Transcript of the talk, from a Whisper transcription of the audio. Auto-generated: speaker names in particular are unreliable. ======================================================================== # A Model of Global Firms and Capital Markets Authors: Loukas Karabarbounis, Bruno Pellegrino, Juliana Salomao Discussant: Ṣebnem Kalemli-Özcan Video: https://www.youtube.com/watch?v=wEt4iKiNNZQ&t=987s ## Talk (00:16:27 – 00:45:52) [00:16:27] People try as well. [00:16:32] What we have to do is between E and K, [00:16:34] it's easy to write also the rate of return [00:16:36] on financial capital, which we call new. [00:16:39] I'm gonna call it later expected rate of return. [00:16:42] So, new J is gonna be the expected rate of return, [00:16:44] which is how much profits you get per unit of equity, [00:16:47] okay, not per unit of capital. [00:16:48] And it's gonna be Chi times the physical return I saw. [00:16:52] And then on the kind of the supply side [00:16:55] of the asset market, we're gonna have here [00:16:56] is where the mean variance preferences come. [00:16:59] And there's gonna be a bunch of investors [00:17:01] that invest globally. [00:17:03] E is gonna collect the, [00:17:05] this is a vector of all the equities sent [00:17:07] in firms one, two, three until capital J. [00:17:11] And depends intuitively on the rate of return, of course, [00:17:14] relative to the risk-free rate. [00:17:18] These are the endowments of wealth. [00:17:20] Depends intuitively on risk aversion, [00:17:22] the variance covariance matrix of returns. [00:17:25] And the new aspect here is gonna be this delta. [00:17:27] Okay, so delta is the first time we introduce it. [00:17:29] It is the international financial friction. [00:17:31] This is, think of it as a capital control, okay? [00:17:35] That basically squeezes the returns [00:17:37] if I send from one country to another. [00:17:46] So in terms of the question, [00:17:47] like when we present the paper, we separate them, [00:17:49] but you can merge them because it's quasi-linear. [00:17:50] It doesn't really matter. [00:17:53] That as linearity, we had to think about wealth effects. [00:17:56] So eventually when we do welfare, [00:17:58] we merge them in one economic unit. [00:18:00] Yeah. [00:18:07] That's a friction. [00:18:09] I mean, chi is reduced form here, [00:18:11] but it's a rate of transformation. [00:18:13] We take it as given. [00:18:15] It's how much correct, correct. [00:18:24] But what is this? [00:18:25] All the magic goes here, okay? [00:18:26] How we clear all the global markets together. [00:18:28] So let me summarize that. [00:18:30] Okay, so we wanna clear J capital markets in the world [00:18:34] with how there is a demand function and supply. [00:18:36] The demand is what I showed you, [00:18:37] the downward sloping demand for capital [00:18:39] where instead of a row now I have new, [00:18:42] but I have this chi rate of transformation. [00:18:44] And it's really productivity times the difference [00:18:46] between price and marginal cost. [00:18:50] Now the price of the price vector here [00:18:53] has a collection of the P1, P2 to Pj. [00:18:56] This price is dependent on what everyone is doing. [00:18:59] It depends on the equity vector. [00:19:00] Why? [00:19:01] Because if a firm gets more equity, [00:19:02] they're gonna have more capital, [00:19:03] they're gonna produce more. [00:19:05] And then prices decline because of the demand function. [00:19:09] I'm gonna show you later. [00:19:10] So mu depends on E and then the second equation, [00:19:14] E depends on mu, taking as given the vector of wages [00:19:17] that pins down these two. [00:19:19] And then we have to clear J product markets. [00:19:21] And the demand comes from the hedonic demand system [00:19:25] that I described before. [00:19:26] This is intuitive. [00:19:27] This is kind of the highest marginal utility you can derive [00:19:31] is the willingness of consumers to consume [00:19:34] minus the price vector per tube by tariffs. [00:19:37] And kind of all the interesting interactions [00:19:39] are gonna come through this matrix sigma. [00:19:41] So sigma collects all the little sigmas [00:19:43] I showed you before in the utility function, [00:19:46] all the product similarities. [00:19:47] So all the interdependencies of this demand system [00:19:50] are gonna come here. [00:19:50] And this has variable demand, elasticity. [00:19:53] It has larger firms with lower elastics of demand [00:19:56] and all the standard stuff that are interesting here. [00:20:00] And the supply function is really, [00:20:03] the production function, after we say, [00:20:04] we recognize that capital depends on equity, [00:20:07] equity depends on returns, [00:20:09] returns depend on prices. [00:20:11] So if a firm has a high price, [00:20:13] they're gonna have high profits, [00:20:14] it's gonna receive more equity. [00:20:15] The return is gonna go up and capital is gonna produce [00:20:19] more output. [00:20:20] So this is kind of an upward sloping supply curve. [00:20:33] Because you go, [00:20:43] the price in Apple, [00:20:45] we'll ask you why, [00:20:46] but I'm sorry, [00:20:47] but I think the price will be increased [00:20:53] because the price of Apple, [00:21:02] I don't see it. [00:21:11] Because it's not a better expression. [00:21:15] It's a problem. [00:21:21] It's a problem. [00:21:23] So everything here, sorry. [00:21:26] Yeah, yeah, let me finish it. [00:21:28] Yeah. [00:21:33] So I have to read the first. [00:21:34] Now of course, [00:21:35] I'm gonna go home. [00:21:38] Yes. [00:21:39] There is no trade here. [00:21:40] It's like Apple hires Indians from India [00:21:44] and they pay the Indian wage. [00:21:47] And I'm gonna show you data on that later. [00:21:48] So there is like, [00:21:49] you work at home and you work for Apple. [00:21:51] So you don't have to move. [00:22:21] You can make sense, expect the elastic. [00:22:23] So if you don't know what it's actually gonna be. [00:22:29] If you are surprised with one, [00:22:30] it has, you know, [00:22:31] this is gonna be whatever it has to be [00:22:33] for the market to clear. [00:22:34] So think of it as a perfectly elastic supply. [00:22:38] The important assumption here is not that. [00:22:39] The important assumption is that no one spends [00:22:42] the resources earned by the outside goods [00:22:45] on the global firms. [00:22:54] Analytically conditional wages. [00:22:56] The only thing you need to put on the computer [00:22:58] is the labor market clear in a condition. [00:23:01] Why? Because these are okay. [00:23:03] So these have curvature, these two things. [00:23:06] Labor supply has curvature, [00:23:07] the non-traded labor demand, [00:23:09] everything else is linear and we're done. [00:23:10] Okay. [00:23:11] So before I move to the quantitative result [00:23:13] and show you one important kind of implication [00:23:15] of the model is gonna discipline [00:23:17] the whole kind of reallocation effects [00:23:19] we're gonna talk about later. [00:23:21] And these are, this is the same equation [00:23:23] shown in two formats. [00:23:24] And I think most people don't like [00:23:26] the matrix algebra I have to say, [00:23:28] but it does have something useful. [00:23:29] So I'm gonna insist that we present both. [00:23:33] So this is the vector of allocations of equity. [00:23:36] And this is the one representative, you know, [00:23:38] EJ in there. [00:23:40] And it depends in a very intuitive way [00:23:42] on the fundamentals. [00:23:43] Okay, so, you know, firms with a high B [00:23:47] relative to their cost of labor [00:23:51] kind