Transcript of the talk, from a Whisper transcription of the audio. Auto-generated: speaker names in particular are unreliable. ======================================================================== # Foreign Exchange Risk Management Spillovers Across the Production Network Authors: Wentong Chen Discussant: Adrien Verdelhan Video: https://www.youtube.com/watch?v=wEt4iKiNNZQ&t=25307s ## Talk (07:01:47 – 07:47:53) [07:01:48] So we're going to start with that. [07:01:50] standard. I want to thank all the authors and all the sponsors for participating and making the program. [07:02:05] Thank you very much for raising your hand. [07:02:09] All right, so we're going to start with the last page, very important exchange management [07:02:13] be the way the production network and something [07:02:17] very good. [07:02:21] Thank you for staying until we leave our message out today. [07:02:23] And if that's great, please, I'll be happy to be here and thank you so much for listening to my [07:02:27] feedback. Now, this is our eight-time long following exchange management [07:02:32] via Lola's platform and production at work. [07:02:34] Yeah, it's a good process. [07:02:42] Well, that's my goal. [07:02:44] You might have a good idea maybe. [07:02:46] Maybe I'll just use it as a help. [07:02:48] I think it's stock. [07:02:54] I think it's stock. [07:02:57] Okay. [07:03:08] Great. [07:03:11] So, [07:03:15] international trade 93% of the U.S. [07:03:21] for the U.S. dollars. [07:03:25] Therefore, the public literature [07:03:28] is making U.S. firms have a [07:03:31] very limited percentage to engage in foreign exchange hygiene. [07:03:34] However, in reality, the U.S. [07:03:37] firms have a large amount of the foreign exchange hygiene. [07:03:41] So here I used the hand-clicking data [07:03:44] from the first annual reports. [07:03:47] We can see that from 2000 to 2022, [07:03:50] the percentage of the U.S. [07:03:52] firms engaged in hygiene, which is the red line, [07:03:55] has increased from less than 12% to over 18%. [07:04:00] And of those hyders, the hygiene [07:04:03] notional amount, the dollar amount, has increased [07:04:06] from less than 20% of the sales [07:04:09] to over 60% of the sales as shown in the blue bar. [07:04:14] So why do those U.S. firms hedge? [07:04:17] So think about that. [07:04:19] When the U.S. firm import goods from foreign countries, [07:04:22] if the goods is priced in non-U.S. dollars, [07:04:25] the U.S. firms must convert the price, [07:04:28] a price back to the U.S. dollars to pay, [07:04:31] which corresponds to the traditional [07:04:33] price-risk discussion in international finance literature. [07:04:36] At the same time, if a U.S. [07:04:39] firm exports to the foreign countries, [07:04:42] even if the export is priced in U.S. dollars, [07:04:46] from the purchaser's perspective, [07:04:48] they have to convert the dollar price back [07:04:51] to their own currency to pay. [07:04:53] So when the U.S. dollar exchange rate changes, [07:04:56] their demand will fluctuate. [07:04:58] Therefore, the U.S. firms also suffer a large amount [07:05:01] of the foreign exchange demand risk. [07:05:04] U.S. firms are aware of that. [07:05:07] If we look at Intel, the tech company, [07:05:10] it states on its annual report that [07:05:13] an increase in the value of the dollar [07:05:15] can increase the real cost of our customers [07:05:18] in those markets outside the U.S. where we sell in dollars, [07:05:22] which corresponds to the demand risk. [07:05:25] At the same time, a weakened dollar [07:05:28] can increase the cost of the non-U.S. dollar capital expenditures, [07:05:32] which corresponds to the price risk. [07:05:36] Besides those direct price and demand risk, [07:05:40] U.S. firms can also suffer indirect foreign exchange risk [07:05:44] from their suppliers or customers. [07:05:47] Even if a firm does not trade with foreign countries, [07:05:50] if its connected firms are trading, [07:05:52] it suffers those foreign exchange risk indirectly. [07:05:56] So the standard view of the cooperative risk management [07:06:00] when we discuss that is the hygiene can help to stabilize [07:06:04] the firm's own financial stability. [07:06:07] However, if we take the production network into account, [07:06:11] a firm's hygiene can stabilize its financial positions [07:06:15] and therefore its operating activities. [07:06:18] Then the question is, can a firm's hygiene help to stabilize [07:06:22] its suppliers or the customers operating stability? [07:06:27] Therefore, my research question is, [07:06:29] when a firm hedges, does it also stabilize its suppliers [07:06:33] or customers and what is the mechanism [07:06:36] and also does it matter in aggregate? [07:06:40] So in this paper, I first developed [07:06:43] a foreign exchange risk management hygiene spillover effect [07:06:47] across the production network. [07:06:50] There are two elements. [07:06:52] First, firms have incentives to hedge [07:06:55] because the firm's borrowing rate [07:06:57] when they use external financing to finance their production. [07:07:01] Their borrowing rate depends on the firm's net worth conditions [07:07:05] and hygiene will change the net worth. [07:07:07] Therefore, the firms have incentives to hedge [07:07:11] to reallocate resources across the states, [07:07:14] especially to those high value states. [07:07:18] The second element is about the borrowing cost channel [07:07:21] and spillover. [07:07:23] Think about a firm that hedges. [07:07:25] It will change the borrowing cost [07:07:27] and therefore it's pricing and demand decisions [07:07:31] and it will change the suppliers and the customers [07:07:35] operating outcomes. [07:07:37] I use our new data, which I hand collect those data [07:07:42] from the firm's annual reports from 2000 to 2022 [07:07:47] and link those data to the supply chain. [07:07:50] I offer the evidence that when firms, [07:07:54] connected firms, their suppliers or customers start to hedge, [07:07:59] the sales of the firm's volatility [07:08:02] and also the cost of the goose volatility [07:08:05] will decrease by 3.1 percentage points. [07:08:09] So it shows hygiene has the stabilization effect. [07:08:13] I further use the quantification to apply the model [07:08:17] to a 60 industry U.S. input output production network. [07:08:21] And I show that the other firms [07:08:24] connected from industry hygiene [07:08:27] can explain half of the industry's [07:08:31] output stabilization effect. [07:08:33] Well, the firm's own hygiene can explain [07:08:37] almost all of the industry's net worth stabilization effect. [07:08:41] In other words, foreign exchange risk management [07:08:44] is a private stabilizer for the firm's financial positions. [07:08:49] Well, a network stabilizer of the operating outcomes [07:08:53] or the real activities. [07:08:56] I further show some connected results [07:08:58] like the hygiene of the more centralized [07:09:01] industry's micro more. [07:09:03] And also if we consider how important [07:09:07] this borrowing cost of China is, [07:09:09] it implies that the total value added by around 10% [07:09:14] and