A Framework For Geoeconomics

>> Speaker 1: But welcome to the start of the new year. It's really exciting to get going, but we're very pleased to have Matteo Maggiori speak to us today. The title is hard to pronounce. A Framework for Geoeconomics, is that correct? Okay, which is joined with Christopher Clayton and Jesse Schreger is on.

>> Matteo Maggiori: Yes, Jesse, I think, is on Zoom.

>> Speaker 1: Fine, anyway, so take it away. We're anxious to hear this novel way to think about the world.

>> Matteo Maggiori: Fantastic. Well, thank you for coming and happy New Year. This actually is the first seminary I gave when I joined Stanford in January 2020.

And George Shultz was still alive and it was pretty impressive to meet him. But it's good to be back. And the paper today, as John mentioned, is with Christopher Clayton, who's at Yale, and Jesse Schrager, who's at Columbia. He's online, so if you have any tough questions, he'll handle them.

So first, what's the topic? So at some level, it's a very old topic. It's the idea that governments use their countries economic strength that comes from existing trade and financial relationships to achieve some geopolitical goals or economic goals towards other countries. Sometimes it's easier to think of the topic as a set of questions.

So first, is this a thing? Like, does it even exist? If this power does exist, in which dimensions does it operate? What are its origins? How does it get wielded? There is a lot of informal talking about this topic, but once you try to dig deep on where does it come from and what exactly do you mean, the answers aren't so obvious.

Similarly, whenever you talk about, particularly geopolitics, there is a tendency to think of it as a zero sum game. I get something out of another country and I'm better off, they're worse off. But is he potentially opposite some game, for example, having a country like the US exert global power, are there positive aspects from that?

Another question that is related is what sectors are strategic? Now, strategic is one of the most loaded words because government tend to abuse them in all sorts of fashions to start justify protection in subsidy, nationalism. Khrushchev used to say, you can make buttons strategic because soldiers need buttons, because otherwise they need to hold their pants and they can't fire.

So you can make almost anything strategic without a theory. And so the point today is to provide a theory and some set of definitions that you might agree or disagree with, but that hopefully push us to confront these questions from a more formal perspective. The ingredients I'm gonna use are pretty standard coming from economic theory.

So a lot of the contribution of the paper is trying to weave them together into a framework to answer this question. So I'm gonna have a collection of countries, I'm gonna have global production networks. So think of it as input output matrixes, where something that I produce gets used as an input somewhere else that creates linkages across industries and countries, and I'm gonna have limited enforceability.

So why do I have it? I want to position the paper and geoeconomics in general between two extremes. At one end is the very blunt threat of gonna war. You don't do what I want, I'll invade you. At the other end of the spectrum is complete contracts where we can agree on fully enforceable contracts on all possible contingencies.

This power operates somewhere in the middle, where the threats are not war, they're commercial threats. But very often what I try to achieve, it's either not enforceable or I don't want to write it down. For example, I want to have a political concession. I'm China, I don't want another country to recognize the Dalai Lama.

I'm gonna pressure them commercially, but I don't want to write this down in a contract. So the main mechanism that we're going to have is that the power arises from the ability to do joint threats, from the ability to use a number of activities that might look rather unrelated.

Exporting and manufacturing and financing or government aid, and join them together to exert pressure on other entities. But also because I'm a large country, I'm gonna ask you, once I have power over you, I'm gonna ask you to take some costly actions. But the reason why I'm gonna ask you for those actions is I'm trying to manipulate the world equilibrium in my favor.

So costly actions are typical things like tariffs, sanctions, export bonds. I'm asking Nvidia not to export ships to China. It's a privately costly action for Nvidia. I'm doing it because I would like China not to develop a particular technology. So for a matter of reason, okay? So that's sort of the broad framework.

Let me try to position it in the literature because it's kind of interesting. This is my own very partial reading of what happened through the history of thought. But we're building very heavily on two books from Hirschman in 45 and 59, who was thinking about national power and trade.

So this was a very big topic immediately after the war. And then my own partial reading is that when formal mathematics came in with Chicago and MIT, pushing the frontier it was very difficult to deal with these concept. And some people reacted badly and went into the wilderness of all the mathematical enterprises useless because they can't quite give me what I want yet.