of get more equity. [00:23:53] Firms that face lower delta get more equity. [00:23:57] Firms that face lower tariffs [00:24:00] get more equity and so on. [00:24:01] Okay, so that's very intuitive. [00:24:03] But kind of all the action is not in the fundamentals. [00:24:06] It is in the transmission matrix. [00:24:08] And the transmission matrix you see in a nice way here [00:24:12] that it depends on these two matrices. [00:24:15] V is the variance covariance matrix of returns. [00:24:18] And sigma is the product similarity [00:24:22] that I discussed before. [00:24:23] Okay, so the reallocation effects here [00:24:26] are gonna depend on these matrices. [00:24:29] And, you know, the kind of a key summary statistic. [00:24:32] It's not sufficient, but it is qualitatively [00:24:34] what you really wanna look at. [00:24:36] It's gonna be what we call the semi-elasticity [00:24:39] of equity with respect to the risk free rate. [00:24:41] Okay, so here the experiment is gonna be, [00:24:43] suppose there is a common shock in the world. [00:24:45] And the common here is the R0, [00:24:46] but when we remove friction, it's gonna be almost common. [00:24:49] Who gains and who loses? [00:24:50] And the answer is, you know, the people, [00:24:54] the firms for which these terms are positive, [00:24:58] they're gonna expand. [00:25:00] And the firms for which this is negative, [00:25:02] they're gonna contract. [00:25:04] Okay, and these terms here, though, [00:25:05] I guess depend on these, you know, [00:25:07] V and sigma as I'm gonna discuss. [00:25:09] Okay, so I'm gonna come back to this object. [00:25:10] The semi-elasticity of equity with respect to the interest rate, [00:25:13] because it does determine, [00:25:14] at least qualitatively, the reallocation effects. [00:25:20] Which, you know, given the kind of the time we have, [00:25:24] I'm not gonna do justice to all the things we did here. [00:25:26] Just a quick snapshot. [00:25:28] So we're gonna use WorldCrop data, [00:25:30] this is pretty standard for listed firms. [00:25:33] GORF data, this is the distribution of revenues [00:25:35] across countries. [00:25:36] The Revello labs that they started to be used now [00:25:40] in many papers, give us a distribution of labor [00:25:43] across firms, you know, data stream for returns, [00:25:46] global capital allocation for the allocation [00:25:47] across countries of equity and bonds. [00:25:50] And the tariff database, okay, you know, for tariffs. [00:25:54] We're gonna use World Bank data for shipping costs. [00:25:58] Sorry, for non-tariff measures and OECD data [00:26:02] for shipping costs, BIS, ILO and so on. [00:26:04] Okay, so the cover is gonna be [00:26:06] what I mentioned in the introduction, [00:26:08] it's kind of big. [00:26:09] And at a high level, what we're trying to do is the following. [00:26:14] So we're gonna first, there are two parts in our methodology. [00:26:16] First, we're gonna try to put numbers [00:26:19] into the endogenous variables on the bottom. [00:26:21] And some we're gonna directly measure in the data. [00:26:23] Things like equity market capitalizations, [00:26:26] gonna be the EJ, or staff costs, [00:26:28] it's gonna be the labor cost that the firm pays. [00:26:31] But some we're not gonna be able to measure directly [00:26:33] and we're gonna infer. [00:26:34] For example, the expected rate of returns, [00:26:36] I'm gonna show you in the next slide. [00:26:37] And then once we have the data, [00:26:39] we're gonna invert the model [00:26:42] and we're gonna say, what is the inversion of the model? [00:26:44] Such that the model exactly reproduces a given cross-section, [00:26:47] the cross-section is gonna be 2023 of firms. [00:26:50] And we're gonna match profits, capital, the labor allocation, [00:26:54] the allocation of revenues, [00:26:56] the allocation of equity and so on, [00:26:58] firm by firm, together with other. [00:27:00] Okay, so. [00:27:01] Yes. [00:27:09] Sorry, something's on the screen. [00:27:11] I'm not sure about that. [00:27:13] But it's kind of confusing. [00:27:15] It's not gonna make sense. [00:27:16] If you go at the bottom of the distribution, [00:27:18] it's a bunch of zeros. [00:27:20] It is very, I'll show you an example. [00:27:22] Okay, so let me show you some examples of quantification. [00:27:25] So an important element is the sigma matrix, as I mentioned. [00:27:28] And here we're just gonna do [00:27:29] what Bruno did in his Joe Market paper. [00:27:31] Okay, so Bruno did a text analysis [00:27:34] of extended business description. [00:27:35] Basically you read what Apple does, [00:27:37] what Samsung, and you correlate. [00:27:38] It's a cosine similarity. [00:27:39] Okay, if there are commonalities in what they do, [00:27:43] the sigma for this pair is gonna be high. [00:27:45] So what we do here is we extend that [00:27:46] and we do it for the whole world. [00:27:49] Our tower measure is gonna include tariffs, [00:27:51] non-tariff measures and shipping costs. [00:27:54] The returns, the expected rate of returns, [00:27:57] we're gonna make an identifying assumption [00:27:59] that within a country, there are no financial frictions. [00:28:03] And then you know, if you see a company [00:28:06] conditional on the variance and the risk aversion, [00:28:09] that's getting a high share, [00:28:11] it must be that has a high expected return. [00:28:14] And the international frictions are gonna basically, [00:28:17] if we know now the rate of returns, [00:28:18] we're gonna invert this equation, [00:28:20] we're gonna say if a company has a higher expected return, [00:28:24] but foreigners don't invest, it has to be that delta is high. [00:28:27] Okay, let me give you an example. [00:28:28] Russian firms, or I'll have high new, okay, [00:28:31] higher than the US. [00:28:32] Why is that because Russians love investing [00:28:35] in Russian companies? [00:28:37] Despite these companies having high variance [00:28:39] and poor heads in profits. [00:28:41] But no one else invests in Russia. [00:28:43] So it has to be that they have high deltas. [00:28:45] Okay, that's kind of the essence here. [00:29:14] So you care. [00:29:16] That's what you care about this. [00:29:17] There are all the interactions in there [00:29:19] that you care about. [00:29:20] But look at the equation. [00:29:26] This is effective. [00:29:29] We can do that. [00:29:30] We can do that. [00:29:31] We can do interactions. [00:29:33] We can see the same. [00:29:35] So, for example, [00:29:38] what is equity? [00:29:41] Sounds like equity. [00:29:42] Like equilibrium equity. [00:29:47] Be my own omega JJ, but what everyone else is doing. [00:29:53] Some results, okay. [00:29:54] So, you know, this is the demand elasticity [00:29:57] implied by our model, [00:29:59] projected on size. [00:30:00] And I'm gonna do a hundred beams here. [00:30:02] And this is very intuitive, right? [00:30:03] So it ranges from minus five, [00:30:05] let's say minus six to minus two. [00:30:08] Big firms have lower demand elasticity, [00:30:10] higher markups if you want, [00:30:11] lower labor shares, okay? [00:30:12] Higher profit margins. [00:30:14] And this is the eight semi-elasticity [00:30:17] that I showed you before, okay? [00:30:19] So this is increasing in size. [00:30:21] And notice, you know, the zero here goes through probably, [00:30:24] you know, around the, I don't know, 75% tile or something, [00:30:27] or 80. [00:30:28] So again, a positive number means [00:30:31] a positive development like a decline in the risk-free rate [00:30:34] or a decline in friction, so trade cost. [00:30:37] Reallocates, okay? [00:30:39] And