hygiene will offset one quarter to [07:09:18] two-fifths of the aggregate amplification effect [07:09:22] introduced by borrowing cost of China. [07:09:29] Okay, so brief on the literature review. [07:09:32] The past corporate finance literature, [07:09:34] typically treated hygiene as a [07:09:37] has an impact on the firm's own performance [07:09:41] and also try to understand what drives [07:09:43] the firm's hygiene decisions. [07:09:45] Well, I show that a firm's connected [07:09:48] firms hygiene can help to stabilize [07:09:51] the firm's own operating outcomes. [07:09:54] And also connecting them to the production network [07:09:57] literature and also the international macro [07:10:00] literature, I show that its only choice [07:10:03] of the hygiene can help or to understand [07:10:07] the foreign exchange risk management [07:10:10] spilled over across the production network better. [07:10:13] So first let's look at the model. [07:10:19] So let's first look at a very simplified version [07:10:22] of the model in which we focus on a stylized [07:10:26] production network, which is a vertical production network [07:10:29] where there is a direct importer [07:10:34] who imports from the foreign firms [07:10:36] and so puts the primary importers [07:10:39] and further from the stream. [07:10:41] So the case when my US dollar is [07:10:45] changing rate of changes, for those who [07:10:48] can import the foreign importer's purchase [07:10:50] from the foreign countries, if they are [07:10:52] comprised of non-US dollars, there is [07:10:55] a foreign exchange price list which [07:10:58] cuts through from the direct importer [07:11:01] to primary importer and further downfalls [07:11:04] and firms will lose when the US dollar [07:11:06] depreciates because of their purchasing power decline. [07:11:09] At the same time, there is a direct [07:11:12] exporter who exports from the foreign countries [07:11:15] and who purchase this goods from primary exporters [07:11:19] and also further upstream. [07:11:22] So when the US dollar depreciates, [07:11:25] appreciates the demand from the foreign [07:11:28] importers who decline, so there is a [07:11:31] foreign exchange in memory of those [07:11:33] who probably derive exporters from the [07:11:36] primary exporters and the via upstream. [07:11:38] So the firms will lose when the US dollar [07:11:41] depreciates. [07:11:43] And in the following analysis, I further [07:11:46] apply those class risks and that is [07:11:49] not the risk of channel, so I understand [07:11:52] the impact of each path of the risk. [07:11:55] The reason I can do this is in theory [07:11:58] the class risk and the demand [07:12:00] risk are in different directions. [07:12:02] Firms may do some alteration or hatching [07:12:05] but to the reality, they have a lacking process [07:12:08] and obviously that due to the contract [07:12:11] on the short-term date differences [07:12:13] and also the currency tax differences, [07:12:15] firms will lack enough of favorable rate [07:12:18] by engaging in two contracts, [07:12:21] two different contracts, [07:12:23] on the price risk effect and the demand risk [07:12:25] effect. As a result, we can add [07:12:27] the quality of those two analysis. [07:12:29] And this is for simplification analysis [07:12:32] in the modification part that I further [07:12:35] explain it to the firm production. [07:12:38] Okay. [07:12:40] So in the model, in each [07:12:43] sector I, there is a unit mass [07:12:46] of the monopolistically competitive firms [07:12:49] using the Coptacola's production function. [07:12:52] There is an X, which is the intermediate input [07:12:55] that firm purchases from the suppliers [07:12:58] and also labor with the wage [07:13:01] and the production process. [07:13:03] So there is a Z as aggregator [07:13:06] with the elasticity of war, aggregator goods [07:13:09] and sell it to the next sector. [07:13:11] So the production network exists at the sector level. [07:13:14] So there is an exchange rate [07:13:17] ES, which is defined as how many [07:13:20] US dollars needed to purchase one unit [07:13:23] of foreign currency. So when the ES increases, [07:13:26] it means the US dollar [07:13:29] is the exchange rate. So for the import [07:13:32] import side, I also consider the dollar [07:13:35] in voicing conditions, only a percentage [07:13:38] eta and you voiced in non-US dollars [07:13:41] and the firms have to convert the dollar [07:13:44] bank, non-dollar price back [07:13:47] by timing that price, the X times [07:13:50] the exchange rate ES. At the same time, [07:13:53] for the foreign demand, the demand is [07:13:56] the exchange rate as well in which when the US dollar [07:13:59] appreciates the demand will decline [07:14:02] because the real cost of the foreign [07:14:05] purchasers will increase. Yes, please. [07:14:08] Can you explain more about the exchange rate [07:14:12] of possible literature? [07:14:15] I don't quite get the question. [07:14:20] So you mean by the... [07:14:23] The demand will be less, [07:14:26] the exchange rate will be less, the exchange rate will be less. [07:14:29] So I'm just wondering if you can [07:14:32] explain more about the exchange rate. [07:14:35] First of all, you can think about this case. [07:14:38] When the US dollar purchase is [07:14:41] from the purchases percentage, the real cost [07:14:44] will increase, so there is a demand and a low decline. [07:14:47] So that's what I mean by the demand will [07:14:50] decrease. [07:14:53] So for this part, for foreign [07:14:56] use, the exchange rate will be less, the exchange rate will be less. [07:14:59] So following the positive measure to calculate [07:15:02] is the percentage of the stock exchange rate. [07:15:05] So if it has a non-dollar factor, [07:15:08] so the exchange rate will be less. [07:15:11] Yeah, I'm just saying, so there's the price [07:15:14] of the exchange rate, but the exchange rate [07:15:17] will be less. [07:15:22] So I'm just trying to match that. [07:15:25] So these are the results, the only matter [07:15:28] on the digital model. So I'm just trying to [07:15:31] characterize how important it is. [07:15:34] Yes, yes, we need both results and the money. [07:15:37] Sorry, another question. [07:15:40] So for the important input, so if everything is [07:15:43] in one single dollar, there's no risk of the [07:15:46] purchase, right? But if you say that it's 90% [07:15:49] of the exchange rate, then it's not a risk. [07:15:55] Okay, so there is a price rigidity here. [07:15:58] We are fraction long, the eye of the firms cannot [07:16:01] adjust their price. So they have to use a preset [07:16:04] price realized before the exchange rate [07:16:07] state is achieved. [07:16:10] And also their firms can make hygiene decisions [07:16:13] by purchasing a forward contract from [07:16:16] a foreign exchange intermediaries, which has [07:16:19] the exchange rate, which equals the spot [07:16:22] exchange rate E S minus the contract exchange rate [07:16:25] E F times the purchasing quantity Q. [07:16:28] And this E F is [07:16:31] finally priced, which equals [07:16:34] the sum of the probability of each