 facing to quit to alternative jobs. And this really in our conceptualization of comparative markets regulates wages in market economies with impacts on, you know, all sorts of things outside of the labor market, even prices for activity, output. And the question really, I'm trying to investigate in this question is in this in this paper is how relevant is this in South Africa, especially given our high end employment, it's a labor surplus country, we might think that in developing countries, kind of competitiveness is a little bit different. There isn't that much evidence on these kind of economies. A second kind of implication of this is that I want to argue that that is that these kinds of implication of monopsony in South Africa, which I'll lead to really provides this interesting link between our high level of unemployment and our high inequality, which as we know, there are plenty of historical drivers, for example, through apartheid policy, the job the job structure that was created. But, you know, this might be a link that is a continuing link through firms and through monopsonistic competition that keeps high unemployment that travels from our high employment towards our high inequality. Okay, so to move straightforward into what this the roadmap of this presentation, I'm going to start off with some to talk a little bit about what we think are indicators of competitiveness of the labor market. Two key indicators are firm wage, PMIA, and I'll give evidence on that. A second key indicator is rent sharing. I'll give evidence on that. Then I'll talk about different explanations for these wage, PMIA and rent sharing, focusing on monopsony power, but also addressing unions, which is, you know, there's quite a big South African literature on unions. Okay, so to hit directly into the data that I'm using this study, this is matched employee employee data from I'm currently in the data center right now. So I'm using 2011 to 2016. It's a new universe of all formal sector workers. Sorry, ignore this. There's definitely not any hours. And I'm focusing on workers in the firms with greater than 20 workers. So this is mostly a large firm analysis, but that doesn't really affect the reduce the numbers that much. So if we, this is 2011, 2016, some of these statistics, if you look at 2013, for example, you know, out of around 10 and a half million jobs that are observed in the formal sector in the data, nearly 9 million are still in my sample of large firms. And that corresponds to around 45,000 firms. And about 85% of those firms have sales data, which is relevant as I'll say later for my rent sharing systems. Okay, so to the question about are they rents in the labor market? And just to take a step back here, when we talk about a competitive labor market, as we say, it means that workers can credibly threaten to leave to another job. And that means that similar workers are paid similar amounts. Because if you if you're underpaid, you're lifted to what you can get at another job, then you leave towards that other job that pays you your correct amount. In a less competitive setup, it means that similar workers earn different amounts. And the reasoning here is that, for example, low earning workers, so firms that pay workers under what, you know, similar workers could get elsewhere in the economy, these low earning workers can't easily switch into the good jobs in the economy, even though the worker that does have that good job has very similar characteristics, the worker in the low earning or bad job can't easily find or switch into that good job. And the reasons are, there are a bunch of reasons, but the reasons I'll investigate later on is about monopsony or unions. The bottom line here is that, as indicators of less competitive markets, we can look at firm wage premium. So for similar workers, our very similar workers paid very different wages depending on the firm they're at. That's an indicator of less competitive labor markets. And then rent sharing. So rent sharing is about, are these firm wage premium correlated with the profitability or productivity of that firm? And that really tells us that it's really something about the firm that's causing these firm wage premium, not something else like higher wages in Johannesburg compared to more rural areas in order to compensate for something. So just to take a quick break and to engage everyone, I'm going to launch a quick poll here at this point. It should be on your screen. And I want to ask you what proportion of total inequality? So this is of all inequality, any type of inequality, obviously income inequality, in South Africa, do these firm wage premiums explain? And you've got an option of saying, I don't believe there's any firm wage premium, or at least there's very little part of it. You can say it's very little 5%. Some people might say more 20%. Others might say a bit more 40%. I'm going to wait for maybe 10 more seconds, 20 more seconds. Just choose. Okay, I'm going to end the poll now. And it seems like, so hopefully everyone can see the poll results. And we see, ha, everyone believes that there are some firm wage premiums. So and the kind of over half says 20%. Some people say 5%. Some people say 40%. Maybe I'm bad at setting questions because it was clearly in the middle 20%. That's what I'm going to try to convince you is the approximate answer. But we'll see that later. Okay, so that's interesting. So to head into the firm wage premium. So this is really a non-parametric setup of the first kind of other firm wage premium. And here, this is a very standard kind of