 All right, so my paper is a lab So, you know that As this company Is a problem for do it for me for many reasons One of the reasons why it's a problem because it yes, it's associated with a bunch of Variety of risks also on the sort of politically on the political side of things And so when we think of inequality like that then one question is okay How do people think about redistribution and demanding redistribution right which I think both in South Africa and Colombia would imagine It's a it's a big topic So what explains support for redistribution why do some people want more redistribution than others So this is the demand side of the redistribution equation And here we focus on three On the role of three three factors one is aversion to high inequality the other one is the personal loss of a more equal distribution and Then just as a sort of sanity check the role of luck versus merit when it comes to redistribution I'll explain better in a second so the literature shows that We as humans have fairness considerations in ourselves, so we do There is ample evidence that we care about Fairness even when it doesn't involve us directly, okay and And this links to the inequality opportunity talk we heard from from Chico Ferreira the other day Inequality that arises from things that are considered Less fair is also more objective Objectionable and that means that we would probably expect to see more demand for a distribution if that inequality is considered less fair In fact, this explains a little bit the different attitudes between for instance Europe and the US Where in the US people tend to believe that effort is rewarded in Europe We think a little more that luck has a greater role and that might explain why we accept more taxes than for instance people in the US So our contribution is so we have all of this evidence, but it's mostly from high-income countries So our first order contribution is to ask these questions in a high inequality Emerging country, okay, so that's really what our first contribution is and this is a lab setting in the University of Cape Town where we have three Students and I'll explain well the sample is more than three, but there is three students and I'll explain how it works So why is this South Africa an interesting place to study these questions? Well, so Africa Speaking of fairness and how fair inequality is so that I think is one of those countries where the current distribution of income is a direct result of explicitly discriminatory Institutions so we can't just say well, you know the market is doing its thing Well, the market is doing its thing, but you're still paying the price of a system that did not allow the majority of the population to express their full potential so speaking of You know merit and luck. There is no merit here, right? Despite all of these however in the world value survey, we don't see a high demand for a distribution in South Africa In fact, we see comparable to other countries with much lower inequality less or what explains this puzzle Why why and why are South Africa not demanding more redistribution? So the experimental design works like this I'm gonna spend actually most of the presentation on this because the results are fairly simple So there is a production phase in which two players are paired and they have to complete a series of tasks And then there is a decision-making which is this third player was talking about this third party. We call it a stakeholder That decides how to split the payment between those two players. Okay It's important to understand that the third party player is not Does not complete the task does not participate in the production phase. It's just the person who decides who gets paid more. Okay Why do we need the first phase we could just make up some players and then do some hypothetical questions Well, the literature shows that having real persons Being paid real money makes the decision-making phase more binding and more realistic So that's why we we have the first phase with real people rather than just ask hypothetical questions We had a total of 335 students and we had about 220 in the workers Conditions or production phase and about 115 in the decision-making phase now our sample of interest is these guys because they Are those who make the decision that we are studying So here's the this is the most important slide Each of those stakeholders gets this piece of paper and they have to decide. Okay, listen You have 25 potential split of payment between player a and player b They just completed some tasks in the other room and now you have to decide who gets paid more One of these 25 will be randomly selected as the payment split So each of your decision is important because one of those is going to be the real payment So our sample size is 116 times 25. Okay, so that's how many decisions we observe So here we say we vary that the difference between the winning player and the losing player So this is max difference hundred versus zero But we also vary the cost for the third party stakeholder to Redistribute okay, so for instance here is zero whether I redistribute or not I don't lose anything but here is 45 which means that if I want these people to to be more equal in the payment split I have to pay 45 rents. Okay so We vary but the inequality The difference between the players. So here is 100 down here. It's only 20 the difference But we also vary the cost that the third party stakeholders in curse when deciding to Split the money equally So if I circle yes in the for instance in this line if I circle yes The players get both 50 percent and I get 100 which is my endowment payment minus the cost So 85. Okay, so they do this all the stakeholders. And so we have as I said 25 decisions times 116 stakeholders How does it change by how much it costs me? to affect a fair distribution so We see that It's decreasing meaning the more I pay The less likely I am to care about fairness so to speak So if I lose if you're touching my pocket, then I'm less likely to enforce an equal distribution And we still sorry we still have a difference Between the lack and the merit so for each cost level the lack People still redistribute more than than the the one in the merit condition, right? But it's decreasing in the cost What about by the level of inequality well the higher the split between player a and player b the More likely you are to redistribute. So we interpret this as an adversion to inequality So another important thing about this graph is that you see that the difference between the lack and merit is less relevant at high levels of inequality which we interpret as a sort of Unconditional this state this taste for excess inequality So for the cost we didn't see that for the cost it was the same essentially two parallel line here instead The source of the inequality whether it's lack or merit matters less when the difference is too much I don't care if you were more productive than player B. You shouldn't get the whole pie essentially That's what this is saying. I don't care if you were more productive than player B You shouldn't get 80 20. That's too much of a split So this is what we are calling an unconditional this taste for inequality This table. It's what I