more equity to these guys [00:30:40] and less equity to these guys. [00:30:42] So these guys basically are gonna be squeezed out [00:30:44] from the liberalization. [00:30:45] And these guys are gonna kind of expand, okay? [00:30:48] So it's quite heterogeneous across firms. [00:30:53] But you kind of have some examples of the firms. [00:30:56] We just took the 12 largest US firms by market cap in 2023. [00:31:01] This is the distribution of revenues. [00:31:03] See, this is how much Apple sells domestically [00:31:06] relative to internationally. [00:31:08] And you see, you know, these guys sell everywhere. [00:31:10] They hire labor, you know, everywhere [00:31:14] because this is not a hunger, except for the, [00:31:16] except for a run and buff it, okay? [00:31:18] This is intuitive. [00:31:20] If you ask us, you know, [00:31:21] what are the highest expected returns in 2023? [00:31:24] We would tell you Tesla and Nvidia doesn't sound crazy. [00:31:28] They have low demand elasticities [00:31:30] and they have positive semi-elasticities. [00:31:32] These guys are gonna expand. [00:31:34] If you're curious, what's the kind of the largest firm [00:31:36] internationally with the highest rate of return? [00:31:38] This is Novo, okay? [00:31:39] The producer of OZEMPIC 2023. [00:31:41] And back then it was not so obvious. [00:31:44] Okay, we do counterfactuals. [00:31:48] Like the model doesn't require crazy frictions [00:31:52] to kind of rationalize the data. [00:31:53] So here we show you the cost that the households pay [00:31:57] in terms of income for trade costs, [00:32:00] which include, again, tariffs, non-tariff measures, [00:32:03] and shipping costs. [00:32:03] And this is the financial frictions, okay? [00:32:05] So this is like a thousand dollars [00:32:07] for the average household, okay? [00:32:08] So the model rationalizes, you know, [00:32:10] the kind of the data without crazy, [00:32:13] at least in equilibrium frictions. [00:32:16] Can you? [00:32:20] Yeah. [00:32:26] That's the... [00:32:28] So this is matrixes. [00:32:30] So it turns out the big guys are not, you know, [00:32:33] the small guys are face a lot of competition [00:32:35] from the big guys. [00:32:37] Their sigma is more similar. [00:32:39] It's one example, but it depends on everything really. [00:32:41] But, you know, the variance covariance matrix [00:32:44] and the sigma are kind of the important determinants. [00:32:47] And that, again, we're gonna measure external to the model. [00:32:52] Axial, which is a financial liberalization, [00:32:54] let's look at the cross-section of firms, okay? [00:32:55] So here we're gonna remove the deltas. [00:32:57] Baseline is the black. [00:32:59] The dash line is the counterfactual. [00:33:02] So first thing is, you know, equity expands, okay? [00:33:04] And it is intuitive because, you know, [00:33:05] we have an elastic choice between equity and bonds. [00:33:08] And so investors are gonna reallocate resources, [00:33:11] their wealth to equity when the financial frictions. [00:33:14] But the more interesting result is look at the dispersion, [00:33:16] okay? [00:33:17] So the change in the standard deviation [00:33:20] is 25 percentage points. [00:33:22] And the 1090 moves up by 57, okay? [00:33:25] This is the reallocation I was talking about. [00:33:27] And the same holds with revenues, [00:33:29] with the only difference that the kind of a mean [00:33:30] of revenues goes down. [00:33:32] So why is that because production expands [00:33:34] but prices fall, okay? [00:33:35] So consumers benefit in this world from lower prices. [00:33:40] But again, the dispersion in revenues across firms [00:33:42] increases quite a bit. [00:33:46] Effects of the, in the aggregate, [00:33:49] the distributional statistics across countries [00:33:52] of the financial liberalization. [00:33:54] And let me start with the first panel, [00:33:55] which is the total welfare effect, okay? [00:33:57] So the median country gains 5.5%. [00:34:03] The mean is seven. [00:34:04] There is a lot of dispersion. [00:34:05] I'll talk a little bit about who gains later. [00:34:08] But you know, some are three, some are 10 and so on. [00:34:10] Now we decompose the welfare gain [00:34:13] into the underlying components. [00:34:14] And the point of these figures is the second panel [00:34:18] captures the welfare effects from changes [00:34:20] in consumer surplus due to increased variety. [00:34:23] It's the Q, okay? [00:34:25] So the second panel looks very similar to the first. [00:34:27] It says that all the welfare gains [00:34:29] almost all come from increasing consumer surplus. [00:34:33] And the basic idea here is you remove the frictions, [00:34:35] okay, Brazilian firms now become [00:34:37] their effective marginal cost falls, okay? [00:34:39] They can produce more, you equity finance them. [00:34:41] They're gonna produce more. [00:34:42] You send them, they send their goods to the US. [00:34:45] US consumers benefit from lower prices and higher variety. [00:34:50] Now the other adjustments, you know, [00:34:51] the non-traded goods go down, but that's small. [00:34:55] The labor surplus, which is related to wages, [00:34:57] don't move much. [00:34:59] Returns, equilibrium returns change, okay? [00:35:02] But again, nothing compared to the total effect. [00:35:04] And then there is diversification, okay? [00:35:07] As you change your portfolio, [00:35:08] there is a variance component in preferences, okay? [00:35:11] That changes, but again, this is small, okay? [00:35:13] So the entire gains come from the consumer surplus. [00:35:20] Realization. [00:35:21] We're just to be clear here. [00:35:23] Trade has, trade costs have three components, tariffs, [00:35:26] non-tariff measures and seeping costs, [00:35:28] and we remove the first two. [00:35:30] So our view is seeping costs are technological constraint. [00:35:33] We don't change them. [00:35:34] And then we just change things related to policy. [00:35:37] So seeping, it's a liberalization. [00:35:40] This spreads out the distribution quite a bit. [00:35:42] One difference is equity doesn't expand here. [00:35:45] It's not that investors, they still have the same portfolio [00:35:49] shares between bonds and equity. [00:35:51] It's just within equity. [00:35:52] The big guys get all the equity [00:35:54] and the small guys get squeezed, okay? [00:35:56] This is the dispersion going up. [00:35:58] And the same with revenues, [00:35:59] with one difference related to financial liberalization, [00:36:02] revenues go up. [00:36:02] So why is that? [00:36:03] Because financial liberalization is a supply shock, okay? [00:36:06] Makes the marginal cost lower. [00:36:08] Whereas this is a demand shock. [00:36:09] So both price and quantities go up, [00:36:11] so revenues have to go up. [00:36:18] It's now from a trade liberalization. [00:36:23] Percent. [00:36:24] I would say that's comparable to what the trade people [00:36:26] have done, but it's a different experiment, okay? [00:36:30] So an ACR formula if you take, [00:36:33] it moves the domestic CR, it doesn't tell you [00:36:35] if it moves because of tariffs, seeping costs, [00:36:37] or preferences. [00:36:38] So here we are very experienced. [00:36:39] We keep preferences fixed. [00:36:41] We keep seeping costs fixed. [00:36:43] We keep substitutional assets fixed [00:36:44] and we just remove the tariffs, okay? [00:36:46] Despite that, the numbers are close. [00:36:48] I guess the closest way to think about the analogies [00:36:51] to calculate the trade elasticity in our model, [00:36:54] we get a trade elasticity