graph in the literature to take you very slowly through this graph. If we look at, for example, this line over here, so this line over here is showing a worker in a quartile one firm. So we're dividing firms by the average wage in this in their firm by the lowest quartile one to the highest quartile four. So the lowest origin quartile, so the way that workers start at is all in red. And this, these people over here start at a low wage firm, and they go at the event there towards a high wage firm, a quartile form firm. And you can see the Y axis, this shows the mean log wage of these people who are switching firms. And it's kind of incredible, because it's kind of something like a 60% increase in these people's wages, just the same person, the same people, just switching firms from a low wage firm towards a high wage firm. And kind of even more interesting, perhaps, is that the opposite occurs where you have someone going from a high wage firm to a low wage firm, they also get a penalty of exactly the opposite, the opposite magnitude, the same magnitude, the opposite side. And so it's really speaking towards these firm wage premium. Another kind of way we can look at this is a kind of matched event study design. And this is really to say, well, in this, in this figure, are we really taking care of are we comparing like with like and so on the old sorts of problems you can think of, you can, you can, you can point out this, this is a much more kind of parametric study where we're trying to control for a bunch of things. And over here, we can take a bunch of people who start at exactly the same firm who are in the same age category of the same sex, they earn within the same wage bracket. So these people are very closely matched. And we find that at a certain date, one separates to a high wage firm, and another separates to a low wage firm, where high wage firms determined by the average wage in that, in that firm. And using that kind of design, we can see here that a kind of there's a large correlation between when the people who switch to a high wage firm actually do get a high jump in their wages, despite controlling for all sorts of things. And we have flat pretense to really show that there's nothing else kind of, there's no like pre characteristic indigeneity that's driving these kind of reasons. So to get back to the polling question, what does this tell us? This tells us that if we kind of take another approach, we decompose these firm wage premium by saying that wages can be decomposed into a worker fixed effect, a firm fixed effect, and a bunch of other stuff, we can do a kind of very standard decomposition of variance decomposition. So that we can set the variance of a large wage and so Africa is around, you know, one comma three two, what does that mean? But of that, firm fixed effects account for 23%. And sorting from high wage, from high wage workers to high wage firms, the covariance between firm work exhibit account for another 11%. Worker fixed effects. So these are kind of immutable characteristic of workers might include kind of skill, maybe education, of course, it also includes things like race and sex and so on. It's not, it's actually 43%. It's not that much. And just to answer the poll, this is kind of, if you add up to 23 and 11, that's 34% of formal sector wage inequality. But, you know, there's a bunch of, you know, this research, for example, by Arden Finn and Marie Labyrinth and Ingrid Willard that, you know, something like 65% of inequality is accounted for by wage inequality. And the rest is about the difference between unemployment and employment. So once we adjust for that, we can say that something like 20%, one-fifth of total inequality is accounted for by firm wage premium. So that's kind of massive, in my view. It's something to take seriously. And it's a large part of this kind of, the kind of crisis of inequality that South Africa has been having for the last two decades. But we need to be talking about that, a lot of that inequality coming from the firm side of the labor market. Okay. Now to look at grand sharing, and this is really to say, as I said earlier, are these wage premium really something driven by the firm? Is it about the firm's profitability? And then in that case, it must be something about the firm setting wages, or maybe unions optimizing at the firm level. But the point is it's about a firm characteristic. And we see here that there's really a strong correlation between value added per worker, or if you take sales, or if you take profitability, no matter what measure you use, and wages, it's a strong positive correlation. And if you look at the kind of estimates that come out of this, if you regress, for example, just a simple regression of wages on value added, you get something like 0.3 elasticity of 0.3. But we can do all sorts of other things to adjust for all sorts of other things. So for example, standard military chase, we use the firm wage Pima estimated directly. And we can see that the estimate comes out to be about 0.15. If we use only closed firms, so this is to avoid indigeneity of who's switching also 0.15. We've also got some I've also got some estimates of if you if you if you take out, for example, the part of profit that's to do with industry, or, or location. So you look at really something that's quite specific to the profitability fluctuations year to year. So first differences, in a firm in a firm's profitability, if you just take that kind of push, you might argue that's a kind of somewhat exogenous push, then you can see wages do change on the order