just told you but it's a linear Probability model. So essentially this is saying that higher inequality more redistribution Higher cost less redistribution and this is what I was saying earlier that there is an interaction between between In the inequality condition where sorry the inequality and the merit condition interact meaning that in What we saw here essentially, this is what this means, right that inequality is more important in As that the source of inequality is less important the higher level of inequality So this is just the regression version of these two graphs Does this matter so what is that you can calculate the genetic efficient out of all the potential redistributions, right? And so the genetic efficient that we started with we made it similar to the real g-link efficient or something a point six So what happens after people play and redistribute? Well in the lack condition the genie goes down to about point three and here it goes down to about point thirty five So you see how the redistribution changes the inequality in this Fictional world that we created So it's important to understand what determines these levels because then the redistribution is kind of the That the way you bring down the genetic efficient And that's kind of the sort of second message of our paper So the first is that the source doesn't matter at high levels of inequality the second one is Well, even in context like South Africa where there is a lot of understanding of a big a very Deep understanding of the unfairness of of the inequality in that country Clearly you have personal loss of a potential limiting factor So this experiment actually was motivated because when we started this experiment there was front-page news in South Africa about something called expropriation without compensation, so it's part of the South African debate whether the government should take land from mostly white farmers and Give it back for free to other people without compensating the white which in other countries will sound like a crazy radical distribution Idea, but within this context we did this experiment because the idea is to think okay Let's see everybody says they want more equality, but what if I start charging you as For for that higher equality and so essentially where we're interpreting this as Personal loss as a limiting factor for the equalizing power of redistribution now because this is an experiment we wanted to make sure that People were thinking the way we thought our experiment would make them think so post post experiment we had some some questions about You know what you know what made you? Decide the way you decided and things like that so what we find is so we analyze those those responses And we in a sort of qualitative way and what we find is that? first of all our sample We had some questions that are similar to the one you find in the world value survey Which is supposed to be nationally representative and we find it for those questions those questions are like the ones for instance Do you think the government should be responsible for reducing poverty? Or do you think the people should be responsible for reducing poverty? So those are the type of questions right and we find that our sample is not systematically different than And nationally represent the sample in the world value survey so that's good It means that our students are not weird in the Joe Herrick sense We also found because we didn't prime them our our questions were very open We find that they mentioned the word fairness lack Deserving good they mentioned these words a lot Which means that they were deciding about redistribution in the way that we wanted the experiment to make them think So essentially we conclude this by we we we interpret this as a sign that We were eliciting social preferences, which is what the experiment supposed to be doing And it also clear from the post-experiment survey that the The student that the players understood what the experiment was about because that a little Decision schedule the show is fairly complicated right, but the student it was clear from the responses in the post-experiment survey That they understood what they were doing So let me summarize This is an experiment in a high inequality context in the middle of a national discussion About fairly radical redistribution policies. So that's the context in this experiment And when I say front page front page, this is stuff that was still today actually it's discussed But this is something that again it's part of almost everyday discussion How do we redistribute and how do we redress the wrongdoing of apartheid? so what we find is that the higher the Payment split between The winning player or the more productive player and the losing player or the less productive players the more the willingness to redistribute result number one result number two The higher the personal loss of redistributing the less redistribution so essentially personal loss is a limiting factor for redistribution More redistribution the lack treatment, which is what I said. It's sort of very consistent with the literature And then that's sort of interaction effect that I was talking about earlier There seems to be an aversion to extreme disparities essentially meaning that Even when you think that player a should earn more than player b because they deserve it Even in that case if the distance between in the payment split between player a and player b is too much Then I'm still more likely to distribute Connect this to the discussion for instance in the US about the CEO being paid What is it 400 times more than the person working? I don't know in the in the office, right? So even though perhaps the CEO deserves a little more money does that mean they should earn 400 times more, right? so that's kind of the The analogy and then this negative impact or personal cost Which is not influenced by the source So it doesn't matter whether you deserve or don't deserve to be paid more if you're affecting me personally then I'm I'm gonna redistribute less Why is this Relevant for policy well as I showed you we have more in the paper But here I just showed you one graph it does change the post redistribution level. So like we saw was it yesterday in the policy session about the I think it was Santiago levy talking about the post The post tax and the pre-tax inequality. So as we see it's important I think he was making the example the difference between Colombia and Europe, right? So as we see it's very important if there is redistribution or not because it does change the the Gini coefficient So again, this is an experimental setting, of course But the intuitions are in our mind applicable to the real-world redistribution And essentially the conclusion is that Even in a setting like ours where fairness concern are prevalent as showed by Surveys and by our experiment support for major redistributed policy like the one I was mentioning May not be guaranteed by the the existence of this fairness concern. So essentially It's not enough for people to care about fairness. We need something that is we need a way to redistribute to that is compatible with people's personal Utility and and and not wanting to lose essentially and I think that's it