And the topic kind of went out of fashion. And I'm gonna argue that today what we're gonna try to do is bring it back into fashion, particularly because over time we've developed substantially more sophisticated theoretical methods. And we're gonna borrow heavily from the theory literature to in some sense show that there are tools that we have developed over 50 years that actually make these concepts very easy to analyze, or at least possible to analyze.

So what I have in mind, well, first I'm gonna use threats. So the way to exert power here is gonna be threats. Now, 50 years ago, threats were a difficult concept. Today we have meaningful concept like off the equilibrium threats to induce good behavior. I'm gonna think about the idea of using multiple industries or multiple activities, as being very connected with the work on why having multiple points of contact can provide high power incentives.

Okay, some of that work is done here at Stanford. Similarly, when I think about costly actions because I'm trying to distort the equilibrium, I'm gonna build very heavily on the macro literature that has developed plenty of tools in public finance and in macro. Where we want to apply some wedges in first order conditions to affect prices of aggregate quantities.

Those are the tools I'm gonna use. Now let's get started. And the way I'm gonna run the talk is first, please do feel free to jump in. This is not a paper where I can do 45 minutes of talking and then you ask me question at the end because trust me, this will be unclear because I'm not good enough at presenting it.

So please stop me. But really point one, two and three there's a lot of models set up. Okay, you need to bear with me for a while because I need to define the environments. Then the paper really comes alive in point four when we start thinking about there's a big country and he can make this threats and he can manipulate things and let's see what it does, what is his optimal contract and then trying to go through applications.

But as I go through the setup, please stop me if anything is unclear. I heard Peter Navarro talk about economic warfare all the time.

>> Speaker 1: Yeah.

>> Speaker 3: This what you're trying to capture?

>> Matteo Maggiori: Yes, coercion warfare. I'm not sure I'm gonna talk about it in the same way, but.

 

>> Speaker 3: I hear this sort of stuff all the time as well. One motivation is we need to win the actual affair. Don't give them parts that they can use to fly planes with. And the other part, what I hear all the time, is we need to win the strategic economic competition, which God knows what that means.

So are you after your economy as a source of military power, or just somehow we need to-

>> Matteo Maggiori: There's gonna be no military actions. It's going to be entirely on the economic side or political connections. And I think I share your sentiment, for a long time I was interested in this topic.

I hear it all the time. I couldn't make heads for tails if that was me. So this is my attempt to move past words that you and I will quickly disagree on what they mean into a question that you and I will still disagree on, but at least we know what they mean.

So I have an infinite horizon model. I have n countries, and there is productive sectors in each of the countries, okay? So here, a productive sector, think of it as Russian oil and US oil are two different sectors, okay? So sector here isn't an industry, it's a sector in a country, okay?

And I have some local factors. Think of it as labor or local endowments. I have a unit mass in each sector of producers. What they're doing is they're buying inputs from other sectors. They're using some local factors that they're producing some output. And that output might end up being consumed or might end up being an input into somebody's health production, okay?

To make it simple, in each country I'm gonna have a representative consumer. So here the focus is really gonna be on the sectors of production. And then the first thing I'm gonna do, which goes to trying to put content on what John was mentioning, is I'm gonna have a vector of aggregates that I'm gonna use to track externalities.

So what am I really doing? If you're familiar with Rothschild Stieglitz, I'm just borrowing that one entirely, saying there's a bunch of aggregates that might affect me. You can think of it as climate, you can think of it as national security, and I take them as given as an agent.

But our collective actions affect these aggregates, okay? So it's a very simple, reduced form way to capture all sorts of externalities that we're gonna use. And you'll see exactly what it does. It's a repeated game, so I have a discount factor beta. Good, so this is the easy part, the consumer, he has pretty standard preferences over a consumption of all the goods that exist in the world.

And you can see the first appearance of the Rothschild Stiglitz term. I have a utility component that-

>> Speaker 3: Rothschild Stiglitz paper you're talking about?

>> Matteo Maggiori: 86.

>> Speaker 3: No, no, not the adverse selection.

>> Matteo Maggiori: No, no, no, this is pure externalities. How to track externalities. So you can see there, I'm putting a vector of aggregates into the utility function.

So simply screwed interpretation is, as a consumer, I'm afraid if I see the military sector in China being big, okay? It's been most used in economics to say there are some aggregates that I don't like. That's all it does. Okay, I'm not gonna use it a lot, but it's there.