of around 0.2. And these estimates of 0.15 to 0.2 elasticity. So what does that mean? That means that if you know, your, your, if profitability increases by 10% wages increased by something like two percent, that's higher than what we find in the international literature. You know, generally international literature something between 0.05 and 0.25. Okay, so the bottom line here is really that it does in my view, I there's some kind of substantial evidence that there is a firm wage premium accounting for a large range of concerns, which I haven't gone through totally in this in this in this presentation, but happy to fill questions on it. And then, and we can see it almost more directly in the rent sharing as well, higher in sharing than then in other countries. And so at this point, I really want to go into explanations and one explanation that I just went to quickly over before entering another poll is monopsony power. And what is monopsony power? This is, so remember, if you, from the side of the side of this presentation, I talked about how a key, a key constraint on market economies is that workers have the ability to detect and to quit. And monopsony is simply saying that quitting a job is costly. And so that means that firms can decrease wages while still retaining their workers. There are many reasons for this, a bunch of models that account for it. You can look at search costs, you can look at job differentiation. But the point here is, is, is for example, that you can test this directly. So if you look at how separations respond when wages increase, if, if separations decrease, then it's really a worker supply side response to wage setting. And we can also see that monopsony does, as I said earlier, hinted earlier, perhaps form a link between unemployment inequality for South Africa. So the link is really saying that in high unemployment settings, if you follow a range of models, high unemployment means that there's more monopsony power. One way we can think of this is in terms of search costs. If there's a lot of unemployment, you like, you get a job, you have to hold on to the job. If the firm decides to decrease the, decrease the, your wage, you're still not going to let go of that job because your alternative is something like unemployment. You can't easily switch towards another good job or another job at all. And so higher unemployment implies more monopsony power. But more monopsony power also means that there's higher end sharing. And that really comes from, from, from kind of model predictions that if you, if your firm is more productive, you're really going to be increasing the number of workers a lot, a lot more in order to take advantage of that kind of profit per worker that you're making. And so that higher end sharing translates into higher inequality. And this is a kind of interesting link to firms between unemployment and inequality that I think warrants a bit more attention in the South African context and perhaps for developing countries generally. Monopsony is, you know, something that's, for those people who are less familiar, it's something that's pervasive in, that's been found around the world. There's quite experimental evidence in international literature. There's concentration evidence. So monopsony really occurs, can exist, you know, with many firms. You can have even in dense labor, labor markets. So for example, in a related paper, I have with co-authors looking at US urban centers, we actually find that urban centers have higher monopsony power than even some, some rural areas. You know, in a meta analysis developing countries, there are a few key studies, Brazil, India, Mexico, where substantial monopsony power has been found. Question is, maybe I'll convince you that there's monopsony power in South Africa. Really interested to hear from Rulof whose paper really, I think, plays well into monopsonistic markets. But at this point, I want to field another poll and to ask, given that everyone is now up to scratch on what monopsony is, why is the South African labor market not more competitive? So you can say, I don't really buy your results on wage-premia. I don't really think there's that much inequality. That'll be your first thing. Maybe you think it's monopsony power. Maybe you think it's unions. Wow. Okay. No, I need to adjust for something. Okay. Don't jump in. Very interesting. Okay. I'm very interested if people have answered other, if perhaps you can put in the chat box what those other models you have in mind are. It's an extremely interesting question to me. Hmm. Okay. Let me give it 10 more seconds. And perhaps if anyone, if those people think it's, who are unsold by the evidence presented earlier, would love some feedback on the kind of issues you'd have with that. I would really appreciate as much feedback as possible. This is to be one of my dissertation chapters. So any feedback is extremely welcome. Okay. So I'm in the poll and it seems that the modal response is unions, which I think is a part of the story. But as I'll say later, I think is only a part and not the whole story. Okay. So I guess I'm going to try and convince you that monopsony power is more important than what the polling currently suggests. So to go back to those key tests, one thing we can think about is if you look at two opposing models, if you look at unions versus monopsony power, for example, if a firm increases their wage, what do you expect the response to be in terms of separations? Well, if unions increase, if we're traveling up the labor demand curve as in kind of a right to manage union