Redline Transfer Agreement with airSlate SignNow

Redline transfer agreement

okay hey everyone and welcome to uh this week's cpr vid webinar with orianna bandiera and who's gonna talk to us why poor people stay poor before we let orianna get going i'm just gonna give you some rules or some introductions for how we do it and maybe by now many of you have joined us so um the seminar is one hour and 15 minutes uh orianna is gonna present for one hour and then we're gonna have 15 minutes of q a at the end and at that point we're going to open up their mic for participants who wants to ask the questions and you can then raise your hand and and we will let you in to ask the question toriyama during the one hour that orianna is presenting if you have questions please write them in the q a section and we will make sure to couple up uh to take a couple of pauses during the talk and i will pass on these questions if they're clarifying uh to the speaker and um you know if you think there's a good question you're also free to vote up that question as well um and we um will um um we will we have done our homework and looked at ariana's slides so if there are some questions that we know is coming later in the in the talk we're just gonna move that to respondent questions um and orianna is gonna cover that later on uh so this is uh this is it for instructions this is the last vid cpr seminar for the fall and then we're gonna take a little bit of a break for the job market and let everyone attend our younger scholars who's gonna be having interviews and then we're resuming on march 2nd and then we're going to have every other week seminars again until the beginning of june and we're going to post the schedule for this uh later on so you're gonna see that as well but for now super happy to have you here ariana and the floor is yours thank you so much for inviting me let me just uh share the right file okay so uh so again thank you so much it's great to be here so this paper is co-authored with claire robin my trash and anton was actually here today so he'll also be helping out with questions in the chat or the q a i'm not sure how it works so let me see if this starts okay starting point is that if we look at the poor in poor countries lack of employment is not the issue most of the global poor work especially in the poorest countries so the problem with poverty is not as much the lack of jobs but rather the type of jobs that the poor do okay so the fact that the only by definition labor is the only endowment of the poor understanding the link between jobs and poverty is of the utmost importance so the question that we ask in this paper is a classic chicken and egg question which is basically whether the poor stay poor because they're only able to do bad jobs or whether they only do bad jobs because they're poor and as you can imagine the answer to this question is fundamental to understand which policies should be implemented to fight poverty and many of you will have already recognized the type of circular reasoning that's associated with the idea of poverty traps okay so poverty traps or multiple study states have a long history both in macro and micro development theory okay starting from rosenstein rodan in 1943 all the way to my my co-author and colleague matress katak a four years ago there is a long tradition of theory papers that look at the possibility of poverty trials now empirical investigations include calibrations with cross-country data structural approaches and micro studies with observational data okay a recent field experiments also relate to big push approaches suggest that these type of approaches work for reducing poverty so all of these points to something consistent with poverty traps but not quite the evidence of the existence of a poverty trap so just to make things precise to start with imagine that you have two people one on the higher dot and one on the lower dot if i put axis it might be more convenient let's say that on the horizontal axis you have assets today and on the vertical axis you have assets at some point in the future okay so one possibility for why people find themselves in two different equilibria is that one is less productive than the other okay so the lower person or lower country has a productivity parameter al and the higher country has a productivity parameter a h with a h larger than a l because that's one possibility the alternative is that they both have exactly the same productivity a common parameter a but the production function that they face is not concave everywhere rather he's got this s-shape or a discontinuity somewhere and what this can produce is two equilibria for two people who are otherwise identical so the l person in this case has started somewhere below the k hat point and ends up in the kl equilibrium and the other person must have started somewhere to the right and ends up at the high equilibrium now finally the answer is key for policy because in the first world in the concave world people with the same productivity will always reach the same steady state so it doesn't really matter where they start they will always climb out of poverty in this world the people who do not climb out of poverty is because they don't have the skills to do so so in this world anti-poverty policies must support consumption in the second world wealth of birth determines the steady state in this world there is no way out without a big push because if the transfers don't push you past that k hat threshold you're always gonna slide back into poverty okay the positive side of this is that a big enough transfer will eliminate the need for future transfers forevermore because once you are above the threshold on your own you will converge to the high steady state what we do in this paper is that we use the randomized control trial of a large asset transfer program in bangladesh and trace the effects of over 11 years to test directly for the existence of a poverty drop and then we estimate the structural model of occupational choice to back out the value of the missile location if any so some of you maybe many of you will have already seen the program because we have a previous paper in which we evaluate it this program takes place in the poorest region of bangladesh the sample originally is 23 000 households in 1300 villages in northern bangladesh this is the poorest area of one of the poorest countries and what we do is that we collect a five-year