Penn Calendar Penn A-Z School of Arts and Sciences University of Pennsylvania

Dropping Out, Being Pushed Out, or Can’t Get In? Decoding Declining Labor Force Participation of Indian Women

in partnership with the South Asia Center

Ashwini Deshpande
Professor of Economics and Founding Director, Centre for Economic Data and Analysis (CEDA), Ashoka University
Thursday, December 2, 2021 - 12:00
A Virtual CASI Seminar via Zoom — 12 noon EST | 10:30pm IST




(English captions & Hindi subtitles available)

About the Seminar:
The stubbornly low and declining level of labor force participation rate (LFPR) of Indian women has prompted a great deal of attention with a focus on factors constraining women's labor supply. Using 12 rounds of a high frequency household panel survey, Prof. Deshpande demonstrates volatility in Indian women's labor market engagement as they exit and (re)enter the labor force multiple times over short periods for reasons unrelated to marriage, childbirth, or change in household income. She demonstrates how these frequent transitions exacerbate the issue of measurement of female LFPR. Women elsewhere in the world face a "motherhood penalty" in the form of adverse labor market outcomes after the first childbirth. She evaluates the motherhood penalty in the Indian context and finds that mothers with new children have a lower base level of LFPR, but there is no sharp decline around the time of childbirth. Blinder-Oaxaca decomposition of determinants of female LFPR suggests that none of the total fall (10 percentage points) in her study period is explained by a change in supply-side demographic characteristics. She suggests that frequent transitions, as well as fall in LFPR, are consistent with the demand-side constraints, viz., that women's participation is falling due to the unavailability of steady gainful employment. The high unemployment rate and industry-wise composition of total employment provide suggestive evidence that women's participation is falling as women are likely to be displaced from employment by male workers. She shows that women's employment is likely to suffer more than men's due to negative economic shocks, as was seen during the fallout of demonetization of 86 percent of Indian currency in 2016. Her analysis contests the prominent narrative that women are voluntarily dropping out of the labor force due to an increase in household income and conservative social norms. Her results suggest that India needs to focus more on creating jobs for women to retain them in the labor force.

About the Speaker:
Ashwini Deshpande is Professor of Economics and the Founding Director of the Centre for Economic Data and Analysis (CEDA) at Ashoka University. Her Ph.D. and early publications have been on the international debt crisis of the 1980s. Subsequently, she has been working on the economics of discrimination and affirmative action, with a focus on caste and gender in India. She is the author of Grammar of Caste: Economic Discrimination in Contemporary India (Oxford University Press, New Delhi, 2011) and Affirmative Action in India (Oxford University Press, New Delhi, Oxford India Short Introductions Series, 2013). She is the editor, along with William Darity, Jr., of Boundaries of Clan and Color: Transnational Comparisons of Inter-Group Disparity (Routledge, London, 2003); Globalization and Development: A Handbook of New Perspectives (Oxford University Press, New Delhi, 2007); Capital Without Borders: Challenges to Development (Anthem Press, UK, 2010), and Global Economic Crisis and the Developing World, with Keith Nurse, (Routledge, London, 2012). She received the EXIM Bank Award for outstanding dissertation (now called the IERA Award) in 1994 and the 2007 VKRV Rao Award for Indian economists under 45.

FULL TRANSCRIPT:

Naveen Bharathi:

Welcome all to the CASI 2021 Fall Seminar Series. My name is Naveen Bharathi, I'm a postdoctoral scholar at CASI. Along with my colleagues, moderate this series. Today, I'm delighted to welcome Professor Ashwini Deshpande, professor of economics and the founding director of the Center for Economic Data and Analysis at Ashoka University. Today's talk is in partnership with UPenn South Asia Center. Professor Deshpande has been working on the economics of discrimination and affirmative action with a focus on cost and gender in India. She has published extensively in leading scholarly journals. She's the author of Grammar of Caste: Economic Discrimination in Contemporary India, Oxford University Press, New Delhi, and Affirmative Action in India, Oxford India Short Introductions Series. She's the editor of Boundaries of Clan and Color: Transnational Comparisons of Inter-Group Disparity, and Globalization and Development: A Handbook of New Perspectives, Capital Without Borders: Challenges to Development, Global Economic Crisis and the Developing World, et cetera.

She regularly contributes to leading newspapers in India and abroad, the most recent being at the Op-Ed on the recent round of NFHS in the Hindu. She has received EXIM Bank Award for outstanding dissertation, and in 2007, VKRV Rao Award for Indian economists under 45. Today's seminar, she will be talking about Decoding Declining Labour Force Participation of Indian Women. This is her work with Jitendra Singh, who is also a grad student at Ashoka University.

The stubbornly low and declining level of labor force participation rate, LFPR of Indian women has prompted a great deal of attention with a focus on factors, constraining women's labor supply. Using 12 rounds of a high frequency household panel survey, Prof. Deshpande demonstrates volatility in Indian women's labor market engagement as they exit and reenter the labor force multiple times. Her results suggest that India needs to focus more on creating jobs for women to retain them in labor force. Prof. Deshpande will present for 30 minutes, and we'll have questions from the audience for around 30 minutes.

Please keep your questions brief and to the point, so that we can get to ask many questions as possible, and apologies in advance if I can't get to everyone. So if you have questions at the end, could you please use the chat box to send them directly to me, Naveen Bharathi, and I'll call on you to pose your questions directly to Prof. Deshpande. Finally, please be mindful about muting your mics through the duration of the presentation, and also please remember that you cannot record this presentation without prior permission from the presenter. With this, I'm going to hand over the mic to Prof. Deshpande. Thanks Prof. Deshpande for taking time and being here with us today.