model, it would be that as wages are increased, then firms need to adjust by decreasing firm size, and so separations would increase. However, in a monopsony model where you're really riding up the firm labor supply curve, it means that workers who, when wages increase, workers are more attracted towards the firm, and so separations actually decrease. Now, it's a bit confusing, but the point is here that also even in a kind of, we do find a decrease in separations in South Africa, but we also find that the responsiveness in terms of separations is slow compared to other contexts, and that really points towards the low ability of workers to easily switch to higher wage firms. And so we can see here, if you regress, for example, a very simple regression of wages on separations on wages without controls and adjusting for hires, for example, you get an estimate of around 0.5, so the labor supply elasticity, so how much is the kind of dynamic firm size respond to increase in the wage. If we add controls, it gets closer to like 0.9, 0.85. Even in first differences, we get something quite close to 0.8 with large standard areas, but it's bounded below, say, 1.5, and this first differences estimate is really, as I said earlier, if you think about the rent sharing shop, when firms get a specific shop that's different to their location and different to the industry, and you see their profits change, or sorry, their value added, change across years in a kind of stable manner, that's the isolated first differences profit shock, and we use that shock as a pass-through onto how does that profit shock translate into wages in a kind of insurmountable variable regression to kind of isolate exogenous increases in wages. We see that separations decrease in a manner which really implies there's a lot of monopsony power in the labor market. And finally, if you look at the re-separations kind of design, which I'm running out of time to explain properly, but the point is that this uses the kind of matched event study design that I spoke about earlier, where you match workers very finely, as I said, and then two workers go to two separate firms, and you see that wages change a lot, that's kind of exogenous wage increase. What is the re-separations? How quickly do they separate from that new firm? Again, it's kind of higher than all the other estimates, but compared to the low, compared to the international literature, it really suggests a lot more monopsony power in Africa than other countries. So finally, at least I'm going to have to ask you for two more minutes, please. That's perfect. So finally, I want to say a little bit about unions, which is the alternative kind of explanation of these firm wage premier. And as one model of unions is that as unions bargain for higher wages, this might give firm wage premier. I mean, one question is that in South Africa, you have unions are really largely organized around industry and location. And even once we take out industry and location, we see really strong dispersion in firm wage premier, which really shows that unions are an important part of the story because those industry location average firm wage premier are high, but it's really an insufficient part of the story. If I recall correctly, it's something like 40% of the total inequalities accounted for by location or industry firm wage premier, sorry, location industry wage premier. The other question about unions is that these non-union firms, so if you kind of take industry locations that have an extremely low density of unions, non-union firms actually have extremely similar dynamics in terms of substantial dispersion in firm wage premier, and they have large rent sharing. And so one example of this is to show that if you look on the X axis here, as the share covered by the bargaining council actually decreases, go towards the left, then you actually see an increase in the dispersion of wage premier, which really shows a kind of points towards kind of uncompetitive dynamics that are outside of unions, even though unions may be a large and important part of it. Okay. And so to end off, I wanted to say a couple of things in discussion. One is that I've looked at the formal sector only here, but we can look at very suggestive evidence on the informal sector, for example, from the survey of employed and self-employed. And we see some kind of, if you look at the very crude rent sharing regression, we see similar results. We also see similar dynamics of transitions of workers between formal and informal sector. It's suggested because they're kind of fit to it that I can't sufficiently justify, but I think it's consistent with the formal sector evidence. And the last thing I want to say is that, as I said, monopsony perhaps offers a link between the kind of high unemployment in South Africa and the high inequality, which might really generalize the development process. I really talked about that higher unemployment implies higher monopsony power. Another kind of reason it might contribute to inequality is that when you have uneven industrialization, it means that some firms are getting much higher productivity than other firms. And so it means that given a constant rent sharing elasticity, there'll be higher kind of rent sharing effects on the inequality distribution given uneven industrialization. And so both of these contribute to really higher wage inequality through firms, through this firm wage pymia, through the kind of monopsony channel that I've been talking about. Okay. So to end off with, I hope I've convinced you that, you know, firm wage pymia