What is the budget constraint? It's pretty straightforward. Per period, I have money that I spend on consumption of the goods on the left hand side, you can see it here. I have the profits from my domestic sectors. Here I'm gonna assume that you own your domestic sectors. There is nothing difficult in this paper if I assume that you arbitrarily own some foreign industry, but I'm not gonna solve for indigenous ownership.

So I'm gonna assume that you own the domestic firms, and of course, you get the revenues from the factors, okay, please.

>> Speaker 4: Matteo, you've already assumed away one thing which I would have thought was important in this context, and that's the distinction between a regime and the populace.

So sometimes geo-economic threats are exercised with the intent of undermining the regime rather than harming or helping the populace. That seems to be off the table.

>> Matteo Maggiori: So, no, let me tell you. So there's one part that is definitely off the table because it requires a real model of debt, and there's one part that is not.

So in a lot of the literature that you have in mind, for example, these aggregates here can be used to be political concessions. I don't want you to take an action, and I don't like your government being blue, I like it being red. If you switch to red, I'm happy.

Now it's cheap, so I wouldn't call it a model of that. But when you don't want to provide a full model, making that one the real thing, this is a cheap way, for example, in trade, to track it. So I have it to that extent. I certainly am not gonna have good models or lobbying or elections at all.

We decided intentionally not to do it. But you're totally right. Okay, out of these, all I get off to keep track is the typical Marshall in demand. Cuz I have prices, I have my wealth, I decide my optimal consumption. I have my indirect to deal with off the stage game, okay?

So the consumer here does very little. Now here's the interesting part, is my suppliers versus clients game. So this is the production side. So I'm gonna set it up as a stage game, and then I'm gonna build a repeated equilibrium with threats, okay? So I'm gonna do the stage game informally in the slides just to get a sense of how it works.

So I'm thinking of a firm, I, playing against a set of suppliers J. Okay, so these are two links in the input-output. I'm buying an intermediate input from them, they're selling to me. At the beginning, what do I do? I place an order. I say I'm firm, i, I am ordering Xij from the suppliers in j.

They can accept or reject this order, okay? If they reject, nothing happens. If they accept and I get the delivery of this order, so I get the goods or I get the capital if it's a loan. And then I have a choice, I can pay the suppliers or I can steal from them.

And I'm gonna assume symmetric strategy. So this is a game between a single firm and an entire sector. If I pay, I pay all suppliers equally in the sector. If I steal, I steal from all of them. Okay, that's essentially the first part. Now what happens if I pay?

Well, if I pay, they will trust me next period, okay? So I'm gonna define a trigger strategy that is based on whether I trust you or not. If I trust you, I think it's possible that you're not gonna steal from me. If I don't trust you, I think that with probability 1, you'll steal from me.

Those are my beliefs at the beginning of the period. So the updating is, if last period you paid me, I will trust your next period. If last period you stole from me, I will not trust you next period. And since it's self fulfilling, I'm not gonna trust you ever again.

 

>> Speaker 4: So is this stealing or not meant to be a proxy for breaking any rule between firms? But I'm stealing intellectual property. That's very different, like I have it forever. Does it work for things like that?

>> Matteo Maggiori: Yeah, I mean, we are at the end of a long school of thought on this, but these are pretty general.

They're essentially just repeated relationship where you give me something, we sign a contract. And if I deviate, you punish me in the future thinking that I'm always gonna screw you. Okay, so it's pretty general in terms of doing any repeated setting. It could be a loan, it could be you provide me inputs.

What we are excluding, to be clear, is paying upfront, right? So this doesn't work very well for small transactions that can be entirely collateralized. This has to be in a space where there's a reason that you need to trust me a little bit at the beginning, okay?

>> Speaker 3: What about larger ramifications, just having some of your one period output stolen.

What about the threat of ruin, annihilation or?

>> Matteo Maggiori: As I mentioned at the beginning, I'm gonna intentionally avoid those coz I wanna think of it as almost a different topic. Here, I wanted to sort of try to focus the model, on that intermediate area where the US is pressuring other countries commercially.

Invading is a pretty expensive proposition. Not selling you some goods. Think of Russia right now, nukes are off the table, thank God. Commercial sanctions are what we're doing. Well, I'm not gonna get into that.

>> Matteo Maggiori: I know you people have opinions, because that, I'm gonna stay to the part I can talk about.

 

>> Speaker 3: It's not governments, so we're not even talking?

>> Matteo Maggiori: For now, these are sectors. Governments are gonna be some of an entity that is gonna coordinate the threats.