Ashwini Deshpande:

Thank you so much, Naveen, and thank you Tariq and Naveen for inviting me to this seminar series, which is absolutely exciting, and so I'm thrilled to be here. Also, thrilled to see friends in the audience. Thank you so much for joining. I'll get right down to it because there's a lot in the paper and I'm going to not be able to anyway, present all the results from this paper. This is a joint book with Jitendra Singh, my co-author, who's also here. And if there is some immediate question that you have for us, please type it in the chat box and Jitendra will answer it as I'm speaking. So with that, let me start.

The issue, as Naveen already outlined in his introduction, is the fact that Indian female labor force participation rate is low and declining over the last two decades. And the issue here that I want to point out is that when we think of the problem as labor force participation rate, which is really a binary indicator, which is either somebody is in the labor force or out of it, implicitly, it poses the issue as one of labor supply, which actually it's not, it just measures participation. And for women, especially what happens is that women typically tend to report lower rates of open unemployment than men.

So in other words, they might be looking for work, which means that they should be counted in the labor force. They might just say to a surveyor that they're not working and that's it. And so it's often, especially in developing countries, often might get recorded as not being in the labor force, whereas, if they had said that they were unemployed and looking for work, they would get counted as being in the labor force. So not working for women often gets recorded as not being in the labor force, and implicitly, it's seen as a voluntary exit from the labor force, which is often we see that in India.

Now, when you look at the literature on the question of women not being in the labor force, and I'm going to qualify that with several qualifiers later on, as I go along. What are known as supply side explanations dominate the literature, which is that there must be something that constrains women's labor supply, they get married and they can't work. They become mothers and they can't work. They might be conservative social norms that prevent them from working either communities or their husbands, or just themselves internalizing social norms. They don't think that it's right for them to go out and work. There might be prevalence of sexual violence on the way to work, at the workplace that might prevent women from working.

There's also a stigma of being seen as a working woman. In other words, it's seen as an indicator that your husband can't provide for you, and therefore you're compelled to go out and work. And of course there are demands of reproductive labor inside the household, which means domestic chore, reproductive tasks, care giving tasks, et cetera. So all of these factors are what broadly economists will classify as supply side explanations. And when we look at the literature on women's labor force participation rates in India, supply side explanations dominate the literature, and each of these issues are important undoubtedly.

Myself I have worked on the demands of reproductive labor and they define women's lives in many, many critical ways. So there's no denying the fact that all of these issues are important and they exist. The question is, do they result in a lower labor supply? They might. So that's a question that we need to ask ourselves. And one thing for us to think about when we think of data on women's participation in the labor market, there are large data sets for India called the National Sample Survey, and there are repeated rounds of the National Sample Survey. I have done some work using a primary survey with Naila Kabeer, and there are other independent surveys.

All these surveys document Indian women's willingness to work, if work was made available or near their homes. I should qualify by saying that when I say work, I mean, paid employment, because women are working 24/7. So it's never the case of women are not working. I'm using it in the form that economists use it, which is paid employment. And so what this data highlight is that there's massive unmet demand for work on the part of women. In other words, they want to work, but they can't find work that's compatible with a whole lot of constraints that they're facing. And in fact, my paper with Naila Kabeer show that it's a predominant responsibility of reproductive labour, domestic chores. That's what constraints women's labor supply.

Not that that's a full explanation of labor force participation rates, but when you think of supply side constraints, more than any of those other constraints that I mentioned, it's demands of reproductive labor, and so we call our paper, Norms That Matter. Because of the demands of reproductive labor, women are unable to access work if it's very far away from their home, because there's no transportation, they have to get back and get all the work done or do it themselves, and it makes it difficult to access work, provided work is available. And we'll come to that point a little bit later.

Now there are very excellent data sets for developed countries that show that women transition in and out of the labor force more than once in their lifetimes. So often around the time of childbirth, a woman might either completely withdraw from the labor market or partially withdraw and then reenter at another time. But because there is panel data, longitudinal data available, you can see these patterns. And one of the questions that we asked ourselves, Jitendra and I, when we got this high frequency panel data, is could Indian women also be exhibiting frequent transitions in and out of the labor force? And if yes, what might those be due to? And this question has not been answered or even asked so far actually before our paper that's because all the other data set that I mentioned are not panel data.

So you don't observe the entry and exit of individual women in and out of the labor force. What might be those factors that account for frequent transitions? Should transitions actually occur? Could there be supply side? Would there be demand side? Of course, they have to be both. And supply and demand side factors are interlinked in a way that I'll talk about a little bit later. But our interest was in knowing which factors are more amenable to policy interventions. If the labor force participation rate is low, what can policy do to change that scenario? And so that's really the motivation for this paper.

We use a high frequency panel data that comes from a private data source, Center for Monitoring Indian Economy. And they have this Consumer Pyramids Household Survey data, and we use four years of data from January 2016 to December 2019. And so all of it is pre-COVID. In other words, it's not marked by this unusual shock that all economies received in the world. And this data is collected three times in a year. So for four years, we have 12 waves of data. In terms of the employment status, individuals are asked about their employment status, and they're classified into four categories, employed; unemployed, willing, and looking for work; unemployed, willing, not looking for work; unemployed, not willing, not looking for work.

Now, what we did was employed is straightforward, which is you are employed. What we did was we took number two and number three, we club them together and called these individuals unemployed. And the reason for that is the following, is that the CMIE data, the Consumer Pyramids Household Survey data, asked about employment status yesterday. So it's equivalent to the daily status if people here are familiar with NSS data. And I discussed this with the CEO of CMIE, Mahesh Vyas, and he mentioned that even if somebody was not actively looking for employment yesterday, which would put them in category three, but if they were unemployed and willing to work, it's safe to classify them as unemployed.