state [07:16:37] pi times E S out of each state [07:16:40] over the exchange rate S. [07:16:43] And there is a complex hygiene cost gamma. [07:16:46] So this gamma is quadratic in the hygiene quantity [07:16:49] with several predetermined [07:16:52] styling factors. Think about it this way. [07:16:55] When a firm engage in a contract, they can [07:16:58] hedge other positions of the exchange rate changes. [07:17:01] So they can either long or short the contract. [07:17:04] No matter they short long or short, they have to suffer [07:17:07] on cost. And if they have [07:17:10] higher hygiene quantity, the cost will be higher [07:17:13] and it is harder for the market makers to find [07:17:16] a counterpart to engage in the contract. [07:17:19] So it is a three period model [07:17:22] and T equals zero before the uncertainty is realized. [07:17:25] The firms can choose the forward quantity [07:17:28] and also the pre-stated price and T equals one. [07:17:31] The exchange rate state and the firm's [07:17:34] pricing type are realized. And then the forward [07:17:37] contract cycles and also the firms [07:17:40] can purchase the [07:17:42] to finance their working capital [07:17:45] from a lender which costs the rate. [07:17:48] And the firms can produce and resets price [07:17:51] if there are flexible price firms. And T equals two, [07:17:54] the firms just collect the revenue, [07:17:57] retain the loans, borrow it with interest [07:18:00] and also pay out the terminal net cost. [07:18:03] So the key friction model is the [07:18:06] financing and the borrowing cost of cost stream. [07:18:09] The production cost of the firms have to be [07:18:12] externally financed. So think about it [07:18:15] as a working capital constraint as in the corporate [07:18:18] finance literature. If you are a construction [07:18:21] firm, you will collect the revenue from the [07:18:24] customers after the building has been built. [07:18:27] And before that, you have to borrow from the [07:18:30] financial institutions to finance your production. [07:18:33] And here the total external financing B is [07:18:36] the cost of the intermediate input plus the labor cost. [07:18:40] And the lenders, when you observe the firms [07:18:43] hygiene contract, the hygiene settlement, [07:18:46] and they will evaluate the projected net worth, [07:18:50] which equals the initial value of the way [07:18:53] which equals the initial wealth [07:18:56] times the hygiene cash flow minus the hygiene cost [07:18:59] plus the revenue [07:19:01] price times quantity minus the borrowing. [07:19:04] And they will offer a borrowing rate, which I call [07:19:07] a credit study, which is decreasing [07:19:10] on the convex in the projected net worth. [07:19:13] So think about that, once a firm [07:19:16] has a better net worth conditions, [07:19:19] so the financial institutions think the [07:19:22] firm is in good shape and is less likely to [07:19:25] default. So they will charge a relatively [07:19:28] lower rate. And [07:19:31] the flow of the [07:19:35] private schedule will be more steep if the firms financial conditions [07:19:39] are more tight. [07:19:42] And to summarize the firms problem, [07:19:45] firms are risk neutral, they want to hedge [07:19:48] because of your financial predictions in the model. [07:19:51] And at the key equals zero, firms choosing the hygiene [07:19:54] quantity, the price at the price, [07:19:57] and also they try to maximize [07:20:00] the expected payout by taking [07:20:03] the sum of the probability of the net worth [07:20:06] in each of the exchange rate states [07:20:09] and also the pricing tax states. [07:20:12] And at the key equals one, then when the real [07:20:15] ester states and also the quoted rate they face [07:20:18] from the financial institutions, they will choose [07:20:21] the intermediate input they purchase [07:20:24] and also they will reset their price if they are [07:20:27] flexible price. At the key equals two, the firms [07:20:30] will pay out the terminal net worth, [07:20:33] so which just equals the same [07:20:36] results after hygiene plus the revenue minus [07:20:40] the borrowing cost including the interest rate [07:20:43] component. So the first component [07:20:47] is what is the optimal hygiene of the [07:20:50] phony. So before we go there, let's [07:20:53] consider the first marginal value [07:20:56] of one extra dollar which I define as [07:20:59] M. Think about that, for one extra [07:21:02] dollar the firm has the value equals one, [07:21:05] the dollar itself, plus a bank term [07:21:08] which is a product of three terms [07:21:11] considering the hygiene impact and the conditions. [07:21:14] It is our function of the finance [07:21:17] bill B, in the firm, but to finance [07:21:20] more the revenue and the value of the [07:21:23] money, if the firm want to finance more [07:21:26] the dollar extra dollar values more. [07:21:29] And at the slope of the credit [07:21:32] schedule side here, so if the [07:21:35] firms have hygiene, a net worth conditions [07:21:38] can help to reduce the borrowing [07:21:41] rate more, it has a higher value. [07:21:44] And finally, the net worth for extra [07:21:47] dollar, how much each extra dollar [07:21:50] can contribute to the firm's net worth conditions. [07:21:53] In the firms which cannot [07:21:56] choose the price of free side [07:21:59] this term just equals one, [07:22:02] the third term equals one. [07:22:05] So think about that, for one extra [07:22:08] dollar it will value more [07:22:11] for the firm, if its extra [07:22:14] financing bill is larger and if the credit [07:22:17] schedule is steeper, in other words, the net worth [07:22:20] matters more for the borrowing rate. [07:22:23] And the firms will choose [07:22:26] the hygiene quantity until the [07:22:29] value where the marginal hygiene cost [07:22:33] equals the marginal hygiene benefit. [07:22:36] The marginal hygiene cost is a function, [07:22:39] linear function of the hygiene quantity, [07:22:42] and the marginal hygiene benefit is a [07:22:45] covariance term between the marginal [07:22:48] value of an extra dollar and the [07:22:51] exchange rate. [07:22:54] So if we look at the sign of the hygiene, there are some [07:22:57] interesting results. The sign of the hygiene [07:23:00] queue actually depends on the [07:23:03] covariance between the marginal value of an extra dollar [07:23:06] and the exchange rate. [07:23:09] Just to give another question, [07:23:12] we are talking about the cost of hygiene [07:23:15] is today, the benefit is more, is there like a time [07:23:18] or? [07:23:21] So this is trying to understand the problem [07:23:24] and keeping it stable. [07:23:27] And this is going to be a problem from [07:23:30] key people zero before the exchange rate is [07:23:33] generalized. So the firms are evaluating [07:23:36] how much to hedge. Because [07:23:39] they are estimating the expected [07:23:42] time to come across different states. [07:23:45] But the cost is going to be, they're assuming [07:23:47] tomorrow the cost is going to be. [07:23:50] So if we look at here, [07:23:53] let's look at the [07:23:57] budget price