of minus four. [00:36:56] These are maybe on the lower bound of the literature, [00:36:59] but many papers like Mike has a minus four, I think. [00:37:02] If I remember your paper, and Costino in his handbook [00:37:06] and other people have minus five. [00:37:08] So this is a reason of trade elasticity. [00:37:11] Yeah. [00:37:12] The P10 refers to countries. [00:37:14] Not, yes. [00:37:16] So these are, okay. [00:37:17] Every country gains, but there. [00:37:19] Every country gains, okay. [00:37:23] Most of the welfare gains, [00:37:24] gains come from consumer surplus going up. [00:37:27] One difference is, and that's in line with the trade, [00:37:30] people, right? [00:37:30] Weights just go up. [00:37:32] So the surplus, here the surplus means labor income [00:37:35] minus this utility of labor, excuse me, [00:37:38] contributes quite a bit as well. [00:37:39] So, and what drives this difference [00:37:41] is you have a lot of global firms, you gain more. [00:37:43] Is that, going back to then, [00:37:46] then ask whether global firm as a country [00:37:48] belongs to a country. [00:37:50] So what drives this difference across countries? [00:37:53] Okay, so, [00:37:55] I've seen how much your consumption trade cost falls. [00:38:03] I mean the trade cost and an expenditure weighted. [00:38:05] So some countries have huge tariffs to begin with. [00:38:08] So then it's all on the cost of the, [00:38:12] whereas for that, for the GDP, [00:38:14] whatever is production is the production. [00:38:16] So this says, maybe American firms are taxed more than others. [00:38:20] And whereas this says maybe American consumers [00:38:23] face higher tax. [00:38:29] Done, so the first is, [00:38:31] there is an interesting interaction here [00:38:33] between the two frictions or the two liberalizations. [00:38:36] So let me pick one row and explain the last one. [00:38:39] So in the baseline model, [00:38:42] the typical, what's that? [00:38:44] Okay, fine, I don't need them, I didn't fight. [00:38:47] So the typical, [00:38:49] the typical poor country, [00:38:54] gains 4.5%. [00:38:57] Now, if you did the same trade liberalization experiment [00:39:01] in a model without financial frictions, [00:39:04] you would get kind of a quite different number, 3.1. [00:39:08] And the same here, for the financial liberalization, [00:39:11] the typical poor country experiences, [00:39:14] a substantial gain of 7.4%, [00:39:17] whereas if you did the same financial liberalization experiment [00:39:20] in a model without trade frictions, [00:39:22] the number would be kind of 1.6% that's point different. [00:39:26] Now the difference is not so big for the rich countries, [00:39:29] okay, there is a little bit, [00:39:30] but especially for the poor emerging markets, [00:39:34] is a good example, you know, we're gonna be here, [00:39:36] the difference is big. [00:39:39] Okay, so what do we learn from that? [00:39:40] We learn that the liberalizations in our model [00:39:43] are substitutes. [00:39:44] And intuitively what's going on is, [00:39:46] look, the consumer doesn't care, [00:39:47] in the end the consumer cares about the utility [00:39:49] from consuming these global firms. [00:39:51] And both tariffs and financial frictions [00:39:54] are gonna basically raise the price of inputs. [00:39:56] Okay, so once you remove the one, [00:39:58] you've done most of the work of, you know, [00:40:00] removing the other. [00:40:01] Or another way to see it is, you know, [00:40:03] emerging markets have larger welfare gains [00:40:06] because the frictions amplify, you know, each other [00:40:09] and emerging markets have both higher trade costs [00:40:12] and financial frictions. [00:40:16] So last time that I have to conclude, [00:40:19] you know, for the reallocation effects, [00:40:21] but also the welfare, which I'm not gonna show you now, [00:40:24] the underlying details of constructing the Sigma [00:40:27] and the V matrix matter a lot. [00:40:28] Again, this determines the kind of the reallocation [00:40:31] as I show you. [00:40:33] So I'm gonna illustrate it in one example. [00:40:36] So I'm gonna show you how revenues change, [00:40:38] distribution of revenues under financial liberalization [00:40:41] and trade liberalization for the baseline, [00:40:44] which is this, you know, very dispersed distribution. [00:40:48] And then I'm gonna show you two nested versions [00:40:49] of the model, while in which we set the Sigma matrix to zero. [00:40:53] Okay, so there is no substitution. [00:40:56] And the other where we set the diagonal of the V to zero. [00:40:59] So there is no asset market correlation. [00:41:01] And as you can see in these two nested versions [00:41:03] of the model kind of the distribution is very different. [00:41:06] So first you don't have a lot of losers. [00:41:09] So now moving under a very large number of firms, [00:41:13] not big firms, small firms, [00:41:14] but they kind of go down after the liberalization. [00:41:17] The same holds under a trade liberalization, [00:41:20] but only for the, [00:41:21] and this is more obvious, perhaps, result. [00:41:23] You know, without product marketing competition, [00:41:25] you would get kind of a bigger effect [00:41:27] and almost everyone would win. [00:41:29] Okay, whereas in our model, you know, some firms lose. [00:41:35] So let me just go through the first part. [00:41:39] Okay, so in this model, [00:41:42] financial liberalization generates a median welfare gain [00:41:45] of 6%. [00:41:47] This is a little bit higher, I should say, [00:41:48] relative to the literature. [00:41:49] Okay, so the financial liberalization literature [00:41:52] is all over the place. [00:41:54] And we have some experts here. [00:41:55] Maybe they can explain why. [00:41:57] It's not like the trade people who have kind of agreed [00:42:00] what are the welfare gains. [00:42:01] Okay, so 6% is relatively high. [00:42:04] There are higher, like the recent paper of Clay [00:42:06] in Manitale who has a higher number, [00:42:08] they have different mechanisms. [00:42:09] They talk about dynamics. [00:42:11] We talk about reallocation and product marketing competition. [00:42:14] It's driven by, you know, the higher consumer surplus. [00:42:16] That's a new mechanism we believe in the literature. [00:42:19] They're larger in emerging markets. [00:42:20] That I think it's intuitive and accompanied [00:42:23] by a larger allocation of economic activity [00:42:25] towards the larger firms. [00:42:27] Now the trade side, again, I would argue [00:42:30] the welfare gains are kind of reasonable [00:42:32] in light of what the trade people think they are. [00:42:35] We do see again reallocation. [00:42:37] This is different, the mechanism, [00:42:39] however, is different relative, say, in Melitz. [00:42:42] You know, people think about entry, exit [00:42:44] and all these things and selection. [00:42:46] Here the reallocation is really about [00:42:48] variable demand elasticity and bilateral capital flows. [00:42:51] So the mechanism is new. [00:42:54] Although it's similar, the reallocation [00:42:55] towards larger firms. [00:42:58] The kind of jointly, even if you don't care [00:43:01] about trade and you are an international finance person, [00:43:05] we would argue you have to have trade frictions [00:43:07] in your model. [00:43:08] And if you are a trade person, [00:43:10] we want to argue you want to have financial frictions [00:43:12] because doing the counter-factuals, as I saw you, [00:43:15] produce very different results. [00:43:17] And finally, all these, the welfare effects [00:43:19] and the reallocation