are a large part, an important part of South Africa's total inequality. It seems to me, I think quite strongly that competitive dynamics should not be the baseline assumption when we do economic analysis. In my head, the primary alternative models are monopsonian unions. I hope I've convinced some of you that monopsony is something to think about properly. I agree with you, with a lot of you that unions are an extremely important relevant continuing part of our labor market. And of course, there are several policy implications that lots of people are hopefully going to be doing work on this stuff about ETI, Josh Badlien, I know he's doing a lot of work on that. I'm doing some work on unions and monopsonistic contexts and so on. Okay. Thank you so much for listening. Sorry I had to rush. Very happy to engage with questions. Thank you. Wonderful. Thank you so, so much, Isan, for a really wonderful presentation. We will head over to Rudolf as a discussant. He'll have about just under 15 minutes. And thanks to those who have posted in the chat function, let's attend to those after Rudolf's presentation. Over to you, Rudolf. Thanks Isan. As I told you before the presentation, I looked forward to reading this paper. I've been hearing about it for a long time. I think like a lot of people who work in labor, I've been waiting for someone to write this paper for about a decade now. So I'm very thankful to UniWider and Treasury and to you that we finally got this, we're starting to fill this big hole in the South African literature on wage inequality. So I think that's great. Rudolf, I'm not sure if it's only, oh, there we go. Okay. So the motivation for this paper, I think firstly, is that there's this large empirical literature that dates back to the 1950s that have observed that people who look the same and who do similar jobs can earn very different wages if they work in different industries or different types of firms. And for a long time, economists have speculated that this may be mainly due to unobservable human capital differences. So if some firms attract more resourceful, more odd working, more intelligent workers, then maybe that's why they pay higher wages and it's all just a productivity story. And I think part of the reason why economists clung to this explanation for so long is that it allowed us to maintain this assumption that markets are basically competitive. But the useful thing about this text data is Son observes the wages of workers who switch between different firms, which allows him to control for time-invariant unobservable human capital differences. And I think most of the unobservable human capital differences that we care about are going to be time-invariant. And importantly, his paper demonstrates that this is only a part of the story. Even after you control for these things, there are these important firm effects and that's kind of a big deal. So I think that's the first motivation for the paper. The second part of the paper is all about why these firm effects exist. And the reason why that's important is if firm effects are in the indication of monopsony power, then that implies that workers are being exploited and it suggests a more active role for policies to improve the bargaining power of workers, remove informational frictions, reduce the effect of transportation costs. It might also suggest that minimum wages can increase employment, although only if wages are below a certain threshold. But importantly, there are other potential reasons for these firm effects. They can also occur as a result of differences in capital intensity, skilled capital complementarity, efficiency wages, certain job search models produce them. And if that's what's really going on, then that would require a different policy intervention. So the main results in the paper firstly, there's kind of the descriptive bit. And there, Isan tells us that about 25% of total wage inequality is due to firm effects. That can explain about 60% of the gender wage gap. I don't think you said anything about that today, Isan, but I thought that was quite a striking result. And it explains about 40% of the wage gap between workers at the bottom and the middle of the income distribution. And then Isan also argues that the results are well explained by a model of monopsynistic competition. So let me start with the descriptive analysis. So I read a different version of the paper. I think the one Isan presented today was hot off the prices. So in the paper that I read, Isan found that 71% of wage variation was due to worker effects. So in the current incarnation of the paper, it's down to 43%. But that's still a lot. So we know that internationally and in South Africa, we can usually explain about 20% of wage differences with observable human capital, education and experience basically. And that means that in South Africa, about 23% of the wage gap must be due to these unobservable worker effects. So that includes racial discrimination, gender discrimination as well, obviously, differences in school quality, also background, social networks. And although this isn't what this paper is about, and it's not what Isan focuses on, I think that's a big deal. So the paper demonstrates that these things really matter and I think that's an important finding. And then the other result is that 23% of wage variation is due to human effects. So it's less important than worker effects, but it's still an important source of wage inequality, especially in a country like South Africa where there's so much wage inequality. That's a larger gap than in many other