All it means is that yesterday they didn't put any effort in terms of looking for work. So we classify categories two and three as unemployed, category one as employed and category four as out of the labor force. Now for everyone, but especially for women, these boundaries are a little bit fuzzy, and I'll talk about that in a little bit, but however, we need to have a starting point for classification. And by this classification, categories one, two and three are individuals in the labor force, either employed or unemployed and category four are individuals outside the labor force, OLF.

So this is what the trend for these categories looks like for the period that we are studying, and this is urban and rural women. And so if you look at the green line, which shows the female labor force participation rates for urban and rural India, according to CMIE, you see a drop in both labor force participation rates in the first about four weeks. For men... Okay. Before that I have these numbers here. So female labor force participation rates declined from 22% to 12.8% for rule and to 11% in urban India, according to CMIE data. And these numbers are different from NSS. If you go back to the graph, you see here that the decline in labor force participation rates really is mostly accounted for by a decline in unemployed women, right?

And so we have something to say about that a little bit later. And so is it that it's discouraged workers, that women who were classified as unemployed then eventually said, "Okay, we're not going to get work." And so they get classified in category number four, which might show a discouraged worker situation, where you think that looking for work there's no point, because you're not going to get work. Male LFPRs also declined, but the magnitude is much lower. It's just five percentage points, and the decline in male labor force participation rates is mainly due to a decline in employed men, as opposed to women, you see a decline really in unemployed women.

Now, notice that here in CMIE, you find a declining labor force participation rates, both for rural and for urban, whereas, in NSS the decline has been predominantly for rural. These are the male figures, and I already talked about the trends in this figure. Now, as I said before, our interest was in examining transitions, entry in and out of the labor force, and because there's a high frequency panel, which is an individual is interviewed three times in one year, if they're entering and exiting the labor force between waves, this shows a very high degree of entry and exit. And there must be some immediate factors that might be leading to these transitions.

So we define entry into the labor force, as a transition from the OLF status, which is out of the labor force into the labor force. So if an individual was out of the labor force in the previous period and is in the labor force in the next period, that individual would be classified as making an entry into the labor force. And similarly vice versa is exit, and if you were in the labor force in period T, and you were out of the labor force in period T plus one that would classify as an exit from the labor force.

Now, when we calculate this, we can calculate exit rates by wave and gender, and similarly entry rates. So for example, this is a graph that shows exit rates, and how does one this graph? So if you look at the first two bars for men and women, you see the height of the female bar is 0.22. So what that means is female labor force participation rate was 22% in January to April, 2016, of this, 0.79 of 0.22, exited the labor force. So the red bar shows you exits every wave of the women who are in the labor force.

So you see that in every wave, both for men and women, you see exits from the labor force. And the quantum of entry, exits varies by gender and by class, but similarly, you also see entry into the labor force in every round. And the way to read this graph is exactly the same as you read the earlier one. So for example, the first wave, which is January to April 2016, you see, 78% of women were out of the labor force, okay? Of them 9%, which is 0.07 of 0.78. 9% joined the labor force when they were observed in the next wave. And so you see that just as women are exiting the labor force in every way, there are women entering the labor force in the next wave as well. So basically you see a volatile situation as far as women's entry and exit is concerned and is far more volatile than that for men.

So the average exit rate for women is about 30%. It means about 30% of women, one in the labor force in a given wave, leave in each period and 70% remain. Exit rate for women is about six times higher than that for men. So 4% out of the labor force women join the labor force in the next period. Entry rate for men is four times higher than that for women. So men are less likely to exit and more likely to enter than women are, which is to be expected now. However, remember that when we are presenting these figures in proportion or percentage terms, remember that these percentages are being calculated on vastly different base. So if you remember, out of the labor force, women in the first round was 78%. So when we talk about 7% or 9% of that, you are really talking about 9% of 78% of women.

So the numbers look very different for men and women, because the percentages are calculated by gender on their respective basis that are different. Now, when you document the total number of transitions for rural and urban men and women, here's a picture that you see. So for example, rural men, about, I don't know, 67% or so men make zero transitions, which means that if they're in they stay in or if they're out, they stay out, et cetera. And then you see this distribution of the number of transitions. And for women, you see that women make... The total number of transitions that women make, either entry or exit is higher than that for men. So women's labor force participation is more volatile compared to men.

Now of the women who are in the labor force... I mean, of the individuals who are in the labor force, in at least one wave, right? So they were not out of the labor force the entire time, they were there at least one way, which makes up 95% of men, and about 44% of women. 70% men were in the labor force in all the 12 waves, and 20% made more than two transitions. So 44% of women were in the labor force at least one time, all right? And 5% of these were in the labor force the entire time, in all the 12 waves. And 95% made at least one transition, 34% made one, 36% made two, 25% made three or more transitions, okay? And our interpretation of these patterns is that labor markets in India are extremely informal. Jobs are very precarious. So you have some work today and you may not have work tomorrow.

And it really reflects widespread informalization and precarity, where men have to work out of compassion, but women work only when work is available, and obviously also when it's compatible with their other demands of reproductive labor and domestic chores. Now, because of what we showed you in the beginning, a question could be asked, is this a real shift in women's work status, or is it just change in reporting? So as I talked about earlier, the division between out of the labor force and unemployed women who'd be classified in the labor force is fuzzy, because as I said earlier, a woman might classify herself as a out of the labor force when she's unable to get a job. And so could it be that what we record as transitions is just changes in self reporting?

So nothing is changing on the ground. It's just in one way, a woman is classified as in the labor force, in another wave, she's classified as out of the labor force, whereas, she's really not been working the entire time. Now, the answer to this question is no, because we also examine changes between just employed and everybody else is classified as not employed, and we examine transitions as well, using these just a binary classification, and we find trends similar to labor force participation transitions. So in each period, roughly about 3% of not employed women are employed in the next wave, and this not employed includes what we had classified as out of the labor force earlier. And more than 18% of employed women are not employed in the next wave. So employed is actually working for pay. So the conclusion from all of this is that women are indeed frequently joining and leaving the workforce and frequently joining and leaving employment.