panel over the period of 11 years we actually also have a small fund survey in 2020 to see how they were coping with copied i can tell you that towards the end and we plan to go back to the field next year what let's see how the pandemic evolves next year or the year after next to keep following them so in 2007 we did a census and the census covered 100 000 households that formed the sampling frame for our first wave which was collected in 2007. and that was before the program was implemented the program which i will tell you in more details in the next slide was implemented right after the baseline and then we observed people two years four years seven years and 11 years later now in between 2011 2014 actually in 2013 the program was also implemented in the control villages because that was a randomized rollout it wasn't an exclusion now let me tell you some facts about these settings which will be important to understand how we can identify poverty traps here so the first is that the poor stay poor only in the control villages only three percent of poor households reach after four years median level of middle so the median of the middle class in terms of assets so you basically are born poor and stay poor this there is a hierarchy of jobs that's correlated with poverty the poor are casually employed in agriculture and domestic services whereas the rich are self-employed in livestock writing and land cultivation and this hierarchy means that better jobs require productive assets and productive assets is really what sets apart the rich and the poor the richer classes in our villages have on average 94 times as many productive assets as the poor and it's not just a matter of numbers so it's not a matter of like the poor households having one chicken and the rich household having a hundred chicken but rather there is a hierarchy of assets and the richer households own more expensive indivisible assets such as livestock and land so this is a descriptive table this is probably better to illustrate how striking the difference in asset is so the other lines on this graph are the factor of consumption so the rich consume food consumption twice as much as the poor uh savings and non-business assets so mostly household assets 30 times as big productive assets are about 94 times as big occupational choice reflects the difference in asset ownerships what we do here is that we plot the log of baseline productive assets on the horizontal axis and then the share of total hours spent in these different occupations and you see that the very richest people towards the right of this graph only do land cultivation and the very poorest people mostly do made casual labor in agriculture casual labor outside agriculture okay more assets as i was telling you earlier mean more expensive assets so people have a very low value of productive assets the people on the left of this graph mostly own poultry then as they get bit richer they start owning goats as they get richer they start owning cows and finally the very rich own land very rich is always very relative here poor people stay poor this is the first three waves and this is the total value of assets in a log in the first three waves and you see there's basically no change okay those who have little assets to start with have little assets four years later and so on and finally if we look at the distribution of productive assets in these villages this turns out to be by model this is essential i mean this is necessary but not sufficient for the existence of a poverty trap because it suggests that there are two equilibria one at log point six and the other one around log point one now how do we test for the existence of poverty trials the ultrapoo program was randomly allocated across villages actually across areas branch of the black offices which com comprise more than one village the beneficiaries are the poorest women in these villages the program transfers a cow and training on how to take care of the cow the value of the asset is enormous it's worth one year of per capita consumption which is about five times the average microloan that brak gives in these areas and contrary to the loan it doesn't have to be repaid within a week this is a grant it's a gift from black to the beneficiaries now entirely by chance i wish we could claim that we were clever enough to think about this but this is entirely by chance the program moves all the beneficiaries which is around 4000 people from the low mode precisely into the lowest density area of the distribution now the reason this is important is that if there is a poverty trap that is a threshold that threshold will be in that area because by definition the poverty threshold is an unstable equilibrium is repellent if you get close to it from the right you escape away from poverty if you get close to it from the left you fall back into poverty that also illustrates why it's so difficult to find evidence for poverty traps in observational data because by the mere fact that the equilibrium is unstable you never observe people in its neighborhood so you can't observe how people behave around that neighborhood luckily for us the program places a lot of people bang in the middle there this means that we can use pre-existing differences tiny differences in wealth a baseline to see how people respond to the transfer and basically plot out a transition equation this is after the transfer you see after the transfer in control people stay where they are because the poor stay poor in treatment which is the red line you get a bump of people that are pretty much half of the people who are at the low mold get moved to the low density point uh this shows that the argument that i was making earlier that shocks of this magnitude are very rare these are from control villages and these plots the share of households that experiences a change in low productive assets of each of these sizes so you see that the vast majority experience no change then most of them experience very small changes this is the value of the transfer the red line is the value of the transfer there is fewer than five percent of control households who experience a change of that magnitude so that is again to make the point that you very rarely observe people in the hollow part of the distribution because it doesn't happen by chance so poverty traps and differential productivity are observationally equivalent in steady state however they produce different