depend on the product market [00:43:23] substitution and the capital market correlation [00:43:25] which we quantify. [00:43:26] So it takes some work to do that, but it is sensible. [00:43:56] Everything is linear here, [00:43:57] so I don't want to take the limit, I think, I guess. [00:43:59] But I think the point here is, [00:44:03] let me not take the limit, but let me say, [00:44:04] suppose trade frictions are very high [00:44:06] and you cannot consume Brazilian. [00:44:13] Then, you know, financial liberalization [00:44:15] is gonna make the production cost [00:44:18] of Brazilian firms lower. [00:44:20] And maybe they cannot cause export it [00:44:22] for a given level of trade frictions. [00:44:26] But because it's linear, [00:44:27] I cannot take the limit as you want. [00:44:48] Actually, we tried at some point, [00:44:50] we can go back, I believe, 2014 or, [00:44:55] most of the data is 2023. [00:44:58] We tried at some point to go back in time, [00:45:00] we have missing data for a lot of variables. [00:45:02] We could, when done, we could do, [00:45:12] yeah, yeah, no, no, this is, [00:45:13] that's why I mentioned, you know, [00:45:14] Nvidia and Tesla kind of smell right to me, [00:45:17] but yes, we could do it more formally, that's good. [00:45:26] I think it's very possible. [00:45:28] Seven times, it's almost, [00:45:34] same as the other one. [00:45:37] So, Lucas, like, I finished 15 minutes early, [00:45:40] gonna have many five minutes at the end. [00:45:42] I'm gonna do it for questions. ## Discussion (00:45:52 – 01:02:47) [00:45:52] I'm sorry, I'm looking out. [00:45:58] In my case, I guess I have to ask. [00:45:59] All right, okay. [00:46:00] I mean, you can see that I love this paper [00:46:03] because Lucas said, I love that line, right? [00:46:07] You said if you're a trade person, [00:46:08] you have to care about financial frictions. [00:46:10] If you're an international finance person, [00:46:13] you have to care about trade friction. [00:46:15] This is great. [00:46:17] So that's what is great about this paper. [00:46:22] So how do I know this? [00:46:52] Hold up, anyway. [00:46:53] That's, [00:46:56] I'm great at this. [00:46:57] Okay, it seems perfect. [00:46:58] So this is great. [00:46:59] So, I mean, super rich, super interesting paper. [00:47:03] Let me just summarize in one slide. [00:47:05] Question, how do trade and financial frictions [00:47:08] jointly shape the allocation of activity? [00:47:11] But you have to have this special word there [00:47:13] because it's a static model. [00:47:15] It's a cross-sectional exercise. [00:47:17] But then, with that still, [00:47:19] this is about allocation of activity [00:47:21] at cross firms, countries, [00:47:23] and what are the welfare gains from removing them? [00:47:26] Right, both on the trade side and the finance side. [00:47:29] So very, very obviously ambitious question. [00:47:31] That requires a very rich model. [00:47:33] That's why they do it in a static setup. [00:47:35] So the good side is going to be this [00:47:37] in a quadratic demand, very, very important. [00:47:40] It's really, all the supply and production side stories [00:47:44] in a way kind of goes through general equilibrium. [00:47:47] Mark was asking this question. [00:47:48] So demand side is going to be very, very important. [00:47:51] But of course the asset side comes in [00:47:53] through these mean variance portfolios. [00:47:55] So in that sense, you are really not missing anything. [00:47:56] So it is a very rich model. [00:47:59] Now, collaboration is going to be hard [00:48:01] on a rich model like that because it's a static model. [00:48:04] They are going to use tons of data. [00:48:06] It is like an amazingly ambitious exercise, [00:48:09] but it is going to be one cross-section, [00:48:10] going back to Esther's question. [00:48:12] So 48 countries, 23,000 listed firms, [00:48:15] like very, very good coverage [00:48:17] in terms of trillions of wealth. [00:48:19] But the year part is going to be a little bit mixed, right? [00:48:23] I mean, most of the data is going to be 2023. [00:48:26] So you'll see in the parentheses, [00:48:28] some data is going to be from 2017. [00:48:30] Like the data comes from global capital allocation project, [00:48:33] but they rescale that to 2023. [00:48:36] I mean, tariff data is going to come from 2021. [00:48:40] So it is, you know, they do a lot of robustness checks there, [00:48:43] but you know, that's what it is. [00:48:46] Now the results are super interesting. [00:48:48] Financial liberalization median gain [00:48:49] around 5.5% of labor income. [00:48:52] You have 3% on India, US India 10. [00:48:56] Trade liberalization median gain is 3.8%. [00:48:58] And again, as Lukas has said, [00:49:00] trade number is consistent with the trade literature. [00:49:03] Financial number is a little bit higher [00:49:05] than the financial literature. [00:49:07] But the new thing here is of course, [00:49:09] but Lukas, you have to have both frictions, right? [00:49:12] Because they are both going to act [00:49:14] in a way that give everything a large firm. [00:49:17] So there is this concentration of large firm [00:49:19] and that's, you know, behind that is this [00:49:22] amplification story to frictions amplifiers. [00:49:24] Okay. [00:49:25] All right. [00:49:27] Yeah, it's freezes. [00:49:29] Okay. [00:49:35] For a new workhorse that finally puts firms [00:49:38] in the international finance [00:49:39] and capital detergents from trade model. [00:49:41] I love it. [00:49:42] Enormous data for transparent identification. [00:49:45] So, and it is really an important contribution. [00:49:47] Why? [00:49:48] Because trade and finance are studied [00:49:49] in separate silos in academia, [00:49:51] but of course this is exact opposite in the post debate. [00:49:54] Everything we talk about and care about in this room [00:49:57] requires both. [00:49:58] So just like very, very clear contribution here. [00:50:01] So what my comments going to be [00:50:03] if I make this thing work. [00:50:06] Oh my God, this is going to drive me nuts. [00:50:08] Okay. [00:50:09] So first, you know, some of you raised this question. [00:50:11] So this V various covariance metrics [00:50:13] of expected returns, that's actually pretty important. [00:50:16] It is going to be doing some Hebel lifting here. [00:50:18] And I'm going to try to explain that. [00:50:21] So the financial wage is also important. [00:50:23] It's kind of residual. [00:50:25] And it is important that everything is loaded [00:50:27] on this international wage. [00:50:29] There isn't any home domestic frictions. [00:50:32] Now I'm going to ask like this two questions [00:50:36] in terms of like, you know, [00:50:37] how consistent the models predictions [00:50:40] with the empirical literature we know so far. [00:50:43] I mean, I put it that data consistency. [00:50:45] I'm going to ask this equity is the capital [00:50:47] where does foreign capital actually go into data? [00:50:49] And if trade and finance subsidies or complements, [00:50:52] you know, this is a big literature. [00:50:53] We have other models saying they are substitutes [00:50:57] with different stories than this paper [00:50:59] or other models in their complements. [00:51:00] And also how to interpret this large firms [00:51:02] get everything concentration result. [00:51:04] Is it efficient reallocation? [00:51:06] And it is in the paper, but is there any room [00:51:09] also to think it as a misallocation result? [00:51:15] And the result is this financial liberalization result. [00:51:18] So I'm going to focus on that