Could this be because of just seasonal agricultural demands? Not really, because we don't see any sharp spikes or dips in monthly data, and certainly not at the same time of the year over the years. So what do these frequent transitions indicate? Is it change in self-reporting? No. Is it seasonal spikes and dips? No. One explanation that sounds compelling is that Indian women's attachment to labor markets might be low. They really don't want to work. And so they work when they really have to, but then they would rather get out and not work if they could do that. And there the supplies and explanations might make no sense. So we tried to look at the... We looked at the cost breakup in terms of the numbers of transitions.

So we already said that 36% women make two transitions, 25% women make three transitions over four years. It's a very short period that we are looking at. Now, when we look at the cast or social identity breakup schedule cost, schedule drive, et cetera, we find 34 and 25% of SC and ST women respectively make three switches compared to 19% of upper caste women. So what this tells us is that groups that have historically had higher labor force participation rates of women, and groups that have fewer taboos on women's public visibility and mobility, are making higher number of switches. These groups also happen to be poorer and in more precarious employment. So it doesn't suggest a low attachment to the labor force. These groups are sufficiently poor that they cannot afford to not be attached to the labor force.

They can't go without work. And so what I suggest is that the nature of work availability has become fractured, and so you do something when you can do something, but otherwise you're really just not finding employment. So what this suggests is the fractured nature of work availability rather than low attachment to work. And this is the picture of what I just said in words. And so finally then we come to this aggregate picture where for women, we find that 44% of women in the sample were in the labor force at least one time. And this is the highest labor force participation rate that we've ever seen for India.

And this is not changing any definition of work. If anything, the definition of work that CMIE employs is stricter compared to NSS, and NSS in all of post independence India had never shown 44% labor force participation rate of women. So even with this extremely strict definition, without changing any way in which work is counted, you see that actual labor force participation rates of women is pretty high. It's just that they are in and out of the labor force, and when surveys like NSS happen, it's just a question of at what point of the woman's life are they being interviewed.

And so often they don't get captured even though actually they are workers. So this whole narrative of Indian women just withdrawing from labor force or being forced to withdraw from the labor force, because of conservative social norms, et cetera, et cetera, all that is just not compelling, especially when you see this data. And of course the average labor force participation rate is 14% and 15%, which is lower by the way than what you get in NSS, which also shows that the CMIE definition of work is stricter.

But even with the strict work, we see 44% of women at least once in the labor force in four years. So that's, the first headline result from our paper, which is the number of women in the labor force in India is far higher than what conventional data sets have been able to capture, and the reason for that is these frequent transitions. Then of course, you get into the next layer, which is what explains these frequent transitions. And as expected, we run estimation, we estimate regressions explaining entry and exit. And what we look at is two things. One is, could there be an adverse income shock to the household, which is the decline in the income of other household members, which is the household becomes poorer and that pushes women out into the labor force?

Or there is an increase in male unemployment in the household, that pushes women into the labor force. Or these a change in the total number of children below five years. These are sort of the convinced explanations of what might account for entry and exit in the labor force. So, because a shortage of time, I'm going to just skip over these results, and I'm going to summarize them later on, but this is the regression result with the probability of exit. All of these results are in our paper. So happy to share that, or you could write to us for the latest set of results.

And so what do we understand from these figures? Now, if you see from both these tables, you see that an increase in the household income increases the probability of exit. So that's the sort of what economists call the income effect. An increase in male unemployment lowers the probability of exit, all right? And number of children has no effect and we'll come to that. And you see the converse results for entry. Now, what is the total magnitude of these factors? So to what extent does the presence of male unemployment and a change in household members' income contribute to women's entry and exit?

So we did some back-of-the-envelope, rough calculation. Now in our data, the entry rate is about 4% for women, and the exit rate is about 30%. We showed that to you already, the presence of male unemployed members, depending on which specification of the regression we use increases the probability of entry by 3.6 to 5.5 percentage points and decreases the probability of exit by 9.6 to 21.6 percentage points. The probability that working age women have at least one male unemployed member in the household is 0.074. So 7.4%. So if you combine these two figures together, it appears that a maximum of 0.4% of the entry rate, and 1.6% of the exit rate in each period can be explained by the presence of unemployed males.

So that's the extent of how much this factor is important, to what extent is this factor important? And similarly, you find that 2% of the entry rate and 6% of the exit rate in each wave can be explained by the income effect. So of course, male unemployment and shocks to household income, do account to a certain extent for these entries and exists, but not really. I mean the overall effect is not very big. So something else is going on, that explains these transitions.

So of course, we turn to motherhood, right? Now, in the developed country context, you see quite a big literature on the motherhood penalty, which documents negative labor market outcomes within the form of labor force participation rates, or wages, earnings occupational levels, et cetera, et cetera, for women compared to men around the time of the birth of the child. So as the child is born and if you track the woman, and the parents actually. So you see a sharp decline in labor force outcomes for women, for mothers and an increase actually in labor force outcomes for men. So we thought we look at this as well from the CPHS data, and we looked at whether the addition of a new child in the household impacts a women's labor force participation rate.

Wages, earnings, et cetera, are harder to do, because we don't have very clear monthly data for wages. Also many women, also men are not in regular wage employment. It's harder to get the kind of data that you need literally months after the child is born. Now, because of the nature of the data set, we can identify a new child when either the woman is a head of the household or a spouse of the head of the household, because the biologically relationships are defined that way in family rosters and in Indian data.