transition equations because we shock the system we put people around the hollow part of the distribution we can estimate the transition a question that is how they respond after they've been put there an unnecessary condition for the poverty traps is that the transition equation is not concave we can test this using the fact that beneficiaries differ slightly in baseline assets i will have some identification slides later we'll exploit this to estimate the transitional question from 2007 and 2011 and then we will test the prediction of the poverty trap model following them 11 years down the line so we identify the threshold using the first four years of data and then we test whether the people who were above the threshold in 2011 behave differently in 2018 so identification is based on differences in initial assets that are tiny relative to the transfer but they're clearly not randomized okay they just happen to be there so we do two things uh to um allay concerns of uh of identification the first is that they could be just endogenous shocks so it could be that people who have slightly higher capital a baseline experience larger increase in capital subsequently because the placement of the problem is randomized we can use the control villages to test for the existence of these shocks that is we can compare two people with the same k0 in treatment and control and see whether the delta k is the same there could be an endogenous response to the program that is k0 could be correlated with how people respond to the program of course we can't use control for this because there is no programming control so we can use different sources of variation to compare people with the same k naught once we get there i'll tell you how we do this but we will be able to compare two people with exactly the same initial capital some in villages where that initial capital puts them past the threshold and some in villages where it doesn't i'm gonna interrupt you and ask a couple of uh clarifying or a couple of questions so the first one is uh what what do you mean by middle class or rich here in the in this paper so the wealth classes were defined by the villagers themselves in order to find the beneficiaries for the program brock run a rural participatory assessment exercise where i don't know how familiar people are with this but basically they get around all the households in the village and they get them to rank all the other households including themselves into four beings poor uh near poor middle class and upper class so this is all very relative if you look at the upper classes in these villages they're still quite poor by any standard okay and then uh two more questions so from uh frank schilbach could other constraints constraints in addition to the credit constraint be binding as well such as hope confidence attention entrepreneurial spirit and so on and this might be affected by assets or cash transfers in additional features of the ultrapurpose absolutely so we are completely in agreement with that we cannot test for these constraints but what we can tell is that if the asset transfer unlocks a poverty trap for sure the binding constraint was assets they could be additional constraints that strengthen this mechanism but we cannot we cannot identify those separately but the fact that well i can't anticipate the results but the fact that the acid relaxing the acid constraint solves the problem suggests that the acid constraint was binding in the first place and then um how should we think about the fact that a cow is a very specific type of asset so it's very liquid and no and not separable and it needs some types of skills to to complement it so in principle in a poverty trap model just giving cash should do the trick but is this then a lower bound effect or how do you think about this so we okay so kind of the best answer to this will be available in a couple of months hopefully because we run exactly the same program in pakistan where we randomized one arm where we gave them the choice to cash now everybody who had the choice of cash chose cash over the cow but then with the cash most of them bought a car so we will find out soon what the difference is i think probably that tells you that they don't value the training so much these people were given cow with training more in general of course i don't think that poverty traps everywhere in the world can be solved by giving people a cow but the equivalent of a cow is the capital that you cannot afford to buy so in most circumstances this would be human capital if you think about it the equivalent in a poor family who has a very smart daughter is being able to pay for university education thanks ariana and now you gave us a reason to continue look at this study uh later on as well very good so this is how the two transition equations look like so the one on the left is without the poverty trap the one on the right is with the poverty trap so on the left basically the higher your capital the less the accumulation of capital in the following period and on the right you have the s-shaped function now with a great fit of computer programming this is what the data looks like so let me blow that up this is the estimated transition equation from our data we have on the horizontal axis productive assets in 2007 on the vertical axis productive assets in 2011. so this transitional question is clearly not concave actually our colleagues at the lsc tatanya hamarova and javier hidalgo have developed a test for non-parametric shapes and we reject the null of concavity with p-value zero so screen not concave and he intersects the 45 degree line at 2.34 if you estimate it as a polynomial of degree 3 you get a very close number now you got 2.36 now k is unstable by definition the people below k hat lose k the people above k hat gain it but that you can see from the figure as well okay everywhere above the k hat the assets in 2011 are higher than the assets in 2007 and below is the other round in controlled villages if we try to estimate a transition equation this is what we get so that is basically the difficulty of identifying poverty traps with observational data you never observe people in that part of the threshold and thus in the part of the distribution where you have the threshold and you get one steady state which is the low mode of the distribution of productive assets that you saw earlier what does the difference in assets correspond to but