for a minute. [00:51:20] So I read the paper, I read it again. [00:51:22] I mean, okay, this is opposite of the textbook, right? [00:51:25] So less value from diversifying the bigger the gains. [00:51:28] So less value from diversifying the bigger the gains. [00:51:31] It's really, you have to, you know, [00:51:34] go through it a couple of times to wrap your head around. [00:51:37] Especially, you know, people like me and Fabio Stiolei [00:51:39] who work a lot on risk sharing and intention [00:51:41] of diversification. [00:51:42] So these are the welfare gains [00:51:43] from financial liberalization. [00:51:45] So the baseline with a V, with that given V [00:51:48] of the variance covariance return matrix. [00:51:52] So you, you know, make it zero. [00:51:54] Then you are going to lose 80% of the gain. [00:51:56] So it's very, very upside down, right? [00:51:59] So the textbook tells us less correlated [00:52:01] with my foreign stock, with the foreign stock. [00:52:04] The more I'm going to get from integration, [00:52:07] here you make them uncorrelated, this column, like, [00:52:11] and then like, you know, you, [00:52:13] but that's the most valuable to you to diversify. [00:52:16] But then you don't get any gains. [00:52:18] Why is that? [00:52:19] So high correlation means home stocks are a bad hedge, [00:52:22] of course. [00:52:24] That's why, like, you know, it goes out of the way. [00:52:26] But in their story, when that happens, [00:52:28] residents hold them domestically because of this mu, [00:52:32] expected acute return is high, which is unobserved. [00:52:35] Foreigners are not holding them. [00:52:37] So interpretation of that, [00:52:38] there must be a very large international [00:52:40] financial friction here. [00:52:41] That's why foreigners are not holding them. [00:52:44] That's why there is a large gain from women, right? [00:52:46] So though, so for you to wrap your head around this, [00:52:49] this is not a diversification gain result [00:52:52] as in our papers, right? [00:52:53] It is a result that, you know, this wage is costly. [00:52:55] There is a large wage needed to rationalize the home bias, [00:52:58] given the variance covariance metrics. [00:53:01] Okay, that's the result. [00:53:02] So, like that. [00:53:08] If you put it in the, like, you know, [00:53:09] tell and rate type of literature, [00:53:11] correlations rise, you integrate more. [00:53:13] Yes, we are, you know, nobody's that perfect [00:53:15] integration, but you integrate more thing like that [00:53:17] and your correlation rise, [00:53:18] you become more and more correlated to stock markets. [00:53:21] So that means like you're like, you know, [00:53:23] like at all target the border. [00:53:24] So it is very interesting result. [00:53:26] I mean, it takes time for you to handle it, [00:53:27] but I think it's super interesting. [00:53:29] But at the end of the day, it is what it is. [00:53:30] Like, you know, Korean and US stocks move independently. [00:53:33] There will be, you know, no gain [00:53:35] from opening the capital market between Korean years. [00:53:37] Okay, so that, I mean, I'm inquired by this. [00:53:41] It's very interesting. [00:53:42] So I said, okay, so let's, you know, dissect this result. [00:53:45] Let's first start with this variance covariance metrics. [00:53:47] This is doing, they have a lifting here. [00:53:49] So this is 23,000, 23,000 metrics, right? [00:53:52] From daily returns in a single year. [00:53:54] In 2023, this is going to drive most of the stuff. [00:53:58] All friction expected returns [00:53:59] are going to be inferred from that single year. [00:54:02] There's nothing wrong with that. [00:54:03] I mean, that's actually a good thing [00:54:04] from the model's perspective, right? [00:54:05] Model's the status of, [00:54:07] this is how you should be doing this calibration. [00:54:09] But 2023 is an unusual year, right? [00:54:12] Think about 2023, Russia was under sanction. [00:54:14] Policy rates were at their peak [00:54:16] after the highest inflation in like four decades. [00:54:19] AI boom was underway. [00:54:20] In video and test, like here, [00:54:21] the highest expected returns [00:54:23] among the largest US funds. [00:54:24] And US share of global portfolios [00:54:26] rose sharply in that tree. [00:54:27] Okay, so it is, you know, it is, [00:54:29] you know, I would do much more here. [00:54:32] Also because this is going to combine [00:54:34] with the fact that holdings and returns [00:54:35] are now going to come from different years. [00:54:37] There is a standard result in finance, right? [00:54:39] Non-synchronous city. [00:54:40] And this is going to buy us everything down. [00:54:43] So this requires a little bit more. [00:54:45] I mean, you know, they are already doing it [00:54:46] using different years, [00:54:47] but one other suggestion I thought of, [00:54:49] they don't have to do it, just a suggestion. [00:54:50] Factor model, weekly returns, five-year window. [00:54:53] You have to, you know, think about a little bit, [00:54:55] like where am I, what am I going to do with the currency risk? [00:54:57] I think that belongs to me also, [00:54:59] but you know, that's something to think about. [00:55:01] All right, dissecting this result now. [00:55:04] Let's go, so this is the variance covariance matrix part [00:55:06] of this result. [00:55:07] Now there's this international wage part, [00:55:10] the financial friction, this delta. [00:55:13] Again, so this is the average wage for foreign space. [00:55:15] So, and it is right. [00:55:17] This is why it is very interesting result. [00:55:18] But Luca said like, look, [00:55:19] we don't require a large friction to rationalize that. [00:55:22] That is correct, right? [00:55:23] They don't require that. [00:55:24] So it is not about like really level of the friction, [00:55:27] but the fact that it is still a residual [00:55:29] and it absorbs everything. [00:55:31] So it is not, there is any problem in the number itself, [00:55:35] but the interpretation behind it. [00:55:36] What is in that number, right? [00:55:38] This is a residual, it's going to have, you know, [00:55:40] everything that mean variance investor is missing [00:55:43] is in this wage, capital controls, currency risk, [00:55:47] hedging labor income, real exchange rate, [00:55:49] information governance, institutions, [00:55:51] you know, I put a bunch of papers there. [00:55:53] All this stuff is going to be there. [00:55:54] It's just fine. [00:55:55] You can say, yes, okay, fine. [00:55:56] But, you know, when you look at the comparison, [00:55:58] there's this US puzzle, right? [00:55:59] A bigger wage on the US than in India, [00:56:02] but the US share of portfolios surging after 2007. [00:56:06] So that's not, you know, I would say [00:56:08] the matter is not quite ending up there. [00:56:10] One suggestion here, you know, [00:56:12] do on the finance side, [00:56:13] is that what you do on the trade side? [00:56:15] Five minutes, okay. [00:56:16] So project this financial wage on capital controls, [00:56:18] tax, trading distance, all that stuff [00:56:20] and liberalize only the policy part. [00:56:22] The policy part will be the capital controls part. [00:56:24] That's the only thing you can on the bill. [00:56:26] Okay, so why no domestic friction? [00:56:29] I mean, this is, you know, kind of like, you know, [00:56:32] I like domestic frictions as you know, my work. [00:56:35] So this new is the expected equity return [00:56:37] that is not observed. [00:56:38] It is going to be inferred from this, [00:56:40] how much domestic investor hold of