>> Speaker 3: What's the private armies?

>> Matteo Maggiori: Yeah.

>> Speaker 3: That's not what we're talking about anyway.

>> Matteo Maggiori: Yeah, that's exactly it.

>> Speaker 5: But in this supplier relationship, do you allow for, say I'm gonna deliver this type of good.

And you're gonna handle in this way, so you're not gonna steal intellectual property. Or you're going to use it or you're not gonna put anything. Think about the Huawei case in Europe, right? At one point they were all buying that, but then there was a threat that Huawei would.

 

>> Matteo Maggiori: Give me a few slides and we'll get into exactly that. But for now, I need to do some more setup.

>> Speaker 1: Okay.

>> Matteo Maggiori: So how does this work out? So it's pretty straightforward. If I behave as a firm, I get my per period profits. I'm using these inputs I'm producing, and I get a continuation value that for now is exogenous.

The next period, you're still gonna trust me, okay? If I'm a supplier and this is important, I'm assuming that the firm is small, so the suppliers could always sell to somebody else, okay? So the cost of opportunity of selling to the firm is very small. I'm essentially rigging the model, to make the threats cheap, which is an intentional decision and I'll tell you later where it shows up in equilibrium.

Now, if I go down, what happens? Well, I get a bump in profits, because clearly I stole this period. I'm gonna assume genetically that they can repossess a fraction data of the inputs. That's like, think of it as some rule, there's some baseline that I can't steal. But then what is the cost?

Is that next period, you're not gonna trust me. So next period, and this is important, you'll see, I can't use you as an input. I can still re optimize among everybody else that still trusts me. So that's gonna allow, potentially, for a lot of substitutability to kick in, in equilibrium.

But you're not gonna trust me, so I'm losing access to your input, okay? Now, if I'm the supplier, clearly I make a loss coz you stole from me. So if you do backward induction, it's pretty straightforward that the suppliers only accept contracts that are not gonna be stolen on.

So that leaves you to an incentive conditions, that trades off the little bump I get from stealing. That's my short run profits, versus this gap here, which is my loss in continuation value. From the fact that, I'm not gonna be able to use you again as an input provider.

Now, I will make it more examples in a second. But you can immediately see here, that this threat is only valuable to the extent that you cannot substitute away from me easily, okay? Threatening you with something that you can easily buy somewhere else, from somebody else, just isn't valuable in equilibrium, but you can immediately see it.

Okay, yeah.

>> Speaker 3: Your reference to j is confusing me, because you say that the suppliers j, there's multiple suppliers in individual j. But now when you talk in j, but now when you're talking about the Incentive Compatibility Constraint, does that mean that's true for all j's?

>> Matteo Maggiori: Now, very good, very good.

So this is important. That's why at the beginning, I tried to be careful about the game here that I'm studying is between a firm, in a downstream sector. So firm i is a specific firm, and a sector of suppliers, of which there is a unit measure, okay?

>> Speaker 3: Yeah.

 

>> Matteo Maggiori: Now, in equilibrium, I'm gonna only have symmetric strategies and representative firms. So in some sense, the representative firm is the sector. But for the off path analysis, it's important to think about, when the deviation is a single firm in the sector, versus the whole sector together. So here, j, is a set of suppliers.

They all have the same beliefs about me, and I treat them symmetrically. So the updating is at the sector level versus an individual firm, okay?

>> Speaker 3: Well, the other thing is, what's x sub ij is that, the quantity from one firm or from the whole.

>> Matteo Maggiori: It's the quantity that the entire supply and sector j supply to firm i, if I steal, I'm stealing from all of them.

 

>> Speaker 3: Okay, okay.

>> Matteo Maggiori: So I treat them always symmetric. That's all, but that's an important subtlety in this, okay?

>> Speaker 6: And the continuation value, if you deviate and do this, this is individual, firm by firm, or I would think of there's like a coalition. I kick them out of swift and they.

 

>> Matteo Maggiori: No, this is individual firm by firm, and I'm gonna build an spe, so for now, I'm taking them as exogenous. And then I'm gonna endogenize them, by building entire spe for it. For now, I'm thinking about, I take the continuation values as exogenous as a firm. This is my action, in a second, I'm gonna need to build where those continuation values come from.