first. [07:24:00] Because it's similar, it has this term [07:24:03] in the zero because there is no price [07:24:06] feedback loop. So if we look [07:24:09] for hygiene workers, we can think about that. [07:24:12] When you have a seller appreciate [07:24:15] it, depreciate it on another risk, [07:24:18] that's higher. So the firms [07:24:21] into the cost is higher. [07:24:24] So firms have a higher [07:24:27] external comparison bills in those [07:24:30] high exchange rate states. [07:24:33] At the same time, the firms have [07:24:36] worse network conditions. [07:24:39] So this time, the [07:24:42] low, the bulk of the credit [07:24:45] value changes with respect to the network [07:24:48] conditions is higher in that state [07:24:51] as well. I don't result, the firms [07:24:54] have a high and reduced, like higher [07:24:57] standard term or the high exchange rate states [07:25:00] but the firms will long the position. [07:25:03] However, for those exporters [07:25:06] things are different. When the [07:25:09] you have a seller appreciate [07:25:12] it, they appreciate it. So firms [07:25:15] total demand will increase. [07:25:18] So this on the [07:25:21] those demand, the firms will have to [07:25:24] have a higher external minus. So VH [07:25:27] is higher than VH. [07:25:30] However, those states are [07:25:33] they have a better network of conditions. [07:25:36] So this time, the credit is value, the [07:25:39] conditions are worse in those high [07:25:42] states. But the [07:25:45] more of that in those states. [07:25:48] So as a result, the firms will [07:25:51] higher the product [07:25:54] term of the borrow and unborrow the [07:25:57] times that impact on the credit [07:26:00] value across the high and low [07:26:03] exchange rate states. And they will [07:26:06] short only if the credit value is [07:26:09] deep enough. In other words [07:26:12] hygiene have a higher impact on [07:26:15] borrowing rate. So [07:26:18] the second element is the borrowing cost [07:26:24] cost of channel. In this analysis, think about that there are four exchange rates. So if [07:26:32] 8-on-4 hydrates, it becomes part of the first network and which enters into the first borrowing [07:26:40] rate because borrowing rate is a function of the first network. And this borrowing rate [07:26:47] is part of the first production cost. They will enter into the marginal cost and therefore [07:26:53] enter into the first pricing decisions. The pricing is marked off tax, the borrowing cost [07:27:00] tax, the other part of the marginal cost. And those prices will impact the first of purchasing [07:27:09] decisions and price decisions. As a result, it will impact the customers and the suppliers of [07:27:16] operating outcomes. Let's look at the overall impact of the hygiene. Before we go there, let's [07:27:29] define our business day term. How does the price and sector eye changes with respect to the exchange [07:27:38] rate? So it is a term of the flexible price expected to share, and of course, can occur [07:27:44] across your real-time or higher possible. And by inputting cost of response in the cell. [07:27:52] Plus, you know, since it is a term which captures how the operating rate, the credit [07:27:58] schedule changes with respect to the exchange rate changes. And this term is a function of hygiene. [07:28:06] In other words, the hygiene will impact the first pricing decisions by affecting [07:28:12] the borrowing rate. So if we look at those first risks of the whole rule, we focus on [07:28:19] a very simplified four sector version and we check how the hygiene, how the exchange rate [07:28:26] impact the sector to surprise and what is the corresponding impact of the hygiene decisions. [07:28:33] We can see that the sector to surprise changes. Now, physically with respect to the [07:28:38] exchange rate changes, these are just the typical price risk and transmission. If a firm [07:28:46] hyges, this transmission will be mutated. At the same time, if the sector one, the suppliers, [07:28:54] they have a hygiene condition, it will also stabilize or mute this transmission. [07:29:02] So that is the price risk. Similar cases also hold for the demand risk. If we check how [07:29:12] made for sector two, the demand changes with respect to the exchange rate changes, [07:29:17] both its own hygiene and demand side, the customers hygiene matters. [07:29:25] So before we go to the data, let's think about the case where I am made a stream from. [07:29:35] If my suppliers hatch, its borrowing rate enterprise will be more stable and my input cost [07:29:42] will be stable. If my customers hatch, they will have similar mechanisms, borrowing rate and [07:29:48] price will be stable. So my sales will be more stable. Therefore, to test this hygiene spill over [07:29:54] effect, a very natural way to hedge is when my suppliers start to hedge, how does the cost of [07:30:02] goods from the volatility of that variable changes? And when my customers start to hedge, [07:30:07] how does the volatility of my sales change? Okay, so let's go to the data. As I said, [07:30:15] for the data, the direct would be if I start to hedge my borrowing cost go down, [07:30:21] would you have evidence of that? Yes, yes. Yes, yes. I will show that before the [07:30:26] amendment, before the quantitative. Okay, so to collect the firms for exchange hygiene [07:30:33] positions from the firm's annual reports, I first use a tax screen, screen of all the 10k [07:30:41] files of all the US firms. And if they have a valid foreign exchange derivative [07:30:47] mentions following the word list from the Hoverganum for 2017, I take those firms and [07:30:54] use the hand collecting data of the firm's foreign exchange hygiene national amount, [07:31:00] including the contracts of forward futures options as well as the cross currency swaths. [07:31:07] And if the contract value is non-US dollars, I convert it back to the US dollars and take the [07:31:14] sound within the firm here across different hygiene accounting designations. And for the US firms, [07:31:21] I consider those non-financial firms and also exclude the utility industries in my [07:31:28] main analysis in the data collection I collect all of them. And the data covers from 2000 to [07:31:35] 2022. And I define the firm's foreign exchange hygiene and intensive margin as the foreign exchange [07:31:42] hygiene national amount, scoured by the firm's total asset. And there is an extensive margin, [07:31:49] which equals one if the firm's foreign exchange hygiene national amount is greater than one [07:31:54] and is zero otherwise. A quick summary is of all the years, there are around 15% of the firm [07:32:03] here, hygiene and among those factors, the national is around the 7.7% of the total assets. [07:32:12] The time series trend is the rough I show at the beginning. [07:32:17] I then linked the firm's hygiene data to the production network conditions, where I combined [07:32:23] two data sets, compute that segment, which includes the major customers conditions, plus [07:32:29] the fact set and reviewer, which has a broader coverage of the supply chain links. And those [07:32:35] combined