their own firms, [00:56:43] how much Koreans hold their own firms. [00:56:45] So this type of identification, [00:56:46] it is very well done. [00:56:47] It is very robust, but that is needing this assumption. [00:56:50] You cannot do this identification with frictions at home. [00:56:54] So then there are no frictions at home. [00:56:56] This is of course, you know, at all with the data [00:56:58] and it's not just an emerging marketing, [00:56:59] both emerging market and advanced economy, [00:57:01] domestic investors do face frictions, [00:57:03] size dependent financial constraints, [00:57:05] going back to my paper with Lucas, [00:57:06] week minor to share older protection, [00:57:08] financial, I mean, you name it. [00:57:10] So this type of stuff is going to hit the residents [00:57:12] and the foreigners alike. [00:57:13] So this is not something that stopped at the border. [00:57:16] And it is important, it matters, right? [00:57:17] Because when you look at this new, [00:57:18] you are going to get some numbers [00:57:20] that are really too high in a sense [00:57:22] that foreign wage is going to be large, [00:57:24] that means gains from the Brazilian too large. [00:57:26] But I mean like 7.9 for Russia, 5.7 for Brazil. [00:57:29] And this combined with the fact [00:57:31] that the counterfactual is not the polls, right? [00:57:33] When you set this thing to zero, [00:57:35] because the residual, you are removing everything, right? [00:57:38] You are removing the bad institution, [00:57:39] capital control, everything, [00:57:40] but the polls, the choice is only the capital control. [00:57:43] So my suggestion here is a very simple exercise. [00:57:45] They already have the data, [00:57:47] look at the dual listed firms. [00:57:48] So the firms listed in the Turkish stock market [00:57:50] and firms listed in the US stock market, [00:57:53] that will be a nice way of at least putting some number [00:57:56] to separate this domestic friction from the foreign friction. [00:57:59] And it is important. [00:58:00] I mean, I'm going to push on this point a little bit. [00:58:02] Both of these, like having no domestic friction, [00:58:05] international friction, picking everything up [00:58:11] as a residual and that's combined [00:58:13] with the single covariance metrics [00:58:15] because think this thought experiment. [00:58:17] Ask yourself, if China opens tomorrow, [00:58:20] the model is going to say, [00:58:21] money is going to pour into Chinese stocks, right? [00:58:24] Not deep seek, that is not a listed company, [00:58:26] but in Alibaba and all that. [00:58:27] That would be your answer, I don't know. [00:58:29] I would think, you know, why the model says that [00:58:32] because the home buyers, the model reads as, [00:58:33] as Chinese return is a capital control on residents, [00:58:37] but I would think Chinese save is four into the immediate [00:58:40] if China opens tomorrow. [00:58:41] And that would be my first guess. [00:58:42] This model makes me think this can happen [00:58:45] and it is very important for us to think that way [00:58:47] because this is a very, very important [00:58:49] first order thing right now. [00:58:50] Okay, but this is probably most of us think in this room, [00:58:54] but this is what is going to happen. [00:58:56] All right, okay, so model constant with data, [00:58:59] this equity capital equivalence [00:59:01] in the model, 100 log point rise in Korean market cap, [00:59:04] it's 100 log point rise in the Korean physical capital [00:59:06] and output firm by firm. [00:59:08] In the data, of course, market cap is going to move [00:59:11] with valuation investment is going to come slowly. [00:59:13] That's okay in this model. [00:59:14] It's a static model, collapsing valuation investment, [00:59:18] but they can benchmark this to the data, right? [00:59:20] I mean, there's a lot of work on equity market [00:59:22] liberalizations going back to Peter Blair Henry's work. [00:59:25] That work tells me that stock price is going to revalue [00:59:28] 38% on announcement, cost of capital falls [00:59:31] exactly as in the model, investment rises with a lag, [00:59:34] growth increase one percentage point a year. [00:59:37] They can actually check that they have this [00:59:38] like elasticity so they can look at the high, [00:59:41] like elasticity firms, are they raising equity, [00:59:44] investing, expand sales more as we know from this, [00:59:47] it's a big literature right now. [00:59:49] I think the original papers Peter Blair Henry's, [00:59:51] but it is big, there are many numbers there. [00:59:53] And also this is combined with the direction, right? [00:59:56] Cross-border capital and portfolio equity [00:59:57] on listed firms is the same here. [00:59:59] Bonds are not doing any financing. [01:00:01] It's completely okay, but of course in the data, [01:00:03] especially for EM's, that's going to be [01:00:05] mostly dead and FDI, and listed firms [01:00:08] are going to be small flies, especially again in EM's [01:00:10] in terms of employment and value. [01:00:13] So I think it's important to realize results [01:00:15] are going to be this equitable relation for listed firms. [01:00:18] All right, and my final point before conclusion is this, [01:00:21] trade and finance substitute are complementing, right? [01:00:23] So here you gain from liberalizing one market [01:00:27] that is larger than the other one is distorted. [01:00:30] That is a very important result. [01:00:31] That is exactly why Lucas said [01:00:32] you have to care about both frictions, [01:00:34] you have to model them together, [01:00:36] but then they are working through the same channel, right? [01:00:39] So there's this substitution story, right? [01:00:42] It is through this cheaper goods, consumer surplus [01:00:45] and imported varieties to get cheaper goods. [01:00:47] So very similar in spirit to the classical [01:00:50] Mandel result of course. [01:00:51] Now there are a lot of stories that why they are [01:00:53] complements, you know, Antras Caballera work. [01:00:56] It can be complements through financial frictions, [01:00:58] complements via risk sharing, going back [01:01:00] to originally Opsvalt's piece, [01:01:02] because of course with financial integration [01:01:04] there is going to be industrial [01:01:06] supply chain more trade. [01:01:07] So all these other stories that I can give you [01:01:09] the complemented here, so this is not in the model. [01:01:12] But I'm thinking to know this, like, you know, [01:01:14] this, the revenue data coming from fact set [01:01:18] that might distort this little bit. [01:01:20] Why? Because that is going to be scaled by GDP. [01:01:23] So when you see Asia getting a big slide [01:01:25] because if that data is under reported [01:01:28] that will look like India and China [01:01:30] getting a lot of this. [01:01:31] And that is going to distort the home bias, [01:01:33] then that's going to distort the welfare again, right? [01:01:35] In fact, if you look at number, there's 53% [01:01:37] Indian spending on the listed from products [01:01:39] that goes to foreigners. [01:01:41] That is a very high number, based on what we know [01:01:43] from the multinational networks. [01:01:45] Okay, so the concentration result, [01:01:49] this liberalization concentrating in large firms. [01:01:51] Again, this is because there is no friction [01:01:53] within countries. [01:01:55] Lucas has this very nice paper [01:01:56] that shows misalocation is about domestic friction. [01:01:59] I love Lucas's paper, that's a very nice paper. [01:02:01] All right, okay, but then