And so we define a new child as a situation where the child is observed for the first time in the household at an age, less than 12 months. That's again, because often women go back to their natal families for childbirth, and they come into their marital families with a child of a certain number of months. And so that's the condition we've used, and so we find in our data set, there are 3,000 mothers with new child defined in this wave. And of course, the comparison group has to be women with no new child.

Now, economists always worry about what they call selection effects, which is that, could it be the case that women who decided not to have children or don't have small children have some other features or characteristics that are different from women who had new children, and therefore even in the absence of the child, their outcomes might have been different. And so we use a technique called entropy balancing to identify women with no new children who are similar to women with new children in all characteristics, other than the fact that one group has had a new child and the other group did not have a new child.

So we created a balanced group for comparison. Before I show you the results of the comparison, this is just what again, Indian data looks different from what you see in the developed countries. So when you look at the lower two, so first of course, the male female gap, the top two lines are men, and the bottom two lines are women. Now you see here, the green line, which is new mothers over time, and the orange line is the comparison group of women, that means women with no new children.

And you see really that the trend in their labor force participation rate is that their lives are parallel over the months. In other words, women who are new mothers just have a lower base level of labor force participation rate. We don't see any dramatic shifts over time. Of course, this graph is not the best way to see that, but we estimated it with two way fixed effects, stack difference and differences, et cetera. I can go into the technical details later on, but I'm going to go through this quickly, because I want to show you the stack difference in differences results, which basically what you do is we created 11 cohort data sets for each of the waves and each wave has no new mothers.

And you estimate a difference in differences within each cohort, and then you append the cohort. So that's a stacked difference in differences technique, and this is the estimation equation, et cetera. And just showing you one of the... Let me show you actually the balance panel. And so what this is showing you, is we have taken the reference period as minus three, which is roughly about 12 months before. So zero is childbirth, minus three is about a year before childbirth. So we feel like if a woman knows that she's pregnant, it's possible that from the time that she's either planning for the child or figures out that she's pregnant, you might are to observe some changes in labor force participation.

The tower is minus three, that's our reference period. And the comparison group is women with no new children, as we said before. And you find really that time period of minus three, you don't really see a big difference between the comparison group and the treatment group. And after the child is born, around the fourth or the fifth wave, so it's about yeah... Fourth or the fifth wave after the child is born. That's correct. You see an increase in labor force participation rates of new mothers relative to not new mothers. And so you do see some sort of reentry, but there's no dramatic difference around childbirth. And certainly it's not the case that childbirth is pushing women out of the labor force relative to the new, not women who don't have new children, do account for the frequent transitions of the kind that we see in the data set.

Right? So motherhood is not really the reason that you see those kinds of frequent transitions. We then do what is called a decomposition exercise where you are account for the labor force participation rates. And here, we now just take a binary variable, which is a woman is in the labor force or not in the labor force. And we try to account for all the factors that are individual and household demographic characteristics, that would normally account for the individual to be in the labor force. And we look at changes over that from our first wave to the last wave. And so the idea is that when... Remember that we are seeing an overall decline between. Despite these transitions, we are seeing an overall decline. So could it be that these factors have changed in a way, or some of these factors have changed in a way that have contributed to that decline?

So that's that's the other thing that we are also interested in. And so we do this decompositions, the standard technique, again, happy to talk about that in Q&A. And really we find that the explained component, which is all of these standard characteristics, that account for labor force participation rates. So when you think of the contribution of the explained part for rural or urban women, it's tiny, right? So it's really, they explain these component. There's no dramatic shift that's happening over those 12 waves. That is explaining the decline in the labor force participation rates or between the two periods.

So in terms of the switches, yes, male unemployment, income. Explains to a certain extent, but not really fully. Motherhood, not really, and change in these characteristics, not at all, actually. So if you look at the graphs, the explained part, you see it's pretty much on zero, more or less. So you don't really see any dramatic changes in characteristics that would account for women's withdrawal from the labor force, a voluntary withdrawal, and so this is urban. The time period that we consider also includes the year 2016, during which time India experienced, again, an unanticipated negative shock in the form of demonetization of India's currency.

And so just as you know, I have another paper on COVID, and where I find that a sudden negative shock disproportionately impacts women's labor force participation rate related to men. So a negative shock leads to a rise in unemployment, leads to a lowering of labor force participation rates, but it disproportionately impacts women. And so we try to see what is the result of the demonetization shock on female labor force participation rate relative to men's? So of course we have to remember that there were preexisting gaps in the labor force participation rates, as we just saw.

And everyone knows, I guess, who's here, what demonetization did, but basically overnight 86% of India's currency was just declared illegal tender. And in an economy that's predominantly works... Its informal works, largely on cash, this was a huge shock for small employers and workers and employers in the informal sector, and led to a massive contraction of economic activity.

And what you see is in an overall declining trend of female labor force participation rate, the 2016 demonetization led to a sharper decline, a women's labor force participation rate relative to men's. And you see that both in rural and urban areas, right? Estimation equation, there's results, I'll skip that. So finally, I mean, I'm probably out of time, but I'm just going to take a few, couple of more minutes to just wind up and show you some now sort of bigger picture stepping back.

So what we find in the paper, there's lots of results in the paper. What we find in the paper is none of the usual suspects that would explain both the decline trend, as well as the frequent transitions seem to be working. And so when you step back and you look at the growth in the working age population relative to a growth in employment, you see a widening gap around the time of 2004. That's around the time when female labor force participation rates started to decline in a big way. So what you'll find is that many more people are getting added to the workforce then are able to find jobs. So that's obviously the background to a decline in female labor force participation rates. Here are some numbers, here are differences in sectors, et cetera. A bulk of the action is in manufacturing.