here is the value the unit value of many assets and here is the difference between uh being above the threshold and below the threshold it corresponds roughly i think this is sorry i'm sure that this is wrong because it corresponds to the value of a rickshaw or applause so i think probably that is uh i know sorry that's the motorized one is the card the card or apply that is the difference that it makes to be above or below the threshold now it cannot be explained by shocks correlated with baseline capital that is if you run that same regression that is comparing capital in period t plus one to capital imperial t depending on whether you're above or below the threshold you see that in treatment villages people above the threshold accumulate capital in control villages there is no difference if anything is negative but is very small and precisely estimated you can put them in the same regression and do a different diff and you get the same exact result now identification so you might worry that this difference this small difference in productive assets is correlated with all sorts of things and in particular it might be correlated with the way people respond to the program so what we do to allay that concern is that we use the fact that in some villages having a cow gives more income than other villages and this is due to the fact that some villages are closer to markets than others so we estimate predicted earnings using all the other wealth classes and then we split the villages in two those were the predicted earnings from a cow are below the median and those where they are above the median now we estimate two transition equations now the red one is above the median and the blue one is below the median and you see that they have a very similar shape but the position of the threshold is different because intuitively in order to escape the trap you need a lower level of initial wealth if the earnings from the assets are higher but that also means that we can run the same regression putting k0 fixed effects so allowing everybody with the same level of wealth to be treated the same and then just use the differences in threshold rather than the differences in k0 to identify the property travel an alternative is to use individual variation and what we do here is that we use the dependency ratio which tells you not how much money you can make but how much money you can save out of the income that you derive from the cow because in order to accumulate assets from one period to the other you need to be able to save so a beneficiary who has five young children an elderly parent can save less than a beneficiary who has no elderly parents and two children so again we split people above and below the median and we find two thresholds when we use the variation in thresholds we get exactly the same number okay so you might argue in various ways and we have a list of arguments which i can put forward for why either the dependency ratios not exogenous or the differences in earnings is not exogenous but no matter what you do whether you use the variation in baseline capital or the variation due to the difference in earnings or the variation due to the difference in dependency ratio the effect of being above the threshold on capital growth is exactly the same it's always 30 percent and you can also use the different thresholds as placebos because being above the low threshold should have no effect because the low threshold does not apply to people with a high threshold and that's what we see here so now that we've identified this threshold i maybe stop here in case there are other questions otherwise i keep going keep going i think you can keep going yes good i'll stop you maybe later on if there's things coming up so now this is the threshold identified from the four-year change right from 2007 to 2011 we see this um s-shaped transition equation which finds this part of k hat where people above k hat accumulate and people below the human right now the true test of a poverty trap is what happens in the long run it is do the people who were above k hat in 2007 keep growing after 2011 and do the people below the threshold keep falling backward now this is the definitive coefficient of the difference in productive assets between people who are above the threshold and below the threshold in 2007 and different survey waves and you see that these are the results that we knew up to 2011 they were in accumulating more assets oh sorry that difference grows by 2018 that is 11 years after the transfer the difference is more than double and this is due to the composition of assets so they accumulate cows at the beginning but then they managed to save enough to buy land so this is due to the fact that the people above the threshold are on a spiral of savings and reinvestment up to the point that when we see them 11 years later some of them have both land the average gap in consumption increases interestingly it's negative at the beginning because of the difference in savings right but then it grows over time and this was at the beginning within the evaluation of this program people who use their 190 dollar a day poverty line we're arguing that the program does not work because it doesn't increase consumption well it turns out that it doesn't increase consumption because people are transiting to a new steady state and so they are saving instead of consuming this is the gap in hours worked it stays constant people work more and in particular the entire gap in hours worked comes from livestock and land cultivation so the important thing here to notice is that this is not just an accumulation of assets this is a poverty trap which is linked to occupational choice okay so what we observe here is the fact that having these assets or not having the assets makes the difference between staying in casual labor or being able to move into self-employment and that's what's driving all the results now over a period of uh 11 years there's many things that change notably people get older i can tell you that because i started working on this precisely 11 years what now is 14 years ago i was there when we did the baseline and over 11 years things change okay so the acid dynamics that we observe reflect two things reflect convergence through steady state as well as aging okay for people below the threshold both go in the same direction as people age they start accumulating assets and because they are below the threshold