you to uh to get access to. Not all countries are going to be available immediately. [00:13:57] But that's uh that's the first step. Okay, good. So that's uh the coverage. [00:14:02] Uh something that I wanted to show you as well because I think it's important when it comes to time use the difficulty is uh it's very hard to know what to validate this information against. uh [00:14:15] compared to all the kind of hard data that we uh especially me as a public economist I'm used to admin data tax data that you can really validate at the macro level against national accounts aggregates and it's very hard uh to know [00:14:30] what is the actual quality of the data that you're getting how representative is it the nice thing is Amori one of our co-authors has recently done an effort on trying to get information on uh time [00:14:42] spent in uh paid work, okay, from labor force surveys. And at least what we can do is see to what extent the information that we get on labor force surveys that we know are broadly representative are very consistent in the definition of [00:14:56] work correspond to what we see in time use data when we measure work. Okay. And as you can see, it lines up pretty well. [00:15:04] Meaning that the information that people recall in this diary about whether they work or not, uh it's um it's it's not bogus. uh it seems to have a bearing uh that you can validate uh against other [00:15:16] sources of of data. Okay, good. So now let me uh >> yes, >> if you plotted that relationship separately for low income and higher income countries, would would the slopes [00:15:30] be the same or >> the slope? Uh that's a good question. We haven't done this, I think. Um yeah, I We need to check as you know the difficulty in labor force surveys is to [00:15:44] reconstruct work hours. So this is what Amory has been trying to do again having in mind a close uh connection with national accounts the notion of what enters GDP and uh I don't know to what [00:15:59] extent we would be able to match it as as well uh that's that's that's a good point uh but actually if you we go back you see that there are a lot of so look at Bangladesh for instance Bangladesh is [00:16:10] not too bad uh down there it's not easy there are a bunch of low income countries that are not doing doing too bad uh in terms of um but Tanzania for instance as you can see [00:16:24] >> over this right that they're using >> okay now it's on um but national income [00:16:43] accounting right includes includes unpaid work on family farm for example. [00:16:48] So how is that can at least in some instance? So are we is those are excluded from the hours of paid work or >> no? So we're going to try to uh follow the exact same definitions. Uh [00:17:02] but actually it's nice because in time use data uh actually these are the um sources that are used by national statistics offices in low- inome countries to uh measure the time that is spent on uh unpaid work production of uh [00:17:17] goods for own use that's the way they call it uh to estimate GDP. So this is actually going to be pretty well captured in some sense here I would say that it's mostly the labor force survey that might not be as good because it's not the labor force survey that is the [00:17:29] main source. Uh so yeah but again this is what we can do at least in some context where we know that labor force survey is a good source we can validate against the time user it seems to be pretty consistent [00:17:42] good okay so now let me uh move on what I'd like to do is start with uh presenting to you five uh interesting stylized fact when it comes to time investment in children so the first fact [00:17:57] is that the arrival of children is really associated assiated with a very large increase in uh your active time. [00:18:04] So it's really an investment when kids arrive. Uh basically your times become incredibly more active. So what I'm showing you here is the overall active time defined as paid work including production of goods for own use. Okay, [00:18:19] domestic work and child care. Okay. uh and this is the aggregate of this at the level of the mother and the father as a function of event time when event time zero is the arrival of the first kid. [00:18:33] What you see is immediately when kids arrive you have a massive increase at the level of the couple of roughly four hours. Okay, that represent an increase in 20 25% of uh of of active time. And [00:18:45] the second interesting fact is just how persistent uh this uh effect is uh over the next 15 years. So yes uh it does take a toll on uh on on your time. Uh [00:19:00] why? Well, first of course because uh you need to take care of uh the uh children and so when you decompose this active time into paid work, unpaid domestic production of goods, foreign use and dependent child care, you see [00:19:14] that really immediately sorry immediately what's jumping up is checkare. Okay. But then what's interesting is that it has declining [00:19:27] trend throughout the next 50 years whereas increases a little