I said I'll come back to that later. And what I want to highlight is that what these numbers show is it seems to us, and putting everything else that we've done earlier in the paper, it seems to us that it's really a problem of demand for female labor. It's demand for labor. In other words, there is a crisis of employment, total employment, but within that, there is an issue of demand for female labor, especially, right? And so what we want to highlight is what we are calling the demand side story, which is Indian women's labor force participation rate is marked by volatility and frequent transitions. There is a measurement issue. In other words, far more women are actually in the labor force than most surveys are able to capture. The reasons are unrelated to childbirth or change in household incomes, et cetera, et cetera.

When you look at the social group composition of transitions, you find groups that usually have lower taboos on women's public mobility or visibility, exhibit higher numbers of transitions. So it's not a low attachment to the labor market story. What about social norms and sexual violence? In this paper, we haven't talked about it, but Jitendra and I have looked at these numbers earlier, and we don't really find a very compelling story, but one thing that I want to point out is when you think of tertiary education in India, higher education, there's been a massive increase in the gross enrollment rate, particularly for women.

And I'm going to show you a graph just now. In addition to that, India's self-help group movement is massive and growing. So there's more than 70 billion women that are mobilized in self-help groups in rural India. And when you look at trends in female migration, even women who work migrate due to marriage, their husbands move and they move with them, end up also working.

So all of these stylized facts don't somehow fit in with this picture of, women are not willing to get out of their homes, are not willing to work, et cetera, because you would have seen that showing up in higher education. That's when women have to go to college, or they have to go to another city, or even within their city and villages, they have to go far away from home. They have to mingle with men. They have to be outside, not protected in their homes. And you see this is Indian gross enrollment ratio for higher education, and the bright orange line is India. And these are the other south Asian countries. And below, you see gross enrollment rate in higher education in India, and by 2018, '19, actually, female GER surpasses the male.

And that's really the story that you see in universities everywhere, India, where classrooms are more female than male. And so these are all the sector that you would see would get affected, if this massive rise in social conservatism were to constrain women from doing what they actually would like to do. Okay. I'm not going to repeat myself here. And so basically I think the story that we want to highlight is that it seems to us, that the story is about lack of sufficient demand for women's labor. Women's education level is increasing, total fertility rate is declining, maternal mortality rate has declined.

So all the factors that would be conducive for women to be in the labor force exist. If you don't see sufficient number of women in the labor force for a sufficiently long period of time, it has to be, we think, something to do with the fact that A, there is an overall crisis of jobs and B, even within that overall crisis, you see women getting particularly disadvantaged. We can no longer call it a jobless growth story because the last seven years there's been very little growth in the Indian economy. And so it's really growth has faulted, and so has the employment story, and that women are the sort of worst victims of this situation. So I'm going to stop here, and then happy to take questions. Sorry [crosstalk 00:46:59]-

Naveen Bharathi:

Thanks, Prof. Deshpande for such a fascinating presentation. We have a lot of questions, one from [Sneha Mani 00:47:10]. She's a grad student here. Sneha, can you unmute yourself and ask the questions?

Sneha:

Yeah. Thanks Naveen. Thank you for this really interesting talk. So in a way I had two related questions. One is, do we have any idea about what kind of jobs do people with more transitions have, even whether it's farm, non-farm, even in the rural sector? And the second question in a way of follow up is, would it be right to think about it as a negative association about income earned and the number of transition? More the number of transitions a woman has in a year, the amount of... Her earnings potential in a way would be negatively associated with it.

Ashwini Deshpande:

So on individual incomes, it's a little bit harder because in any way, the best of times Indian income data is a little patchy. And on CMIE, it's not so easy to translate individual incomes to match the income data set in other words, with the employment data set, because these are two separate data sets and Jitendra can give more details about the challenges of associating income data. So I would not be able to comment. [inaudible 00:48:20] comment on the second part, I don't think we can, right?

Jitendra Singh:

Sorry, what was the question about income?

Ashwini Deshpande:

That women who have higher income will exhibit lower transitions? They probably have more stable jobs, right? Probably the case [crosstalk 00:48:38]-

Jitendra Singh:

Yeah. About if you talk about like industry and by nature of occupation, we clearly see that the number of transition are less for women who are in urbans with service sector jobs and who are having in by collar jobs. And there is small, less number of tradition also in women who are like at the bottom level of education also, who are least educated. Otherwise, are mostly the men who are in like a post [inaudible 00:49:07], they are having a least number of transitions.

Naveen Bharathi:

Thanks, Jitendra. The next question, Bhumi can you unmute yourself and ask the question?

Bhumi:

Hi, Ashwini, thank you for presenting this so late, and for all this wealth of data. My question has to do with age cohorts. I'm really struck by the number of transitions for women compared to men, and I'm wondering if you see fewer transitions for women who are older versus women who are younger.

Ashwini Deshpande:

Yeah. So what [inaudible 00:49:40] our age. We've done it with all kinds of cohorts with age cohorts, et cetera, et cetera. My sense is that, that's what... I don't have the paper in front of me right now. My sense is that that's the case, right, Jitendra? That older women probably exhibit fewer transitions, but it's in the paper [inaudible 00:49:59]. There's a table by age cohorts. So yeah, I'll send you the link, and it's there.

Jitendra Singh:

So, I mean, if you segregate by entry and by exit, we'll see that obviously that exit is higher at 55 plus age loop, because obviously when [inaudible 00:50:15] permanently there, and also the exit is that higher in younger group, like between 15 to 25 where women are sometimes reporting whether they're in education or not, whether they're looking for a job. So there are exit higher rate younger and both at 55 plus age group. Similarly, [crosstalk 00:50:31]-

Ashwini Deshpande:

... with more like a permanent exit. It's not so much about switches. So switches are higher at the younger age groups. For sure. Yeah.

Naveen Bharathi:

The next question, we have similar question from Rashi and [Aditi 00:50:45]. So Rashi, can you unmute yourself, Aditi can add if Rashi has missed any part.