they accumulate assets but for people above the two forces go in opposite directions the fact that you're growing because you're above the attraction makes you accumulate the fact that you're growing old makes you accumulate so we can split the above sample into older and younger and those who are above the threshold and are younger to start with are twenty percentage point more likely to grow assets by year eleven and this is uh the below the threshold and above the threshold the different percentiles for for the youngest so far orianna maybe i'll i'll stop here with a couple of questions so one uh is on um you know a key assumption of these tests is that equilibria points do not change over time and so how do we know that that's the case because they didn't for the previous if you take the history of the village say the four years before as a sign of what would happen three years later we saw that if you draw the distribution of asset it will be the same year after year if you draw it in control i showed you at the very beginning i don't there go back all the way there in the slides because that would take me forever to come back but at the beginning i showed you a histogram of asset ownership in three waves which were four years apart of the ultrapoor the middle class and the reaching control villages and you could see that it was absolutely flat so in this type of economy there's not much changing once you account for crisis changes the real value of assets remains the same thank you so there are a couple of more questions so one is related to whether you see changes in internal migration so when houses buy land the children stand stay on the land and work on it and hence further increase the return to this investment so the children is a very important aspect of this because the intergenerational transmission of poverty you know these are very immobile societies so if you're born in poverty you stay in poverty but then if your mother gets a cow when you're just born when you can does it make a difference and that's what we're working on now actually and we observe i can give you a preview we observe an improvement in health of kids who were in intro or under the age of two when the mother was given the transfer there's no improvement in health for older children there is an improvement in schooling for children who were under the age of 11 when the transfer was given there is no improvement in schooling for older children and we are now following in the what happens once they join the labor market so this uh you just answered the second question with what which was actually about what is the effect on kids education and they were wondering if you know you see reduce reductions in investments for all for education for older kids but you're saying you don't say no no we don't see that because although there is a slight increase in the hours that the kids help with the with the animals this is very small so it doesn't preclude attending school thank you so we don't see the drop in education to the country we see an increase thanks so much and migration so they didn't answer the question question so these women by the way they're selected to take part in the program they're kind of stuck they're mostly women who don't have uh an able husband so and they have young children once they're treated when they're treated in 2007 so they're fairly immobile which also explains why attrition rates are extremely low in these samples we have like 10 attrition over there maybe 13 over the entire period so is not the right population to migrate you can imagine that an escape from poverty would include some urban migration and maybe the kids will but they're too young for for us to tell now so um another question is if you could achieve the same results with cheap loans such as the borga program and land rental as you did with the livestock so loans have the main issue of microloans at least of being tiny and you've seen here that you need a lot to go past the threshold the very last slide will compute the share of households we'll show you the share of households that can be brought past the threshold for every level of transfer so loans have the problem of size and the problem they have to be repaid very quickly there is a variant of this program where they give the cow as a mortgage so they pay the repay after one year i think they start repaying and they use the cow as a collateral but this is a version of the program that's given to people who are kind of richer than these extremely poor people i think that a very long-term loan contract could work in deciding even for the ultra poor but it would have to be very long term and i don't know how profitable i think probably not very profitable but relative to giving cows for free some of the costs could be recuperated i think thank you so we'll let you move on so the final thing that we do is to try to put some numbers on this so as i said the key result here is not as much that you give people a cow they get more cows but rather the fact that they change the job that they do they go from doing casual made and casual laborers in agriculture to be self-employed so what we do here is that we use a structural estimation of a model of occupational choice to estimate individual level productivity and cost of effort and then based on those determine the optimal occupations for each person each beneficiary here in the absence of capital constraints and that will allow us to quantify the misallocation baseline okay so the unique feature of this setting is that we observe the beneficiaries doing both jobs you very rarely observe this because normally people sort in the job that they're best at or if they face a constraint they keep facing that constraint but here we observe them a baseline where they could do nothing but wage labor so everybody's doing wage labor so we can estimate how productive we are in that and then because of a feature of the program basically brock asked them to keep the cow for two years and after two years they could do whatever they wanted with it so a bet at year two we observe almost everybody working with the cow so we can estimate productivity both in wage labor and in cow raring for everybody for every beneficiary then we do that and then we predict what the optimal choice should be and then we evaluate the model both using both the four-year data and the 11-year data as well so the idea is that assume that the