But that's exactly right. Okay, so the one shot game for the firm is it takes this continuation values as given. It's trying to maximize profits subject to incentive constraints, okay? Some notation, Bi is the set of sectors that trust me as a firm i. If I steal a subset of exoderm, this is the set next period that is left trusting me.

And this is gonna be the Incentive Constraint, potentially, has lots and lots of combinations. I could steal one good, three goods, six goods altogether. So let me give you a concrete example to make it somewhat easier. So let's assume, that I have two sectors that are supplying j and k, and I'm looking at firm i, okay?

So let's start from a combination where both sectors trust me, okay? So the set of possible stealing decisions are I steal j, I steal k, I steal both of them, okay? Those stealing decisions, then define a set of Incentive Constraint is that if I only steal from j next time, only k, trust me.

If I only steal from k next time, only j, trust me. If I steal from both next time, all I have is 0. I can't produce anymore. I lost all my inputs.

>> Speaker 5: So j cannot observe when you are stealing from k.

>> Matteo Maggiori: They can, every all actions here, are observable.

 

>> Speaker 5: Observable.

>> Matteo Maggiori: And the set B of who trust you is common knowledge, okay?

>> Speaker 6: Everybody knows everybody's type.

>> Matteo Maggiori: Everybody knows everybody's type.

>> Speaker 3: So this belief that, if you cheated on me, I don't trust you anymore, but I can see that you treat it on Doug.

But I still trust, even though.

>> Matteo Maggiori: That's gonna be exactly, what I'm gonna play around with. My sorcerer joint threats are gonna be exactly those triggers. I could in principle say, look, my beliefs update that if you stole on anybody, you're gonna steal me. Or my beliefs update, where if you steal from me, that's the only time that I'm gonna update.

I'm gonna use those to form joint triggers in a second, okay? Now, this is what Doug was asking me, the value function. So how do I build it? Here, it's literally as straight from Abeo Pearson Stacketti, okay? So if it looks like a magical construction, It is. It's a beautiful paper.

We're not adding anything at all. We're using it.

>> Speaker 3: But, their paper was with lots. Lots of things not obsolete.

>> Matteo Maggiori: Yeah, yeah, using very simple versions 30 years later.

>> Matteo Maggiori: I'm good with that. All I want to say is we didn't invent this, we're using it.

>> Speaker 3: No, no, I know.

 

>> Matteo Maggiori: That's all. Okay, so let me start, okay, just to make it simple, the construction is a typical recursive one. You start with, nobody trusts you, value is zero, I can produce anything. Now, I go up one notch, there's one sector that trusts me. I can use the previous built one as my continuation value.

If only one sector trusts me and I steal, I go to zero, I build my new value function, then I go one step up. What if two sectors trust me? Well, if I steal any one of them, the previous value functions are the continuation values from where I am, okay?

And I'm gonna keep going until I get to the set where everybody trusts me, okay? So that's how we build it.

>> Speaker 3: No, that's not, see, you're using sort of single agent decision making version of dynamic programming to talk about a game. And the thing is that the Bellman principle doesn't apply to games.

This business of iterating backwards in the value functions because there might be multiple fixed points or multiplicities. So-

>> Matteo Maggiori: I'm certainly not.

>> Speaker 3: Your story about building. Whereas what another intuitive way to build is to assume that everybody trusts you. But then you find out conditions under which incentives don't work and then you shrink down to the building.

You're talking like building interior, that there's no theory that says that that's gonna get you the full set of equilibria, everybody trusting your bike and then chopping off.

>> Matteo Maggiori: Hold on one second, let me get John and I'll get back to this.

>> Speaker 7: So mine's very simple, if you could take 30 seconds to read that equation, for those of us who have forgotten what all the.

 

>> Matteo Maggiori: I will do both, let me start from one and then go to the other. So what am I doing? I am maximizing profits per period, okay? That's my per period stage game profit. If I set up people that trust me, I need to know, because those are the people that will accept orders.

 

>> Speaker 7: Which is the set of that sigma of si.

>> Matteo Maggiori: No, no, that's bi, here. Sorry, John, you're not looking at it.

>> Speaker 7: Here, I'll look bi.

>> Matteo Maggiori: No, it's okay.

>> Matteo Maggiori: Those are the people that trust me, okay? I choose my inputs and I choose my factory usage.

That tells me my per period profit, okay? I don't steal, I have my continuation value. Next video where the set of people that trust me are the same as today, okay? When don't I steal, when if I stole, this is my bump in value that I get from stealing.