countries. And understanding why that occurs and what the policy implications are, I think, is very important work and issues that we haven't really thought about before. So I think that's great. The other results from the descriptive analysis that Isan didn't focus on, he finds that 60% of the gender wage gap is explained by firm effects. So that blew my mind. I was very surprised by that. So I think if I understand that correctly, it suggests that most of gender wage differences exist because most women end up in bad jobs or bad firms that also pay mean similarly low wages. So I'd be curious to know Isan, how much of that is due to industry effects rather than firms within an industry that pay different wages? But I do think there are some policy implications to this. I haven't thought about this a lot, but I think it implies that a legal framework that focuses almost exclusively on getting firms, a specific firm to pay men and women the same wage for the same job, is missing the bulk of why gender inequality exists. So I thought that was interesting. Let me maybe skip over this next point just for the purposes of time. The second part of the paper is all about why these firm effects exist. And I think that's more ambitious, but it's also more difficult to make that point persuasively. So firstly, Isan finds evidence of rent sharing. There's this strong, positive relationship between profitability and wages. And he showed us that graph, and that suggests an important role for institutions like trade unions and helping workers capture a larger share of the profits. Otherwise, all of that ends up in the pockets of the owners of capital, and that's probably not what we want. The one question I add there is in the one regression, it seemed to be that much of that correlation disappears when controlling for capital stock. I don't know whether you've done more work on that Isan, but I think that really matters. The result clearly stands, clearly there is this correlation. People who work for more profitable firms earn higher wages. We know that, but to know whether that's a rent sharing story, I think we need to rule out some alternative hypotheses. For example, that firms just have different technologies. Something like complementarity between skilled labor and capital I think would give you the same result and is also consistent with this sorting of higher skilled workers into higher paying firms that appear by the face of it to also have more capital. So I think that would be interesting to see. Secondly, at least in the paper that I wrote, the next thing that Isan looks at is evidence of monopsynistic wage-sitting powers. So the first bit of evidence there comes from a result that firms that are important employers in an area industry sell don't pay much lower wages. So that's evidence against monopsony. So there is a relationship, but it's not very strong. In that regression, I'm a little bit concerned about the concentration measures. So one of the things that if I understand the text data correctly is you don't actually know where the workers work. So my understanding is that all 100,000 shop right workers will be treated as if they compete in Cape Town, whereas because that's where each operator is registered as a tax entity. But most people will be kind of dispersed in the country. And if that's the case, I think that could explain why the correlations that you're finding isn't stronger. So I'm not convinced that this is evidence against monopsynistic wage-sitting power. And then the other bit of evidence comes from the fact that Isan sees that firms with higher firm effects suffer fewer separations. And that's exactly what we'd expect in a model with monopsony power. Again, I think they might consider the same as it was in the previous analysis. I'd be very interested to see what happens when you control the firm size and capital because I think that relationship is also what's predicted by some other models. And it would be just useful to rule out some of the alternative hypotheses. So just some final thoughts. I think this is a very important contribution. I'm going to be thinking this over for the next few months, definitely. I think the descriptive results are very interesting. And so estimating these models, I know, isn't easy. And I think, Isan, you were very transparent and very rigorous. And I think, so that part of the paper is great. The tricky thing is when we try to understand why exactly these firm effects exist. So you make the point that monopsony power fits the data very well. And I think that's true. We also know there's a bunch of other things about monopsony power that fits well with other things in the South African labor market. We know that we have very high transportation costs. We know there are big information asymmetries. So my assumption before I read the papers that monopsony is probably a big deal. But it'd be great to see whether that's the main thing or whether it's just kind of incidental. And I think as things stand, it's not clear to me that we can rule out things like efficiency wages, job search models, differences in technology. I think it's very difficult to know exactly what the most effective policies are to respond to your demonstration that these firm effects are really important until we have more clarity about why exactly these firm effects exist. And that's it. Thank you so much, Rulof, for your insights. Let's open the floors for some additional Q&As. Okay, Jaycee. Hi. Can you all hear me? Yes. Okay, wonderful. Okay, so thanks, Desan. I guess, Rulof, the things I have to say