network covers the period from 2003 to 2022. I further construct the two extensive [07:32:43] margins of the upstream hygiene and the downstream hygiene, where the upstream hygiene equals to [07:32:50] one if at least one of my suppliers hatch. And the downstream hygiene equals to one if at least [07:32:57] one of my customers hatch. As we see in the mainstream view slide, to check how the firm's [07:33:05] operating stability changes following my suppliers and customers hygiene, I construct the downstream [07:33:12] firms hygiene impact to check the log depth of the firm's sales itself. And then for the upstream, [07:33:19] I checked the log change on the cost of goods sold. And I also constructed the firm's [07:33:26] specific foreign exchange exposure by using the contextual analysis of which country the firm [07:33:34] mentions they import goods from and export goods to. And I use that mention, the percentage of [07:33:42] the mention of all the top country related measures, times the foreign exchange volatility [07:33:48] of that particular currency. The advantage of this method is it can capture two directions, [07:33:55] both import and export set. So to check how the first suppliers or customers that connected [07:34:05] the firms hygiene can help stabilize the firm's performance, I use the first [07:34:11] adoption of the firm's suppliers or customers hygiene and check how the first bonus change [07:34:19] following this aggressive adoption of the connected firms hygiene. To avoid the impact of the hygiene, [07:34:27] I control the measure group, which is the one to one nearest neighborhood hygiene, take a firm [07:34:34] with the same year same industry, but it's not worth as no connected firms hygiene suppliers [07:34:42] or customers hygiene. And the firm is mesh related on the outcome variables history and [07:34:48] slow, the outcome variable changes as well as the firm characteristics including side leverage, [07:34:56] cash, technology, and also the firm's own hygiene conditions. And I use the U.S. VALY method, [07:35:03] which captures how the firm's performance changes controlling for the higher level [07:35:09] outcome variable differences, the higher fixed effect and also the 10th fixed effect. [07:35:17] And the identity of the adoption is that it without those connected firms hygiene, [07:35:23] the treatment group and the control group, they were the firm's delivery similar. [07:35:28] And in after this mention, I have unlocked 1,300 of the firms that in the suppliers who [07:35:38] act with the customer group. And for those firms, they are connected with the start hygiene [07:35:46] here and their industries have a great variety of thoughts on the same periods and industries. [07:35:53] So for that identifying something that we need some things that they are deciding to start [07:36:01] the business with things that they do have to find to implement. [07:36:06] So, yeah. [07:36:18] So two big answers to that question. So first, when the firm started in [07:36:24] Ohio, I can observe whether it started in Ohio or not. And to avoid the cases, the firm [07:36:32] doesn't know if this belongs to whether it's hygiene or not. I just exclude those firms. [07:36:40] The second answer is for the leverage conditions, they are writing control [07:36:46] setting fact and this partially the firms financing conditions may impact their hygiene [07:36:52] conditions because the treatment and the control to really have our same language [07:36:58] conditions and not damaged by those in conditions. [07:37:07] Yes. So, yes. [07:37:10] So there are other circle reasons for choosing this kind of. [07:37:15] Well, the decision is not true. So based off the suppliers and their customers, [07:37:21] right? So I agree there can be some factors that may contribute to the connected firms [07:37:27] space stages, but I try to control that. Okay. [07:37:37] So here is the result. So we can see that after the firm connected firms started to launch, [07:37:47] there is a 3.1 percentage point, big one in the first past and the cost of losing [07:37:54] thousands of dollars over the years from zero to three years. This is our long [07:38:01] plan to 14% of the free adaptation work. [07:38:06] And I further check whether there are some additional component factors here, [07:38:13] like does the firm phone hygiene contribute to that? [07:38:16] Whether the firms space different for these change conditions, risk of conditions, [07:38:22] whether it's network changes, new firms were existing suppliers, [07:38:28] or just how persistent this already is and also whether there are any [07:38:34] impacts of the matching ways and also does not change for the magnetic robot. [07:38:42] Okay. Now before we go to the model, let's go to the mechanism. So first, the firms own hygiene [07:38:50] can help to indeed stabilize the firm's performance, specifically of those firm [07:38:56] hygiene for 10% increase of their hygiene national than their sales. The absolute annual [07:39:02] growth rate of their sales cost of goods. So the cash and the net worth declines from [07:39:09] 1.2 percentage point to around 3 percentage points and this effect also pose within firms. [07:39:19] And the second mechanism I want to test is the borrowing cost channel. [07:39:25] When our firms net worth no matter increase, no matter what measure it is using the book [07:39:31] or market value of those net worth, the borrowing cost decline. [07:39:37] And also we can check how the firm's hygiene conditions change with our firm connects to [07:39:43] more suppliers of customers. Of those two graphs, the horizontal axis shows that the number of the [07:39:49] customers and suppliers and the vertical axis shows that the hygiene participation rate is [07:39:55] very variable. So I thought from customer suppliers of customers, they are a number of the [07:40:02] hygiene participation percentage increases and is also a whole to get the control room [07:40:10] of firm characteristics. And so the next question is how much does the hygiene elsewhere in the [07:40:24] production network has an impact on my own performance and does it matter in aggregate? [07:40:32] So [07:40:33] Okay. So in the calibration part, I calibrated the model to a 16 industry US production network [07:40:46] using the 2022 BEA for the output accounts, including the gross output conditions in [07:40:53] intermediaries and the trade conditions. And I consider two economies focusing on the [07:40:59] price risk and the demand risk separately. And I calibrated the foreign demand elasticity [07:41:07] by the post-literature, the GT AP based variations across industries. And there are firm [07:41:16] equally likely foreign exchange rate states. And also there is our free-set press share [07:41:23] London and also the credit schedule parameters are calibrated to match our around 4.86 reference [07:41:31] rate from literature. And also there is a hygiene cost parameter. So which is [07:41:38] which is chosen to match our lower hygiene cost of bar from the literature. [07:41:46] And there is a target in the calibration, which is the average hygiene notional amount [07:41:52] as a percentage of the gross output, which is 3.56%. And this yields a gamma of around 7.65. [07:42:03] And to validate the model calibration, I use one untargeted moments after grouping those [07:42:11] industries into 16 different big groups. We can see that the relative rank of those [07:42:18] industries hygiene conditions are pretty good. The model predicted part is consistent with the reality. [07:42:28] So the first result I want to show is what does the hygiene stabilizes. In the import price [07:42:39] economy and the demand risk economy, the firms hygiene can help to stabilize 50% of its own [07:42:48] terminal networks and also 30% if we look at the demand economy. However, if we look at the operating [07:42:58] outcome variables, no matter sales cost of goods or the outputs, the hygiene can only stabilize the [07:43:05] volatility, varying from 1% to 6%. Therefore, the hygiene stabilizes more financing outcomes [07:43:15] than the operation outcomes. Second is how much of our industry stabilization in fact comes from its own [07:43:28] hygiene and how much comes from everyone else hygiene. Think about our two players game, the firms [07:43:37] the sectors own hygiene and also the other rest 59 sectors hygiene decisions. So if we want to check [07:43:48] the conditions, what is the gains of the exchange or the volatility stabilization from no hygiene to [07:43:55] hygiene? We should consider the case where the firm hygiene first on the rest firms hygiene first. [07:44:05] So I take the average of the two cases namely a shipping value of those conditions [07:44:11] and I found that the firms own hygiene can stabilize the firm's own performance which is [07:44:20] like it stabilizes 79% of the firm's total net worth. That's good and consistent with the corporate [07:44:29] finance meal hygiene can help to stabilize its balance sheet. However, if we look at operating [07:44:36] outcome the things are different. The other firms hygiene if we look at the output can [07:44:42] explain 15% of the volatility decline of the firms output. So in other words, half of the [07:44:53] stabilization in fact comes from other firms hygiene. [07:44:56] But the output is much smaller. [07:45:00] The average is much smaller. [07:45:02] It's been half of a very small. [07:45:04] Yes, half of a small. [07:45:07] That's true. Okay. And the third question is whose hygiene matters more? [07:45:15] So here on the horizontal emphasis and plot the forward leon tiff which leon tiffy words which [07:45:22] measures the centrality of each industry and the two plots shows the price risk economy and [07:45:31] the demand risk economy. Vertical emphasis shows that the sum of the value added volatility changes [07:45:39] following the counterfactual analysis. In every time I just remove that particular [07:45:47] industry's hygiene opportunities and let all the rest of the industry optimize. [07:45:53] And we can see that I's a firm is more important, more centralized in production network. [07:46:01] This hygiene has a higher impact on the aggregate [07:46:06] volatility changes. The final result I want to show is how much does the hygiene can [07:46:15] explain the volatility introduced by the borrowing cost channel. I construct our [07:46:20] counterfactual economy where each industry borrows and unmatched the constant rate. [07:46:27] So in that economy, so if the firm does not hedge it boosts the aggregate value added volatility to [07:46:35] around 10% and hygiene can offset 42% in the price economy and 26% in the demand risk economy. [07:46:47] Okay, to summarize and introduce the theory of hygiene's bill wallers across the production [07:46:53] network by the borrowing cost channel and offers the empirical evidence and show the [07:46:59] quantification that the firm's own hygiene can offer or stabilization of the financing [07:47:06] conditions where the other firms hygiene can offer the stabilization of the operating outcome [07:47:13] variable changes. And so hygiene is a private stabilizer of the finances and the network [07:47:19] stabilizer of the real activity. This matters because even if we have a dollar in voice in [07:47:27] the international trade, US firms still has foreign exchange risk. They are trying to hide that. [07:47:33] And if we consider the production network, the hygiene disruption has our impact [07:47:40] spill over effect. And it's also open to the future research on what is the [07:47:45] optimal allocation of the hygiene and the relationship of hygiene across [07:47:49] emerging markets and at once the economies. Thank you. Thank you. ## Discussion (07:47:53 – 08:04:51) [07:48:28] I'll let you do it. [07:48:30] Much safe. Much safer. [07:48:35] While you do this inside the lab, I want to thank you again. [07:48:40] It's a lot of work to organize this meeting and to go over all the papers and you've done a [07:48:43] great job. Thank you very much. So I'm going to discuss one time's paper. [07:48:55] It's not to you. [07:48:58] Yeah, it works. Okay. So you are right to stay until the end. This is a good paper and it's a [07:49:06] good paper on a literature that absolutely needs it. If you take a step back, the big ideas [07:49:13] we basically did that from the eight days like push off, you know, time and hours and [07:49:17] that's been that you didn't work for a very simple reason. First, don't have to report the details [07:49:23] of the packaging policies. And so you open up your stack, you open the work scope, [07:49:27] you don't say it. And so if you don't have the data, it's very difficult to do integral work. [07:49:31] And I think this paper makes progress on both dimensions. I went down as a model and [07:49:36] on top of that she has an amazing integral work. So I'm going to try to describe both [07:49:41] thinking to the science. The vertical idea is very simple. Think about the working capital. [07:49:48] In order to produce something, you need to borrow some money. So today we're going to produce the [07:49:54] desk and produce the desk. We need some wood, but we need to borrow the money to buy the wood. [07:49:59] Now we're going to buy this wood in some foreign currency. And once the dollar is going [07:50:04] to be appreciate, this would cost more. That's bad news. And that's going to have some [07:50:09] consequences because the fact that my network goes down, that means that I'm going to need to borrow [07:50:14] at a higher rate. If I borrow at a higher rate, it may be going to show up in my cost and the price [07:50:22] I'm going to charge. Now, why do I go to the page? It's because my boring rate changes [07:50:29] with my network. If you don't have that assumption, hedging is nothing on the table. [07:50:35] So Wynton writes this in a little model, the network prediction thing. And so she has [07:50:43] speedovers. I'm going to do this in a more simple setting. She also has a really nice [07:50:49] and a good call result. She showed you this figure, that's bigger for in the paper. [07:50:54] Let's basically showing that when your customer hedge or your supplier hedges, it has some impact [07:51:01] on you. So if your customer hedges, your