Rashi:

Thank you so much Professor Deshpande, for this talk. It was so interesting to hear more about transitions. I think, we don't get to hear about labor force participation in such a nuanced way. I had two questions. I think one was this question about social norms versus the income effects. I mean, you're looking at transitions and I was wondering about just maybe, do social norms play more of a role in terms of like getting women into the labor force in the first place, or do they play more of a role in certain regions? I don't know. I was just wondering is it possible that it's both but for different groups? And then related to that question, I also had a question about caste groups. And is it possible that social norms maybe play more of a role for upper-caste women as opposed to lower-caste women, where maybe the income effect doesn't dominate as much?

Ashwini Deshpande:

Yeah. So the thing about social norms is that they don't fluctuate so frequently over months. I mean, these are very sticky norms. All the social norms that get talked about in the literature are very sticky. You wouldn't see every three months or every year a fluctuation of the kind that one would expect to explain this kind of transition. So that's really the point. So the point is that despite... This is not to say that conservative norms don't exist or that conservative norms are not an issue. Of course, they are, right?

But they would not explain the transitions, and I don't think that they explain the decline, because the bulk of the decline started in 2004. So either there has to be some evidence. You don't see that in the crime records bureau, you don't see that anywhere, that there was a suddenly something spectacular that started to happen around 2004 that pushed social norms to account for 10 to 12, 20 percentage point drop in female labor force participation rate of the kind that you see depending on which dataset you're using.

So using the idea of social norms to account for the decline, I find really a bit problematic, because it sounds like a plausible story, but it's not so plausible when you actually think about it, right? That is not to say that conservative social norms may not marginally affect. So for a marginal person, there might be a marginal effect. That's also not to say that sexual harassment, for example, at the workplace would not affect women negatively, okay? But what is the evidence on women dropping out of work altogether because of sexual harassment at the workplace?

So I feel like there are two very important problems that we are getting discussed here, but by putting them together, some of the narrative it doesn't add up, right? And the final thing that I want to say is there was a time when economists didn't look at anything other than the hardcore economic explanations. So no consideration about culture, nothing about... And now I feel like sometimes we go to the other extreme that here is a very obvious economic explanation for the declining labor force participation rate. But somehow we don't want to accept that, because we feel there must be something cultural.

Why can we not be more open about the possibility that really there aren't enough jobs for people to work in, especially a population with rising education levels. Secondly, women's education levels have risen far more than men's over the precise time, where their female, the labor force participations have declined. So being in very low and rigor or just very basic farm environment, et cetera, that's not compatible with what women want to do, et cetera, et cetera. I mean, what I'm saying is that there could be some state forward economic explanations for this story, which I think we at least have to consider.

Naveen Bharathi:

Thanks, Professor Deshpande. I think that answered your question too, right? So next question we have from [Retika 00:55:17]. Retika, can you unmute yourself and ask the question.

Retika:

Sure. Thanks, Naveen, and thanks Professor Deshpande, for your great talk. I have a very basic question. So I was just wondering if you could tell us more about this demand problem, and while there are fewer jobs, are we saying then that women are less likely to be preferred to be employed in these jobs that are available? So is that because of the fact that women generally have lower skills, and is this some kind of feedback loop that's actually happening? So if you can just expand on your understanding of this. Thank you.

Ashwini Deshpande:

So I think this is the part that needs much more unpacking, but I think I can give you some explanations based upon my earlier work. So one is that I think there is a straightforward discrimination that occurs in the labor market, which earlier we used to talk about, but now somehow we've gone completely on the cultural side, and we don't talk about it. Yes, there is the discrimination by employers. So when we talk about cultural norms, maybe they're getting exhibited by employer discrimination of the kind that you find, for example, in the developed countries. And there's definitely evidence of wage discrimination, exactly in line with every country in the world where average wages of men are higher than average age wages for women in every occupation, right? And India is no exception to that.

So it's a much less social norm, but much more safe for business. So that's wage discrimination. The other is a huge constraint is transportation. So it's not as though there are zero jobs, obviously there are jobs, but the fact is that the jobs where the jobs are and where the women are, how do they get from place A to place B, and link to this, is what I was talking about earlier, which is norms that matter, which is I think the norms that we should really be focusing on is, the predominant responsibility of women to either do the domestic work themselves or get it done, right?

And this is the norm actually, and in South Asia, Indian, Pakistan in particular have the most unequal ratio of sharing of domestic chores. If you look at the time that women spend in India and Pakistan on domestic chores versus men [inaudible 00:57:38] is about 10 to 11 times higher, right? And so when you take this, combine it with constraints such as transportation, et cetera, et cetera, that just makes it very hard for women to access the jobs that might exist. And then of course, the larger problem of not sufficient jobs being available.

Of course, there's NREGA in rural areas and there's farming, and there are family enterprises, and women do work on those. But it's when they work on family enterprises, they don't get recorded as workers. Not every woman wants to do NREGA work. So in rural India, then the only option is [ASHA MGNM 00:58:23] workers, which is all women, right? So you do see women doing different kinds of work. And self-employment is basically a euphemism for doing something to get along, get by basically, and you see that kind of work as well.

In urban areas, childcare becomes a concern, which is not such a big constraint in rural area. So it's not really motherhood as such, but it's childcare. So where you leave your children, right? In addition to the transportation issue. So there are, and then good old fashioned discrimination that exists. So I think it's a combination of all of these factors that I think contribute to what we see as a demand side story.

Naveen Bharathi:

Tariq, do you want to go next?