bit less but as upward trend throughout the next 15 years is domestic work. Okay. So what's [00:19:41] interesting is you know 15 years after what is really driving basically the time investment is mostly uh the domestic work rather than the time in China. Yeah. [00:19:54] >> Sorry. >> Okay. Okay. [00:19:58] [clears throat] Uh okay. So what you're seeing is really that it's uh as much child care as domestic work that explain this massive increase in the time investments of of [00:20:09] parents. Okay. The uh counterpart of this is that if you have more active time your overall leisure time uh decreases. Okay. So at the level of the couple you have a decrease in leisure [00:20:22] time that you can again decompose into uh the basic leisure which is your sleep your eating your care but all the other things that you like to do in life entertainment active leisure social activities basically you need to reorganize on all these margins. Okay, [00:20:37] you have a decline in all these dimensions. A decline of roughly an hour of the uh basics but also of 30 minutes of active leisure, of social activities and so on and so forth. If you decompose [00:20:50] the basics, it's uh a decrease in sleep. You sleep much less when you have kids. [00:20:56] Don't know if you are aware. Uh but you also cut on many things. You eat a little bit more quickly. You uh take a shower a little bit more quickly and and and all these things. So that's the beauty of this type of data. We can really [00:21:09] decompose quite granularly all these all these effects. Okay. Now the second fact that I wanted to uh show you um which was a bit of a surprise uh it's just how much child investment rises over the [00:21:24] course of development. Okay. So here let me be precise. It's really the active child care primary or secondary but not counting coresent. Okay. that's going to [00:21:35] be important. So what you're seeing is uh at very low level of development take subsarian African countries that we have in our uh sample uh you are going to [00:21:47] have an increase right away on impact at the arriv of kids but then it declines quite uh quickly and stays uh really uh really flat around an hour of child care uh for the next 15 years. If you compare [00:22:02] that to say uh Europe uh it increases much more the level of the couple almost six hours uh on impact. Okay. And then also declines much more slowly. Okay. Uh and so what that means is if you look at [00:22:15] this through the lens of a very simple graph like this where on the x-axis I have GDP per capita on a log scale. [00:22:22] Okay. you see that there is really this massive increase in the average parental child care time across GDP per capita an increase of roughly three over the course of of development. So that again I think is really important. Yes. [00:22:35] >> Do you have an explanation for Latin America that sort of doesn't >> Yes. So you'll see I think uh there is potentially some issues with the way child care is recorded in Latin America [00:22:50] because we have a lot of recall. In particular, Brazil is a big recall country. Uh and we know that in recall uh there is typically inflation. So basically you have to constrain those recall so that the sum of the activities [00:23:03] doesn't exceed 24 hours a day. Uh so that might be part of the explanation. [00:23:08] Uh but again as you are going to see there is also something quite specific about just how domestic the overall time of women in Latin America is. They're spending much more time at home. Okay. [00:23:20] And you're going to see that it fits actually quite well. I'll come back to that. So they are actually spending very very uh much larger fraction of their time at home than than women at lower [00:23:32] and higher level of development. Yes. I was wondering uh about the correlation with the number of kids and sort of what you make of those estimates >> taking that into account. I'm I'm a little bit surprised I guess. [00:23:46] >> Yes, absolutely. And sorry, can I add one to that because is that related to reconciling with the result by Golin Doug Golin and his co-authors that find kind of an inverted U for average time [00:24:00] spent in child care across country that it's like >> so the the difficulty I would say kind of flat so clearly so to be clear what Claudia mentioned is that there is this [00:24:14] paper go they are trying to of measure of child care but the measure of child care that they are showing are uh not conditioning on being a parent that's average childare per capita. So clearly there uh just everything that has to do [00:24:29] with demographic uh trend are going to be impact when you have kids the number of people who have kids how many kids they have is going to potentially impact these things whereas here it's really conditional on being a parent okay what happens to the time that you uh invest [00:24:42] it is true that it might be surprising that you have so many kids and basically so you have roughly three to four times more kids when you're in Sudan than