ultrapoor had the same assets as the people who are at the upper mode and we use the model to estimate optimal occupation there is three types of occupation you can do wage work lifestyle craving or mixing the pink is the actual occupation that we observe and the green is the optimal that is what they should be doing given their parameters and you see that there is a massive mismatch here their optimal occupation is lifestyle crediting for 90 percent of them but everybody does wage work then we compute the payoff at the optimal occupation and compute the payoff for the actual occupation the difference between the two is the value of the missile location and our estimate suggests that it makes no sense not to give cows and this coming back to the question does the program pay off there must be a way of making these returns pay off because the total misallocation value is 15 million per annum the total cost of transfers that's needed to bring everybody above the threshold is 1 million paid once and that's it now the model suggests that everybody's misallocated and that the value of misallocation is way larger than the cost of eliminating the trap but i think the coolest thing i think about this model is that we can simulate all sorts of things that could destroy the effect of the program like a typical concern is now all these people are wearing cows so there is many more cows so the price of milk collapses and the effects are much smaller so we can see what happens when we allow for general equilibrium price effects we can see what happens if we double the weight rate will more people stay in wage labor or if we make wage labor less painful and sorry i thought that i had the numbers there basically nothing changes for in the following sense that in order for general equilibrium effects to kill the gains of the program so for the program actually to break even so that the gain the cost is 1 million and the gain is 1 million you need prices to fall by 90 so if the price of the product of the cow is the price effect is such that the price falls by 96 then the benefit equates the cost doubling the wage rate uh attracts a lot of people into wage labor so it wouldn't be 96 misallocated it would be maybe 60 percent misallocated but the value of the misallocation will be more or less the same because the distribution is very skewed so there are some people who are really very productive a cow raring who should do cow wedding and having the disutility of wage labor has the same effects now let me finish with that so let me take a question on this misallocation before i let you end so um one of the participants are saying that to say it's misallocation there must be an inefficiency so does the data say that poor people are more productive at raising the cow than rich people if they're given the very same asset so i don't know if they are more productive than the rich people but they clearly are more productive already in cows than being wage laborers and there's clearly a constraint here binding constraint which is the credit constraint they cannot buy cows on their own but once you give them a cow those who are above the threshold keep working with it and they do very well thanks thank you so this last slide i already introduced it earlier what it does is that basically for every transfer value here normalized as a share of average annual per capita consumption we compute the share of households which will be brought past the threshold by that transfer so what that means is that with the black transfer we get about 66 percent two thirds of households above the threshold and one third below now the typical microloan is about here its value is less than 20 percent of annual per capita consumption the share of households that will be brought above the threshold by that type of loan is 20 this is i think consistent with recent findings of the evaluation of microfinance that suggests that people who already have a business who are on the cusp of expanding but can't expand thanks to the micro loan they do much better but the vast majority of people they don't change their occupation at all once they're given a micro loan the value of the payment of andre on average is similar and these are the ultra poor programs that have been evaluated the pilots of the ultra put programs have been evaluated by abhijit and co-authors in the in the recent science publication and you see that they're all they're all poorer than black the one that's most similar in terms of results is also the one that's closest if you screwed honduras i had some implementation problem which is west bengal in india and this is chris plattman's uh transfer for micro entrepreneurs and we should put more because in the meantime there have been more programs being implemented of this nature but basically what this illustrates is that the classic approach microfinance cannot do anything to solve this problem because relative to the size of assets that people have to start with the micro loan is way too small to push people past the threshold so i think we've learned that people are not unable rather they lack the required capital and the key conclusion is that we really should focus on production and jobs that is to enable people to escape poverty on their own because many of them have the talent to do so and a one-off large transfer might be painful but it's way cheaper than keeping propping up consumption in the long run i'm done thanks eliana this is great so um there is questions in the chat but i suggest that we open the floor and people can ask questions themselves so if you want to ask questions please just raise your hand and we will unmute you and you can ask the question directly to eliana so i i see that annette john has a question and i'm gonna unmute you and you can ask the question directly to orianna sorry yeah um hello ariana this is fascinating thank you thank you very much um so i'm just wondering whether no sorry i'm just saying um i'm just wondering whether an alternative explanation of the s shape that you see is not that the cow gets you across a threshold level of capital k star but simply that the cow itself needs complementary capital to be productive so then the s shape holds by construction because you're giving people an acid which is only productive if they already have capital so suppose the cow needs a shed and only those who can afford the shed will thrive but instead suppose you hold the acid value