Tariq Thachil:

Sure. Ashwini, thank you so much for the talk. It was really fascinating and nuanced. And I think you actually touched on one of the questions I was going to ask just in this answer right now, because I guess I was surprised to hear you say the come down so forcefully on social norms, not playing a role, but I think what you meant there, if I'm understanding correctly now is that it's not playing that much of a role in explaining the trend of declining participation, because to me, it seemed like the whole pattern of volatility that you document and the transitions in and out itself are produced by a particular set of social norms. So social norms are playing a role in that pattern, but you're saying that that pattern is not... That's a kind of general stock pattern.

It's not explaining the trend of declining labor force. So that's kind of the way in which you're seeing norms playing a role, and where they're playing a role, because the time you survey was exactly what I was thinking about as well, when you were talking. Anyway, so I think that's what I heard. If that's the question though, I mean, I guess, if we do that separation, I guess I'm thinking about this jobless growth, which I completely understand is like a natural place to go. But then what would we need the kind of jobs produced to look like? Because imagine if we even had a period of job growth, but if those jobs aren't matching these very restrictive needs that women are still operating with, even to gain employment, then I wonder if jobless growth is itself maybe too simple.

It's not just, you could have even a period of growing jobs, but if they're not matching you could still see it not attacking female labor of force participation. So that was my conceptual question. And I guess my empirical question, just a small one, I haven't worked with CMIE data, but I've seen some back and forth about them maybe being representative of better off households or not fully being as representative of sample, in some ways perhaps as NSS data. I don't know what your taking is on that, just as you've worked so intensively with it. So how do we think about CMIE data given that it is such a high frequency and useful resource potentially for others to use as well?

Ashwini Deshpande:

Yeah. So let me answer your first question first. So the thing is our director... The way in which the norms story gets told in the context of women's labor force participation, it's about A, permissions given to women to work outside the home by the community, husband, slash et cetera, et cetera, and stigmatization of women who work outside the home. And of course, there's straightforward religious explanation. So people will say in Muslim areas, et cetera, et cetera. So when I use the social norms as a term, I'm really talking to that brand of literature. That's one, but of course social norms matter, right? I mean, why would they not?

They matter in every country. Every country has its norms, and those matter. My point is that you can have two separate phenomena that are equally worthy of attention, but it doesn't automatically follow that one [inaudible 01:02:08], for example, high rates of sexual violence against women. Of course, India has it just as the UK has it, right? But why do you see that you don't see high rates of sexual violence against women in the UK leading to a lower labor force participation rate in the UK? It doesn't happen. Whereas, one in every three women in the UK report some form of violence, either from intimate partner or somewhere, right?

So it's pretty high. I'm saying UK, because that's a number that I remember, and it's true in other countries as well, but you don't see that resulting in a declining labor force participation rate. So what I'm trying to say is that there could be two very serious issues, but maybe the connections are more complex than we do. Secondly, when I talk about norms that matter being reproductive labor or domestic chores, in all countries that have grappled with this, and I'm thinking now of China, South Korea, women's labor force participation rate was not only a matter of just creating a job. That doesn't increase women's labor force participation rate.

There have to be changes made, that will enable women to access those jobs. That's why I was talking about what is amenable to policy and for example, provision of childcare or canteens that sell reasonable food. I mean, one of the big revolutions in China was the availability of affordable eating places at the workplace so that women didn't have to cook lunch for the family, to be able to go out and work, right? So that's a workplace measure, but it affects domestic labor. So there can be a number of workplace measures that affect women's ability to access work and change eventually the highly unequal social norms of sharing domestic labor, because how is this going to change?

So one view could be, one day the men have to get up and decide that they've just not shared enough and they'll start sharing, and then change of heart story, right? So we could wait for a hundred years and hearts may never change, but if women start to go out and access work, maybe that will compel a change over time. Even when you look at the US rates of labor force participation, male and female, it took about a hundred years before you see anything like a convergence coming in, right? And so historically, when you look at present day developed countries, which also had very similar norms, not very long back, what is it that shifted those norms?

My own understanding of... And that might be limited, I grant that, but my limited understanding is that for whatever reason compared either by the World War, flu epidemic, whatever it was, women entered the labor force, and that then led to concomitant changes both of at the workplace and inside homes, more gadgets started to be bought, et cetera. Whatever it might be, a bunch of things happened. That then eventually started to shift domestic norms inside the families as well, right? And so I think that's where India needs to go as well, which is get the women out of the house, and in the labor force. That's what's going to shift norms.

I don't think that there's going to be a change of heart story, where somehow miraculously, one day we wake up and we have a change in norms and... It didn't happen in COVID. I mean, my own paper on COVID, in April 2020, that one month of massive lockdown, when domestic helpers were not coming. That was the only month you see a slight increase in men's hours of work based on CMIE data. By the time you come to December, their time had dropped below what it was pre-pandemic. So that tiny silver line that appeared in April completely vanished by December, right?

Now on the question of CMIE versus NSS, I think we'll take that offline or something. It's a long answer, and I'm obviously not a big critic of CMIE because I use the data. So I think all data sets have some pluses and minuses, and so is CMIE is no different also from that sense.

Naveen Bharathi:

Thanks, Professor Deshpande. I think it's late for you, and thanks for joining us so late. With this, we come to an end of this seminar. Also, thanks Jitendra for joining us. Thanks everyone for joining the CASI Seminar Series. With this also, we end 2021 CASI Seminar Series. Please visit our website at casi@upenn.com. Sorry, CASI website to check for our future updates, future events, and the next seminar series would start in late January. And I also thank all my colleagues who made it possible to organize this seminar series, and also all the speakers who shared their work with us. Thanks a lot everyone and have a nice day ahead or good night there back in India. Thanks everyone.

Tariq Thachil:

Thanks, Ashwini. Thanks, Jitendra. Thanks for visiting.

Ashwini Deshpande:

Thanks, Naveen. Thank you for hosting us. Thanks, Tariq. Bye.