let's say in the US okay and yet the total amount that you spend in checkare [00:24:55] is much lower. There's going to be a bit of an explanation to this. I'll come back in uh literally two minutes uh because it depends on who's doing the child care as well. Okay. [00:25:13] Um to what extent is this driven by the fact that in higher income countries you're more likely to have to you have couples? Uh because here you're recording couple. So couple average is [00:25:25] average across couples or is within couple average? [00:25:29] >> It's the sum. It's the sum of both the time of the father and the time of the mother. Okay. If a father doesn't have a mother in the household, that's just going to be the father. But uh here in actually in the data uh the prevalence [00:25:44] of uh households where you have the father and the mother present is actually quite high at low level of of development actually. [00:25:52] >> Okay. Good. So fact three and really important to better understand what's going on with child care. We need to of course make a difference between being present with a child at home and [00:26:05] actively taking care of the child. I think it's really important if we want to think about you know the concrete activities that we're doing. So here again the beauty of the data that we have is that we can record for a lot of countries co-present. So here what we're [00:26:20] showing is very simple. It's the ratio of the reported time in childare to the co-present time with child for different countries. The first thing that jumps out is of course that it is uh you know [00:26:34] very different from one okay actually when you are at home with your kids you might be supervising them loosely but you do many other things. Okay. So basically on average uh for the entire [00:26:47] set of countries that we have it's roughly around 30% of the time that you are spending with your kids that you're spending in active checkare. The other thing that needs to be uh uh refined a bit but there seems to be a bit of a an increase uh in that fraction over the [00:27:02] course of GDP meaning that as uh GDP increases the uh time that you spent with your kids is more dense towards actually providing skill acquisition okay and we're going to actually see that yes [00:27:17] >> do you have any information about the siblings presence like because >> I'm coming to that in a second okay So before I do that, I just want to show you also from the parental side what we know about the activities in terms of [00:27:30] childare because child care you can decompose it into many different things such as the physical care that you're providing bathing feeding the kids but uh there are many other things that you do when you're doing childare educational support play leisure or the [00:27:44] types of childare. What's really interesting is that relatively low level of development, the activities of child care are massively tilted towards just the caring, the feeding, the bathing and so on and so forth. And that increases [00:27:57] quite a lot on impact. Okay. But then as kids become more independent with respect to bathing, feeding and things like this uh basically that uh decreases. Okay. And is quite stable when development kicks in. This is now [00:28:11] for India. You see the rise in other type of activities in childare such as educational support, play, leisure. When you look at the US, it's actually mostly what childcare is about. You do a lot of [00:28:24] time in play. Okay. You do a lot of education support and all other types of childare that are also quite tilted towards cognitive and non-cognitive skill acquisition. Yes. [00:28:34] >> So there are two ways to to think about these types of activities. is from the perspective of the couple, the parents, and then the other is from the perspective of the child who is investing in the child. And so if you [00:28:47] were to draw these graphs looking at hours per day by members of the household, then what would be the patterns across across different development taking grandparents or siblings? [00:29:02] >> That's this. Okay. So the other thing that we can do so unfortunately we cannot do it the exact way I think you and I are interested in it which is take children and have a full diary of the children and know basically what they receive in terms of of care and [00:29:16] activities. Okay. But we can do two types of things. The first thing that we can do is look at the total amount of child care that is being done in an economy. Okay. I know basically uh people report in my sample I'm doing childare and I might not necessarily be [00:29:31] the father or the mother of the child. I might be what we call an alo parent. An alop parent is somebody who is taking care of the child but is not biologically related to the well biologically the father or the mother of the child. Okay. So from this what we've [00:29:46] done is very simple. We look at the entirety of the um time spent in childare per capita and then we allocate it depending on whether it