constant but you this time the acid itself contains suppose you only give them a pig but you also give them a peak pen so just the acid package by itself is usable and they can be productive with it under that explanation people might thrive independent of the starting level k because the asset packet already is self-contained would that be consistent uh because you see that in villages where the earnings from the cows are higher the what they need is less so if it were a matter of having the pen or you know the cart to bring the produce to the market we should observe the same threshold everywhere yes so i think it's a combination i mean they certainly are complementarities but i don't think specific to the cow i think you have to have enough to overcome that threshold i didn't quite catch it you observed that if they have less that you better no so uh in villages you know we use the earnings of the middle classes and the rich people to predict how much each cow produces in annual income in different villages and when we do that you find that in villages where the cow yields a higher income the threshold for escaping poverty is much lower ah okay that makes sense so it's not like the fixed presence of a a pen or a cart that makes you escape it's not the bundle okay thank you work thank you and next we have uh miguel espinosa that i'm now gonna allow to talk to you can you hear me good yeah hi ariana so amazing uh material thank you very much for sharing it i i just wanted to know a bit more on the potential spillovers that the uh program had i can expect that you know such a big shock i might share some of this extra income to families and neighbors and what would be the general equilibrium effect of that thing so thank you so in our previous paper where we evaluate the program actually the reason why we collected data on these households every class in the villages was to see the spillover effects and so we found that because the beneficiaries stop working as wage laborers the wage the equilibrium wage increases by 10 percent and that is the only price externality that we find uh because the cows you know the transfer is one cow and the rich people in the village have herds of like 50 cows so in terms of the number of cows transferred as a share of the total cows in the village it's a very small percentage so that's why prices are not affected but with the with the model we can simulate what would happen where prices fall by any amount that you want and they would have to fall by enormous amount 90 percent to to basically make this non-profitable wonderful thank you related to that question um wouldn't these make people in the in the control group to stick to weight flash and productive labor i mean if the wages increase then the people who don't receive a cow are more likely they as wage laborers and that makes them stay in the low lateral equilibrium okay so the randomization is done at the branch level where the branch the area of the block serves so most of the wage benefits accrue to villagers or people just outside the village they don't reach the controlled villages but in any case even if they did a 10 increase in wage is minuscule relative to the income increase that you can have if you switch from wage labor to calgary actually with the model we also simulate that we say what would happen if the wage were to double and you see that there is a bank double so not ten percent you see that there is a group of people that will be actually better at staying in casual labor not everybody is born to race cars thank you very much miguel and jean marco so and next one to ask a question is muhammad basal i'm allowing i'm opening up now and then you can unmute yourself hi orianna thanks for the very interesting paper how are you doing i'm good about yourself this is basically a reunion yes that's it how are you so i'm i'm a bit puzzled with this uh lack of uh sensitivity to the change in prices and like you need huge price reductions in order to offset the effects and uh i'm i'm having two hypotheses here one is whether these households are having a large share of own consumption and that's why you don't see a price sensitivity in the results the second hypothesis is since these are small uh livestock farmers their share of milk production or other kind of livestock production is very low and therefore you do not have in your data a large change in supply of livestock products and if that shock that large shock happens then you might get into bottlenecks in the supply chain like you need refrigerators milk processing units etc and in the absence of those you might get like uh a smaller price reductions to offset your results what do you think of this absolutely so uh factually the they are tiny producers relative to the bigger producers in the same village so that's why we don't observe a change in price what we were doing in the model was simulating a collapse of revenue okay so which could come from the fact that prices go down or because as you say capacity constraints start binding that was a simulation that it would take basically a huge drop for the program not to be profitable but i understand that but i understand that but uh since the model is basically estimated from the data and you don't have such large shocks in the data i'm wondering whether that that's going to impact the simulation as well you mean of course we have a parameter that stands for that multiplies revenue right so we scale that down what you're saying is that it would change the parameters of the production function as well yes what you're saying is like we can't use these parameters to estimate what would happen yes and if you also may comment on the first hypothesis as well the share of own consumption yeah yeah so on consumption no that doesn't matter for this because it's the value okay so it's not the market value is how much the production is worth either you consume it or you sell it is is all the same but i think okay what you're saying is that this model cannot be used to simulate an enormous increase in production because the parameters that we estimate are estimated from a small scale lifestyle craving and that's correct that's like every structural model you have to believe the assumption of the models if the parameters change you cannot you cannot do any simulation so this is conditional on these parameters being correct if the general equilibrium effect messes up the parameters then