are sort of elaborations on things that Rulof, like what the counterfactual is when people say firm effects explain this much of the variance or whatever. Like firms differ in a bunch of ways. Some have good managers, some have a lot of capital, some are located and close to their customers. All of these things can create differences in profitability. So when you say it explains 43% of wage inequality, like what's the counterfactual exercise that you have in mind? Like are we going to make all firms have exactly the same amount of capital and randomly sort managers across firms? I just want to know what the counterfactual is that you have in mind. And then the second point that I wanted to make was that there's been a lot of stress on unions as a potential explanation for monopsony power now. I don't want to take a stand on exactly how much monopsony power there is. Obviously, that's sort of the substance of this paper. But I do think that when you start speculating about what the policy solutions are, it was I think a notable omission was that you didn't even think about sort of supply side policies. Like why not? But like if you really think that the labor market is really characterized by a lot of monopsony, real like radical openness to trade and deregulation so that you get lots of new entrance on the employer side would be a potential solution. So that I mean, I don't think that this is obviously this is speculative but we as we've noted, there's lots of models to potentially use to think about the situation and we aren't quite sure which one is most relevant. So anyway, that's my comment. Isan, would you like to respond to JC's as well as Roulop's points? Yeah, sure. Thanks so much for these extremely valuable interesting comments. Really appreciate it. So just to pick up a couple of things, but to start off with Roulop, your question around what to do about capital specifically and it's something that that result is quite striking. So when you control for capital, the erraturing elasticity does decrease substantially. But it's difficult to know what to make of that for two reasons. One reason is that the test here is whether a worker at a high capital firm is able to go to sorry, the firm wage premium that is that is calculated for a high capital firm is calculated on a worker that is working at that high capital firm and is able to and then switches to another firm and what the wage penalty or gain is associated with that. So it's really kind of looking within workers that actually do work at those firms. And one point here is that in theory, at least in a very competitive labor market, we'd expect even if two firms, one high capital, one low capital, share the same labor market, then there should be no firm wage premium regardless of capital. As you say, a fit to this is the kind of complementarity between these skilled labor and capital and whether that messes up with the estimation and need to think a bit more about that. The second point I want to make about capital is that one kind of hesitation of controlling for capital is that it might be the reinsharing might be kind of endogenous towards the capital. So, for example, unions might bargain differently in terms of rents if we think unions is the correct model versus nobsony when they are when they are high capital firms due to kind of like capital hold up, for example, and it's anyway the implications of that is not clear for the estimates because it can increase it or decrease it. But as you say, I think it's something important to disentangle. Just quickly on Jesse's questions, thanks so much, Jesse. I agree it's kind of, it's like a wild kind of counterfactual to say, oh, what do we do if we add no firm which premier at all. But in my mind kind of, if we accept that labor markets that are really competitive allow workers to switch whenever they want to, and we accept that these firm which premier really are measuring kind of rents that similar workers can get if they were able to switch, then it doesn't seem like an unfair counterfactual to me. In this kind of world where you have a very competitive labor market, you could still have lots of kind of heterogeneity along the firm margin for all sorts of other reasons due to kind of, you know, stuff in the in the in the product market going on or so on, and which will in some ways interact, but at least in principle, I think if I have it right, you could you could have a perfectly competitive kind of labor market where you have, you know, negligible firm which premier with, as you say, even though some firms have great managers and others don't. And then lastly, on the on the policy implication, I mean, just one thing is that I think and as as you know, says this is like a thing that we would be great to get more investigation on and it's one of the reasons I really like rule of your paper on the reference letters is that I think for for South African labor market, a lot of them power is driven by things to do with not on the kind of not necessarily on the on the employer side in terms of, for example, the number of entrance that exists. So as I said earlier, you get the situation where even in highly dense urban kind of industry cross locations, you can still have a lot of monopsony power, kind of, which kind of implies that that the number of firms doesn't solve the problem. A lot of the problem can come down to something like just the ability to to to search at all stuff like transport, as you have mentioned earlier. Yeah, but I mean, only the comments like I mean, extremely helpful and lots to think