is done by alo parents or a parent okay and within [00:30:00] the alo parents we are splitting this between the old alo parents okay and the young alo parents who are essentially people less than 25 who don't have kids okay and what you are seeing is actually really interesting at low level of [00:30:15] development As we knew it takes a village. Okay. Alo parental care is really defining of our specy by the way because we are the only great apes uh to do cooperative breeding. Okay. So yes uh for quite some time in the uh [00:30:30] evolutionary history of our species uh children are being taken care of by a broad set of of alo parents and what's happening with development is that disappears. Okay. [00:30:42] And in particular that disappears because the young alo parents uh disappear. And why? Well, that goes back to the question on on the evolution of demography. You have much fewer uh older siblings that essentially take care of [00:30:56] you uh at lower level of developments. Uh and the other thing is basically uh it takes a village here because you live in a village. Okay. But when urbanization kicks in basically it becomes much more complicated to have a lot of other alo parents take care of [00:31:10] you because you're more spread out and the space of child care becomes really uh the home and so you have this kind of centering focusing of child care on parents rather than alo parents that is [00:31:22] quite defining I think of the evolution of uh childare investments over the course of development. Yes. [00:31:29] >> To me that this reflects a large part of the transition Because babysitter. [00:31:47] >> So let's get at this. That's what we're getting at. So that's the second thing that we can do in the absence of uh explicit diaries for all of these countries at the child level. Okay. What we can do is at least allocate specific [00:32:02] time that we observe okay in some activities. And the two types of uh time that we can allocate is one the type that you spend in education or in institutional care that we can do from a [00:32:15] bunch of statistics that exist in all of these countries and the child care that you receive by all these parents and other parents. And we're going to see basically how these evolve over the course of development. And the thing that is quite interesting is that they seem to be actually broadly complement. [00:32:28] Okay. So yes, clearly there is a rise in alo parenting from institutional care rather than your alo parents. Okay. But that doesn't substitute away from the parental time. Much to the contrary, what you're going to see is that [00:32:41] parental time increases as well as uh institutional uh time. So just to explain briefly how we went there. Oh, maybe I'll take that question before I'm curious. [00:33:17] Um [laughter] >> okay the short answer is we haven't and I agree that's something that we should definitely do. Um what I'd say is that [00:33:32] given the focus that we had that was really on understanding uh gender specialization uh really finally in terms of task allocation nothing can re really replace uh time use surveys here in terms of the consistency the availability across a [00:33:46] broad range of of but I agree that this type of source is really a nice uh way to complement the type of uh insights that we can get from from time use data and we should do more of it but the short answer is we haven't done anything with these If you speak without a microphone, the [00:34:06] millions of people out there will not hear you. [00:34:10] >> No, I just it seems like it also looks at this exactly who does it and how that evolves and how it relates to institutional care. So, I think it's a nice it could be a nice way of validating and then you could see what echoes and what is extending. [00:34:22] >> No, that's a good point. It's an excellent point. Okay, let me just maybe show you what we have which is so we quite painstakingly I must admit reconstructed uh the overall hours of uh [00:34:36] time spent in institutional care okay being defined as pre- primary care or primary or secondary education uh across all these countries and that's the average per week for every child uh aged [00:34:49] 0 to 14 years old. Okay. And the uh uh nice thing about this is we made really an effort to uh account for everything both the variation in enrollment but also the number of days that you spend [00:35:02] at school uh as well as the number of hours per day because of the variation in curriculum. Okay. What you are seeing is that echoes what you are saying. Yes, there is a rise in the time that the kids spent in education over the course [00:35:16] of development basically that doubles. Okay. uh the uh weekly time that they spend in education goes from essentially eight to 16 on average. Okay, 16 hours and that's as you can anticipate mostly driven from what happens very early on [00:35:30] in pre- primary care and also what happens for secondary education where enrollment also increases over the course of development. But then what we can