of course [01:02:02] we have Liliana's work, [01:02:03] that's exact opposite result on the Hungary. [01:02:05] We have Spain, this is Hungary. [01:02:07] So this makes it very interesting, right? [01:02:08] These are two European countries. [01:02:10] So I would love to see if you put a MRPK wage in the model [01:02:14] if you can talk about this misalocation [01:02:16] increasing or decreasing. [01:02:18] Okay, conclude, I am finishing, yes. [01:02:20] An important ambitious paper, [01:02:22] firm level trade meets firm level intention finance. [01:02:24] Super, super important, big picture. [01:02:27] Paper finds the world is close to financial authority [01:02:29] at the border. [01:02:30] My reading is the inside border frictions matter a lot, [01:02:34] but I think the framework is regionals [01:02:35] to separate border friction from the firm. [01:02:37] Thank you. [01:02:38] Well, we have four minutes left. [01:02:46] Okay, I'm on time. ## Q&A (01:02:47 – 01:10:00) [01:02:47] We have four minutes for questions. [01:03:31] Before the shock happens, [01:03:32] like do you tell me before you liberalize [01:03:34] or do you just do it? [01:03:35] And so I'm actually just wondering why aren't you, [01:03:39] like in some sense, my sense is I'm gonna least believe [01:03:42] welfare numbers about big reforms like that. [01:03:45] And it seems to me that the huge advantage [01:03:47] of your framework is that you can answer [01:03:49] important questions about, you know, [01:03:51] let's say I have shit firms, okay? [01:03:53] So they're very uncompetitive [01:03:56] and I have a closed capital account [01:03:57] and you know trade and I have high tariffs. [01:04:01] Now, is there a situation where, you know, [01:04:04] maybe the Washington consensus is not a great policy for me? [01:04:07] So in some sense, you know, maybe you wanna look at like India [01:04:10] was it a good idea to like liberalize trade [01:04:13] before liberalizing the capital account and so forth, [01:04:16] looking at more granular stuff basically. [01:04:25] Side is you're bringing a new framework [01:04:28] like with the characteristic demand. [01:04:31] Then the question is like, [01:04:33] what new does that deliver? [01:04:35] Like when we relative to standard kind of trade models, [01:04:38] is there something like new insight? [01:04:41] Is it the simplification that you guys can do that stuff? [01:04:45] I think that would be really like important [01:04:47] for the trade audience to like clarify that. [01:04:49] I think the variable elasticity is something [01:04:51] that hasn't been thought about much more. [01:04:53] So like building that up [01:04:54] and I can see how it interacts with this stuff. [01:04:56] So. [01:04:58] I would ask a question about the missing part [01:05:01] of some of the modeling choices. [01:05:03] Like we know many of the firms are vertically integrated. [01:05:06] So I think like the vertical integration [01:05:10] how interacts with the tariff costs and other things, [01:05:13] we shy away, but how would it distort the results? [01:05:17] And the second question is about the Revello data. [01:05:19] Revello data is biased towards high-skill workers [01:05:25] and in some countries they are not there. [01:05:27] And like it will also change the global footprint [01:05:31] of the countries and since you're doing the on TF [01:05:34] on the production, it might have some effects. [01:05:37] Like, could you comment on this? [01:05:49] Oh, so. [01:05:50] Sorry. [01:05:51] Sorry. [01:05:52] Sorry. [01:05:54] Sorry. [01:05:55] Sorry. [01:05:55] Sorry. [01:05:57] So. [01:05:58] Sorry. [01:05:59] Sorry. [01:06:00] Sorry. [01:06:01] Sorry. [01:06:01] Sorry. [01:06:02] Sorry. [01:06:03] Sorry. [01:06:04] Sorry. [01:06:05] Sorry. [01:06:06] Sorry. [01:06:06] Sorry. [01:06:07] Sorry. [01:06:12] Sorry. [01:06:13] Sorry. [01:06:14] Sorry. [01:06:15] Sorry. [01:06:16] But that's why we have the non-traded sector in there. [01:06:18] We want to make sure the allocation of labor [01:06:20] is not off in our model. [01:06:22] And this is the data. [01:06:24] We're trying to get the relative size of the sectors [01:06:26] correctly so that we don't mistake the kind of the high-skill [01:06:30] versus total labor. [01:06:32] And Mike, you're right. [01:06:32] You got it right. [01:06:33] That's kind of the relative to standard trade models. [01:06:37] Kind of the number is the same, but the mechanism is different. [01:06:41] And so I don't know if my authors want to add some. [01:06:47] That's right. [01:06:48] So. [01:06:49] I'm thinking about the question right. [01:06:52] So. [01:06:52] So. [01:06:53] So. [01:06:59] That's the same thing. [01:07:00] But I'm going to. [01:07:02] But here, actually, because we have a model with variable markups, [01:07:06] it turns out that actually financing, international financing [01:07:08] frictions have a bite of markups and analysis. [01:07:11] So let me give you an example. [01:07:12] Suppose that the certain country liberalizes [01:07:16] so that it starts attacking capital inflows. [01:07:19] What that happens, what that does, [01:07:20] it lowers the required rate of return to capital in that country. [01:07:23] So in that country, firms endogenously charge lower markups. [01:07:28] And that's why we see an increase in concentration [01:07:30] while the country is liberalized. [01:07:33] So that's one, I think, an interesting aspect [01:07:35] that is missed by current trade models. [01:07:38] Basically, it turns out the financing friction [01:07:39] matters for both the internal competitiveness of a country [01:07:43] but also for the external competitiveness. [01:07:45] Because when a country liberalizes the capital account, [01:07:49] what happens is that a country can also charge lower markups [01:07:52] externally, becomes more competitive internationally. [01:07:54] So I think that's something that is absolutely unique to our model. [01:07:57] And that's also, as in part of the next question, [01:08:01] can we think about what happens to sheet firms [01:08:05] when we liberalize? [01:08:06] That's kind of a question. [01:08:07] Yeah, let's see. [01:08:08] Like, yeah. [01:08:10] Yeah, great. [01:08:11] Thank you. [01:08:12] It's a very new company. [01:08:25] I'm a big artist. [01:08:26] I'm a big artist. [01:08:27] I've been there for five years. [01:08:29] I'm 45 years old. [01:08:31] I'm a big artist. [01:08:33] My life success is taking the bar in for a year. [01:08:35] So it's really glad to hear. [01:08:38] So it's a great experience. [01:08:41] I'm glad to hear that. [01:08:43] Yeah. [01:08:44] I'm glad to hear that. [01:08:46] You want to know what it is? [01:08:48] Yeah. [01:08:49] It's a brand new company. [01:08:51] Yeah. [01:08:52] It's a brand new company. [01:08:54] It's a brand new company. [01:08:56] It's a brand new company. [01:08:58] It's a brand new company. [01:09:00] I wanted to download my slice. [01:09:02] I. [01:09:04] They're on by. [01:09:06] Yeah. [01:09:08] Yeah. [01:09:10] Yeah. [01:09:12] Yeah. [01:09:14] Yeah. [01:09:16] Yeah. [01:09:18] Yeah. [01:09:20] Sorry. [01:09:22] It's impressive. [01:09:23] You have a whole thing memorized. [01:09:25] I just did it. [01:09:27] I put it downloads. [01:09:29] Yeah. [01:09:30] I'll fix it. [01:09:32] Yeah. [01:09:34] I'm not sure how. [01:09:36] You know, [01:09:37] I just need to share this. [01:09:46] I'm just going to see if I can discuss it. [01:09:49] But. [01:09:50] Oh, is he already on here? [01:09:52] Oh, okay. [01:09:54] Oh, okay. Yeah. Perfect. [01:09:56] All right. [01:09:57] Then yeah, you're. [01:09:58] Yeah. [01:09:59] Okay.