sales are now less volatile. If your supplier's hedge, [07:51:07] your cost costs are now less volatile. It's very natural. But in the magnitude, she finds out [07:51:13] awfully big. You remember the 2.1 point that she mentioned. And the mechanism is that just [07:51:20] hedging allows my supplier or my customer to borrow at a lower rate and to offer [07:51:26] all these people their conditions. And then she puts that in a big quantitative model with [07:51:31] many, many sales. So I like the paper. I really enjoyed putting this paper and you'll see I did [07:51:38] a few things. I'm going to try to organize my comments in two parts. One, I'm going to go for a [07:51:44] simplified version of the model. I'm just going to go for one front. Price is going to be rigid [07:51:49] so that I can try to understand what is going on. And then I'll go for the beginning of a [07:51:55] replication attempt on the video. So let's start with the model. I'm using the XI C invitation [07:52:02] as I'm doing. I am a student at the period. But there's only one firm now. And the prices and [07:52:09] the outputs are fixed. See? So moving. So we won't have to worry about that. Now, for video to [07:52:17] your point, in the way I'm going to write it, yes, I'm going to pay the cost of hedging in [07:52:22] period one. And so that's the quality cost. And I'm attached by buying a forward, forward [07:52:29] gas drive. And why? Well, you have to pay a period one. I'm going to borrow some money after the [07:52:37] anything that you just realized. And I'm going to borrow some money to cover the cost of importing [07:52:42] the input bill. And in the last period, just the revision of the individual. So [07:52:52] the key assumption in the model is that there is a critical schedule. And if I look for the [07:52:58] vendors, let me add a higher rate. So you need to know who you look at the time you ask for the money, [07:53:06] which is a date and date one. And so what's going to be the difference between the [07:53:11] value of the exchange rate where the forward gas drive did you have the number of forward [07:53:16] gas drives in both minus the quality cost of hedging. And that minus the amount of the borrow [07:53:23] is the measure of the network. Now, in this model, the screen, the result is absolutely [07:53:31] ambiguous. And the object is the functional value of $1. So you can write the value function [07:53:36] as exactly what I'm going to show up. Let's see, I need to my side. I'm not going to show it. [07:53:40] But think about the functional value of $1. But if you want more, that's pretty cool. [07:53:44] It's going to give you an extra benefit. Because if you look rich, you're going to be able to borrow [07:53:50] a kind of lower rate. Now, this is kind of different in finance. This side, which is kind of the [07:53:56] outmatch, the interest rate you draw changes with your wealth. And the big is the amount of [07:54:02] money you borrow. So now, I think, the actual condition of hedging is you have to [07:54:06] want benefit for the actual amount of money. That's the most amount of cost of hedging. [07:54:15] And what's the functional benefit? Clearly, what you want is transfer money from states of the [07:54:21] world when you don't have that money. So think about, again, my piece of wood, the total depreciates [07:54:28] since I paid this in foreign currency, it's cost me more. That's very, very painful. [07:54:34] I want to make sure that this doesn't happen. And so in times when the total depreciates, [07:54:41] my input bill goes up. And that's what I want to avoid. So let's think about the [07:54:49] compliance between the two. The total depreciation means this is going up. My input bill going up [07:54:54] is the most expensive one that was up. This is when the two things are going up at the same [07:54:58] time, the compliance, positive, so I'm just trying to attach. Q is always going to be positive. [07:55:03] So if you write a model where the prices are rigid, it becomes very nationally that you [07:55:09] want to catch. Now, she doesn't want to do that. Why? Because we should do that. [07:55:15] Since the prices are set in advance, there are no proposition. There's no point in writing a whole [07:55:23] network key when there is no proposition. So that cannot be the model she wants to write. [07:55:29] So the model that went on the price is a model where prices are flexible. And that the mechanism [07:55:35] is going to be a bit more complicated. So it's a same timing issue. And then what you have to [07:55:40] realize is that you borrow money after the infection, and you don't borrow money in day one, [07:55:49] because it's not the one thing we would do to actually come and break it off. So let's [07:55:54] rework this problem now with flexible prices. If prices are flexible, the problem at day one is that [07:56:01] the price over its emotional cost, because that is the market, and the emotional cost will depend on [07:56:06] the interest rate of the labor to transition. So now we redo the reasoning. We just think [07:56:14] again about the emotional value of $1. If I have one more dollar, it's pretty cool if you have [07:56:18] one more dollar. It's also very different. The emotional value of the law depends on how much [07:56:26] the more the value of charge is needed when I look for it. But there's an additional term, [07:56:31] and this additional term is the one that we need to think about. So let's say I can imagine that [07:56:37] the dollar we appreciate, I need to improve this word because it means more. I've never [07:56:41] heard the whole story before. But now, since it costs me more, I'm going to change my price. [07:56:48] My price is going to be slightly higher. I'm trying to go like this, so we're going to buy a little bit [07:56:54] less from me. But if I buy a little bit less from me, I need less money. It's about $10 for the [07:57:00] weekends. I still have two forces, the same one that we did with the rigid price that we've [07:57:04] done before, and this kind of new force that comes from the fact that if I adjust my price, [07:57:08] I'm going to adjust it up and the demand is going to be lower. So once the opportunity side [07:57:12] with you actually do want to hedge down now, and when you actually derive this in this form, [07:57:21] the condition that has to be true so that we do want to hedge demand when the price is not [07:57:26] flexible. So if I want to present the result, this is $10 more than the amount of hedging. [07:57:30] With rigid prices, you always decide to hedge because you really decide the time that [07:57:35] you actually get to pay higher rates when the dollar wall depreciates. With flexible prices, [07:57:40] it's no longer obvious. There's a vision of the model where the fees were exactly what you would [07:57:44] expect. But the vision of the model, when it comes down to the exact opposite, you have to [07:57:48] start hedging, it's like you take 100, some current fees. So I'm using that as a suggestion. [07:57:54] Now, since what we see is the key assumption of the model is this credit [07:57:58] schedule. When your network goes