about. Thank you. Isan, maybe a question from my side. So it relates to the point that rule of made about like shop right is registered and they have their headquarters in Cape Town. But obviously, we have many shop rights dispersed throughout the country. What did then make sense for you to work on the PAYE the branch level instead of the yeah, in the data, the tax reference number? Yes, so this is actually estimated at the branch level in the in the updated version, which is not sorry about that. But yes, precisely for that reason, I switched to the to the branch level to get a finer, finer indication of location. Okay, fantastic. Do we have any other questions or any other comments? Andrew Donaldson, I see you have another point in the chat. If you would like to elaborate on it, you're very welcome. Well, hello. It's not so much a point in the chat. It's just a question on if sun and how you interpret the events, upwards moves of those from lower quartile wage firms to hire, uh, alongside movements down, um, uh, of, of, of workers from higher wage firms. So that these may not be, there's an asymmetry between these two moves. You know, one, if one wanted to pose a human capital explanation or an individual productivity explanation, you could say that workers moving up are likely to be more productive workers who succeed in finding higher paying employment. Workers who move down are possibly being pushed out or are more likely to be the consequence of employer initiated terminations. Is that something that that that that that comes into your interpretation of the results? Uh, if I can answer, at least, um, that's it's definitely something that that I discuss. And so kind of the two tests that I that I do in order to guard against exactly those kind of of, uh, possibilities. The first test is that if you remember that graph, uh, which showed exactly those quartile switches, the P events, uh, kind of came of these people is actually flat. So it's kind of it's it's it's showing that it's it's not like they're on particularly different wage trajectories before they switch. And so, which gives a little bit of confidence that, uh, these might be kind of, uh, uh, uh, you know, not two different workers. And the second and I think more rigorous test is, is that kind of matching, um, that, that I showed you in the second kind of piece of evidence for firm wage premiere, where we really do a lot of, where I really, um, uh, um, match quite finely and show a long pre period of something like six years, um, uh, including other firms, uh, to show that's kind of they aren't it's they're not systematically different wage, uh, uh, you know, firm wage trajectory. So these aren't picking up, as you say, people that are high, that are, you know, climbing up the, the, the job ladder versus kind of down the job ladder, uh, which in my mind certainly does happen, but I don't think that this, uh, that's, that's biasing these estimates of the firm, firm wage premiere. Um, I see there's a comment, uh, released, uh, by, uh, Sharon. Yes, it shouldn't be hidden. Yes, please. Okay. Let me just, uh, read it for, for the, for everyone. Um, uh, do you have insights, insights to share from the data on, uh, within or versus between, uh, uh, firm effects? Can some of the results and patterns be seen in wage ratios between executives and the low lowest paid workers within firms and LinkedIn and Donison's input with institutional architecture, not have the opposite effect. So, um, I'm going to be speaking out of scope here. I can't really say that much about, uh, you know, with, with, within firms. Um, maybe I looked at it at some point, but I, I, it's not, you know, part of the main study. But to talk about the, the institutional architecture, um, yeah, I do think that these bargaining councils, for example, and that's really what I'm, I'm spending most of my research time working on now. Um, uh, bargaining councils really do have, uh, a large effect on these, uh, uh, on, on this dispersion of firm wage premium. In some ways, um, it could increase these firm wage premium. Uh, um, if you think about, uh, you know, what types of firms are in these, uh, these bargaining accounts seem to be higher wage firms. And so even if you make, um, firm wage premium more, more, more uniform within the bargaining council, it means that it's more uniform, but, but more differentiated from the lower paid firms. And so that means that you get higher inequality on it. Uh, on the other hand, in some labor markets, it's, it's creating more, more, uh, more equality, uh, because it means that within that industry, you're getting more uniformity. Uh, and, and as something which I'm really investigating right now, uh, which is, uh, they are the substantial, uh, spillover effects, uh, from bargaining councils, uh, due to monopsony, uh, monopsonistic competition, uh, which really, uh, spills over onto the labor structure. As you say, the architecture, uh, um, which kind of, yeah, has effects on equality. Jury's out on higher inequality or less. Cheryl Lynn, does that answer? Do you have any comments? No, I, I just, I'm fascinated by it. And I think like Gerulov says it, it makes for lots of thinking over the next few weeks and months. Definitely. Thank you. Okay. Thank you very much, Issan, for, um, a really insightful presentation. Um, I'm very excited to read an updated version of the paper once it's done. Um, Gerulov, thank you to you for, um, for your discussion, which is also very insightful. And thank you to all of the participants, um, for joining today and have a wonderful afternoon.