do uh echoing what I was explaining is from the perspective of the child [00:35:44] think okay if I'm a child under three okay over the course of development how many hours per day I'm going to be spending receiving either child care from my parents or other alo parents or [00:35:56] from uh uh school uh and therefore kind of educational alo parenting and what you are seeing is that there is this massive increase in childare but that is actually echo showing a rise in the time that these uh kids also spend in [00:36:11] institutional care. So you don't see uh any form of substitution by which the rise in institutional care would substitute away from um the child care received from uh parents. That's true [00:36:24] for children under three. Even more so for children in pre- primary from 3 to five. Okay. For children in primary education here it's very flat. there is already quite a lot of uh time spent in [00:36:37] primary education over the full course of uh level of of GDP per capita but you still see that rise in childare and that's also true uh at higher level secondary education where there is clearly also a rise in the time that [00:36:51] spent kid that parents uh spend with their their older children. Okay, >> good. That's all driving them to soccer games. [00:37:01] >> Sorry I missed that. I said that was all driving them to soccer games. [00:37:07] >> Okay. Exactly. Yeah. Exactly. Um >> unclear that that's really good time spent with your child. [00:37:19] Yes and no. But so so >> yes yes yes and no. So what what what we can do uh because so that tells us that at a broad macro level the time in [00:37:33] institutional care is a broad complement to uh the time invested in your kids in terms of childare. Can we find this at a more micro level? The answer is is yes but mostly at younger ages. So here we [00:37:46] are looking at an individual level. Okay. uh how much the time that you spend in child care uh increases or decreases where you get access to more institutional care where we're exploiting the cross-country variation [00:38:00] the time structure of education by uh age and what you're seeing is uh that indeed at low um ages so in pre- primary education the fact of getting access to uh pre- primary institutional care is [00:38:15] actually met by a corresponding increase in childcare investment that's much less so at uh older ages and what's interesting is that it actually echoes [00:38:27] some micro studies that exploit quasa random variation in education I'm thinking for instance of the gelburn isn't paper on on head start that also show that when you get access to uh pre- primary institutional care the active time that you spend in activities such [00:38:42] as reading to your child and stuff like this does does increase okay so now Oh wow. Fact five. I'm going to really rush. Uh but the most important uh the [00:38:55] uh implications of parenthood are of course very gendered. So everything I showed you was at the couple level. If we split that for men and women, what do we see? Well, for men, not much changes in terms of their time allocation. [00:39:08] That's the the composition of their active time. Uh basically, they don't do more domestic work. They do a little bit more of dependent childare, but that's that's mostly it. For women, the arrival of the kid is a complete change in the [00:39:22] allocation of your daily time. Okay. With a massive increase in dependent child care, okay? Uh an increase in domestic work. Okay. And uh echoing what I was saying, basically domestic work is [00:39:34] as important in terms of uh the time constraint that it imposes as dependent childare. And uh this is essentially substituted against paid paid work. [00:39:47] Okay, good. So now in five minutes I'm really going to need to rush but I want to show you some important implications for gender inequality. So first on the anatomy okay as I was mentioning for women it's more of a burden in terms of [00:40:02] time constraints the overactive time of women increases that's globally okay in the world population by roughly two and a half hours and only one hour for men. [00:40:11] So the uh increase in uh time demands is uh as we kind of expected uh larger for for women. The nice thing and of course what that means is the counterpart is that of uh they experience a decrease in [00:40:25] leisure that is larger. Okay. The interesting thing about our data is that we can really dig into the granularity of things. What is taking so much in domestic time for instance? Well, it's essentially food preparation. Okay, it [00:40:38] takes time to uh cook food. Okay, but it's also but to a a slower a lower extent cleaning or or doing the laundry. [00:40:47] Okay. Uh the three important things that I want you to remember about uh children led specialization is first that it has significant effects on mother quality of time and well-being. Why? Well, because with this type of data, we can go from [00:41:00] activities to measures of attributes of time. For instance, we can ascribe a certain