you're absolutely right okay thank you thank you muhammad um i'm now going to open up the floor for albert and kata so you are now and you you need to unmute yourself no we can't hear you so albert why don't you write the question in the q a and we will bring it up here so you still get your question answered okay so uh we'll wait for an albert question to come in are there other questions um please raise your hand i'm going to open for peter sanness yes thanks um thank you fascinating uh fascinating uh presentation so could you could you shed light on to what extent you think this is a very specific group and so the fact that they are less likely or can't migrate does that suppress a number of other options that have high returns um that forces them to really believe in this option that is that is the way out it doesn't take away any of the of the strength and the richness of the results but but it does um it would affect the comparison with other programs now absolutely i think you know the general point that we want to make is that it is possible if there is a sufficiently large transfer to enable people to change occupation and escape poverty on their own now in this particular setting they had no choice but to basically do what the richer women in these villages do which is standing cows and cultivating land in another setting or another population the way out might be to move to the city but you know the the main principle is the same so if you apply this to poor areas of london probably the way out is to study to get vocational training or to go to college but the principle is the same absolutely yeah absolutely and what would help is to say more about the possible chances for escaping that they have left right is this the only way they have left it so are they forced to do not force but in a way they're going to be more focused to take this way than anything else if there's nothing else left right yes and that's maybe why in pakistan as i mentioned at the beginning everybody who was given the choice between this program and the equivalent in cash took the cash because many people know better what to do with the cash thanks that's great thanks sorry i couldn't switch on my camera thank you peter um other questions please feel free to uh raise your hands and ask your questions uh in the meantime i'm gonna actually bring up some questions that was uh allende genres there were suggesting that you need a complimentary intervention to increase elasticity of demand for milk like da non as a suggestion i yes that that could that would help it wouldn't help getting people out of the truck because if they lack the capital to buy a cow the returns to caloring could be as high as you can get them that would benefit the people with cows it wouldn't solve the poverty trappist but improving the increase in the crisis once the beneficiaries have the cows of course will help yeah so he so he it was the suggestion was a complimentary as well so i'm gonna um i'm gonna ask a question that came to my mind and it's a little bit related to the first question and then joined with a question that was in the chat so this um these people are they're part of the brac program right so that means that these are in a sense you have you have searched to be part of of black or you're interested so these are more entrepreneurial people than the average can one say that no because they don't select themselves into the program is brac who selects them and this the compliance is basically a hundred percent because you're giving them an asset which is worth a year worth of income nobody says no to that okay uh so one of the questions that were asked in the in the chat was have we assumed uh the knowledge skills as the same for people above and below the threshold and and if yes why is that uh they were all given the same training so none of them had a cow to start with and everybody was given the same cow in the same training so i i think yes we can safely assume that they have the same skills none of them were in self-employment before and basically 93 of these women are illiterate so there's no difference in in education either and and just a quick follow up on that so the they wouldn't have worked as wage workers on on the big uh animal farms either they wouldn't have experienced that right so yes but the same above and below the threshold martina maybe we should try albert one more time okay i'm gonna try albert one more time and i'm i'm muting you know so i'll bet if you're there you can try to talk we can hear you doing something with your mic so please i think it might work no okay can you make me the question yeah you can email the question to orianna no sorry it didn't there's something that is not working so i'm gonna uh yes email the question to ariana and you're gonna get a reply to it um so um the last uh i'm gonna we're gonna end this uh unless someone in the panel but there is what's a question that is a good question to end the seminar with uh is to what extent are these findings generalizable and what major variables are needed to be accounted for when trying to apply these findings to various other contexts i think i think these findings are generalizable of course i can't prove it until i get to evaluate something like this in a completely different setting but the fundamental you know the fundamental criterion that you need for these to be applicable to other settings is uh two things one that the distribution of talent is uncorrelated with wealth of birth and two that there are barriers to entry into the best schools into any level of education so if we are if we believe that both of these things are true and they're certainly true of many societies that i know of then the chances that a one-off investment can make a long-run difference are quite high thanks ariana so as many said this is an impressive amount of work for a very long time major sample and following them with very little attrition and is just incredibly interesting thank you so much uh and for everyone listening orianna is going to be available at her zoom room to talk with anyone interested for for 30 minutes uh starting now and you find the link to the zoom room in the um in the chat where we posted it thank you so much and uh wishing everyone a happy holidays including orianna and we hope to see you all on the 2nd of march when we resume for another uh seminar serie with the vw cpr thanks everyone bye thank you you