down, you have to pay higher rates. I wanted to see the [07:58:04] specification which was reasonable on that. And I think it does. So I just put numbers [07:58:09] to the math formula that actually went on to show you. And it would be, but interest rate, [07:58:15] the firm that you pay would go from something like 3% to 11%. Okay, why not? But I think it's [07:58:22] actually seems to be what the model implies, seems to be relatively in line with what you have in your [07:58:28] business. So I wanted to show it, I said that in the agreement, actually, the model, [07:58:34] the firm doesn't really visit what states where there's a high interest rates. And so that maybe [07:58:43] I'm just specifying. I think I buy actually the whole mechanism, which at least what I find [07:58:50] just more difficult to solve is initial assumption. Why would the interest rates [07:58:58] change with my network in the model where I never default? If I default, who cares? [07:59:06] So in the model where I look, people, like, what, what, right, that, then the whole thing [07:59:12] would actually work. Now I'm going to the data. So that's the figure that when you can start with. [07:59:20] Just to remember on this sheet, both the interest and the extensive margin start with the red line. [07:59:28] That's the extensive margin. But you see that the number of firms actually has [07:59:33] come from 10% to 20% in the US. And the blue is the intensity margin. And what you see is that [07:59:39] all the firms that they cannot have something like this first and the rest of us. So I try to [07:59:44] repeat that. But I don't have the patience of going to this manually, as you can imagine. [07:59:50] So I went through the 10 things that I can go. And I also use the fact that since 2009, [08:00:01] I can find using the next one. I can find the good notes of the financial statements [08:00:17] that firms have to report. And so I use that as a way to to replicate what my job is doing. [08:00:24] I can match, I think, the national amount of project that's actually in five, like all that I [08:00:29] said, I can also match the number for how to get the safe one. So that's the data set on the [08:00:36] company. Now, when you can come here, what you get from the simple text analysis from the [08:00:44] experiment data, and it's not a one to one, the text, the text screen is going to find [08:00:50] 80% of the firm that we bought is at the extent the national, but it doesn't find all of them. [08:00:56] So that's the percentage of what actually would be missed by just the text. But it's also [08:01:00] not perfect in your way around. There are some national that actually can [08:01:04] find 100% of the data that match. That's it. Here's what I found. So this is my repetition. [08:01:12] This is when some say that this is my repetition, as you can see, only it's really easy to [08:01:16] accept because I'm not doing things manually. So I get the same path. I don't get exactly [08:01:24] the same magnitude. Now, once I have the data, I can also read through the first [08:01:28] time, look at what is the first year when you hatched. And when you get out, I'm going to look [08:01:33] at what's in fact, on the top of the list. So I didn't read the whole thing, but I looked at [08:01:39] the number of first time hatchers and what you can see that version did spawn in 2008, [08:01:44] 2001. And so here, you have to ask yourself, is it really the first time they hatched [08:01:49] or the first time you really read both? And that I did not decently. Now, I have first [08:01:56] for me, maybe an issue because think about it, 2008, 2008, 2008, as a crisis. So we all stopped [08:02:02] hatching in very turbulent times, which is what you would expect. In turbulent times, you want to [08:02:08] say that you hatched. So it's not a right time to actually tell your investor that you hatched. [08:02:13] But just after that, the word continued to recurse. So it's not. So it's going to [08:02:18] because you've hatched the first time you went up actually, just started reporting. But I went [08:02:24] on to talk about that. And so she has actually a business like I'm controlling for [08:02:30] realize what I think it doesn't actually change. Now, when I look at the agency, [08:02:35] launching their money that doesn't really make what what she wants. I get much more [08:02:41] ratios of it, which much more overall sales. And if I look at the top pieces of features, [08:02:47] I don't get I don't get what she what she has. So I must be missing something. And I might think [08:02:53] she should actually be at this just show the benefits of the manual analysis, because the [08:02:58] XPR data that can be done right to people. Now, once you get into this, I think you should even [08:03:04] push your push your your work because by the way is manual analysis, you actually can extract [08:03:10] even more information. When I get the amount of that is not long, I don't know if you are [08:03:15] fetching the debt or fetching the cost. And my mind is that a picture is actually [08:03:22] in the debt. So you borrow the point you're in, see you buy the swap by corresponding [08:03:27] then get these are going to exist. And so that has no to do with the with the mall. [08:03:31] So but you would want to be fine. Okay, I have only one minute. So I just go back to [08:03:36] probably to her, she was asking for a direct test of the mechanism. And I think one [08:03:42] you don't have a perfect test. Like what we're going to test is exactly what probably you said, [08:03:46] you want to get exactly and the war and cost. What you are right now is that [08:03:53] lowers the volatility of the sales in the network and then you have another test that goes [08:03:58] from the network in level to the lower cost. What probably shows us is go from the ethics [08:04:04] edge in level to the foreign cost in level. Now, you can have something more. If you believe [08:04:14] my life and it's something very similar, you can actually test this condition. [08:04:19] It basically varies why the person should touch. If it's not, you should actually take on [08:04:23] 20 series. Okay, that's a very nice. Okay, I want more things on the back of the envelope. [08:04:31] Let's see this and that's going to be it's a great person very well written. It's very publish. [08:04:39] It's really you can always don't wait. I think that there's a contribution both on the practical [08:04:46] side in the very simple mechanism and making it more transparent with just one one firm. ## Q&A (08:04:51 – 08:08:59) [08:04:53] And actually, you are super interesting data work and you can actually play that. [08:05:03] You're finished. [08:07:16] You know, in the growth of our economy, [08:07:20] we think that that people are able to take it all, right? [08:07:24] Because we're going to be able to do this, [08:07:26] we're going to be able to do that. [08:07:28] But if you say, [08:07:31] if you want to economy, [08:07:33] it's a good process. [08:07:35] So that high productivity [08:07:38] in the economy, [08:07:40] that you've got over the hundred, [08:07:42] you know, [08:08:03] it's not.