level of cognitive intensity or physicality to different activities. [00:41:09] Okay? And when we do that, we can show that the physicality of women's time increases globally by roughly 5% relative to men when kids arrive. So it's not only different things that you are doing, but of course the attributes of these times are different and that [00:41:22] has real implications for for well-being. The second thing about this specialization, it is very asymmetric. [00:41:29] Okay, when you look at task substitution, I'll go very quickly here, but happy to talk more about the methodology. What we find is when mothers increase their labor uh force uh time, their labor force participation, their time in paid work, that comes at [00:41:44] the expense of domestic work, childcare and leisure almost equally. But when you look at men, labor time only substitute for leisure when they do lessly less labor. It's not like they increase their [00:41:57] domestic time. And in terms of cross spouse substitution as well, what's interesting is we find that when fathers increase their labor uh supply time, okay, mother's domestic investment go up. So there is clearly a pattern of specialization, but the reverse is [00:42:11] untrue. Okay, if a mother increases her labor force time, the father doesn't increase his domestic time. Okay, so that's an important asymmetry. [00:42:21] And finally, specialization increases with development. How can we do this? [00:42:25] Well, we create a very simple synthetic measure of specialization by projecting your daily time in n dimensional space where n corresponds to the number of activities. Okay. So we project this. So [00:42:40] your day is a vector in that n dimensional space and then what we do is we compare the cosine similarity of this vector between the father and the mother. And what we can show is that the [00:42:52] similarity okay decreases much less at low level of development when kids arrive compared to high-income countries. Okay, roughly the the magnitude is a 5% decrease in similarity. So specialization driven by [00:43:05] kids in low- inome countries but 20% in high- inome countries. And the final set of really important descriptives that I wanted to show you uh are these ones with respect to uh development. So this is same type of graph scatter plot of uh [00:43:20] GDP per capita on the x-axis and a bunch of different uh average daily activities taken by fathers and mothers. So here basically this is number of hours in paid work and we're essentially known [00:43:33] territory. We're finding the U shape of mother paid work. Okay, so that's a nice basically validation. But what's interesting is that this uh so sorry this uh uh large increase uh in uh the [00:43:48] time spent on paid work for women is not coming from a decrease in their time spent in childare because this is increasing both for men and for women. [00:43:57] Okay. And there is still quite the same level of specialization in terms of who's doing the childare. So what that means is that to explain the U-shape and in particular the large increase in labor force participation sorry you [00:44:09] really need that domestic time decreases and this is what we're seeing it's an extremely pronounced inverse U shape okay meaning that uh the uh increase in labor force participation of women has [00:44:23] to be substituted against domestic work so at the end of the paper what we're doing is using these moments to discipline a simple model of task allocation and uh get some insights out [00:44:36] of this. I'll conclude on this. Basically, what we're finding is uh that it's impossible to explain the level of specialization that you are seeing without a rising value of childare. [00:44:48] It's impossible to explain the level of specialization without extremely strong comparative advantages. But these comparative advantages are so strong that they can only be rationalized by uh gender stereotypes. And we are finding indeed a strong correlation with gender [00:45:02] norms uh in the data. Just to give you a sense to get the level of specialization between the time spent in childcare for men and women, you would need in this simple model of specialization that the [00:45:16] time input of a woman in childare is five time more productive than the time input of a man. Okay. And when it comes to domestic time, it's seven to eight. [00:45:26] Okay. And also it varies a lot across level of development. [00:45:29] you know primaasia it's impossible to rationalize this with simple stories of biological comparative advantages or or stuff like this. Okay. And the final thing which is really important is that you really need these technology and market substitutes. And what we show in [00:45:43] the data is uh that uh we measure appliances. Okay. Appliances increases a lot over the course of development. And when you do a statistical decomposition just through the uh rise in appliances, [00:45:56] you can uh explain 30% of uh the decrease in domestic time over the course of development. Thank you so much.