And it's to be below what I lose in the continuation value from eliminating a set of players that trust me because I stole from them, okay?

>> Speaker 3: And for all s such that s is-

>> Matteo Maggiori: Right, so that one is what I showed you here, in this example.

In principle, I could steal from one another, both of them. So I have a whole set of those incentive constraints.

>> Speaker 7: You said something about building up where more, more people trust you.

>> Matteo Maggiori: Yeah.

>> Speaker 7: That only seems like preserving the number of people who trust you.

>> Matteo Maggiori: I did it at each point in time, I need to know, sorry.

If I look at the people that trust me, I need to know what happens when I steal from a bunch of them, okay? So this goes to what we were discussing with John. So imagine starting from nobody trust me, there's no production I can do, the value is zero.

Now I go to, okay, let's start to think about an SP where one sector does trust me. If I screw them by stealing, I'm gonna go down to zero, trusting me. So I'm gonna use the value that I built from the first conjecture as what do I go down to if I steal, okay?

 

>> Speaker 7: So there isn't a dynamic process here of getting more people to trust me. Finding the equilibrium-

>> Matteo Maggiori: As an SP, it's the same construction as a bra kit. And to what you mentioned, I love to actually speak more about that. I think the way we built it, I'm pretty confident in, it's the same exact way we build the SPs in all of macro is we're starting from.

I'm trying to compute continuation values and I'm adding one at a time, but I have to think about building it the other way around.

>> Speaker 3: Thing is that in our paper, Shavin and I were not. We talked about the basic, we call the inner approximation, where we grew out and to fill in and we could find no proof that that always converged.

Now what, so what we did though is we looked at the outer and then shrink it down. And that's also what recently there was a paper, a brew, in somebody that did a refinement of pastoral roach bars. And they only looked at the shrink.

>> Matteo Maggiori: I should look at it.

 

>> Speaker 3: I think there was some Minnesota guys that had a paper many years ago and they did just the shrinkage.

>> Matteo Maggiori: Okay.

>> Speaker 3: The building, so I don't know of any paper where the building starting in and going out works. Because, you see, there are possibly multiple fixed points.

 

>> Matteo Maggiori: That I'm happy with.

>> Speaker 3: You could be starting, rebuilding, and then you get stopped at a fixed point. Remember, in this theory, it's the maximal fixed point, that is the one we're looking for. So that's why-

>> Matteo Maggiori: That's a good point.

>> Speaker 3: That's why shrinking works.

>> Matteo Maggiori: I should look at it.

 

>> Speaker 3: But that building this inner out, building out.

>> Matteo Maggiori: Let me look-

>> Speaker 3: Let me fail, because it may.

>> Matteo Maggiori: Let me look at it, that would be interesting to look at. I did the sergeant textbook construction.

>> Speaker 7: But can I make sure I understand Ken's point? If you find an equilibrium by going up, then he's found and-

 

>> Matteo Maggiori: Yeah, that's-

>> Speaker 7: You're just saying is that-

>> Speaker 3: Equilibrium definition here typically is maximum touch point of this.

>> Matteo Maggiori: No, no, that's not what we're using. That's what I was saying, like-

>> Speaker 3: Okay, but I'm saying that's-

>> Matteo Maggiori: No, no, yeah, no, you were using, okay, that's actually used.

 

>> Speaker 3: The set of all Nash equilibria is the max.

>> Speaker 7: No, no, I understand he's not, that's not what we're doing.

>> Matteo Maggiori: Here we're not doing the set of worth threatening you with the worst to sustain the best, that's not what we're doing. Here we're simply saying, in fact, here, this is definitely not the worst possible threats.

 

>> Speaker 3: No, I'm just pointing out that there are multiple fixed points.

>> Matteo Maggiori: I'm totally happy with that. Now I understand where you're going.

>> Speaker 3: With several national-

>> Matteo Maggiori: No, no, okay, now I understand, we're not going there at all, which is what-

>> Speaker 7: It seems like the solution concepts like Shapley values and successful.

 

>> Speaker 3: No, this is not how to do Shapley values.

>> Matteo Maggiori: Okay, now I got your question better.

>> Speaker 3: Only talking, Nash.

>> Matteo Maggiori: Now I got your question better. No, here, we're not doing that at all. What we're doing is, we're building it from the bottom up, finding one.

We're not making any claims that this is the maximum one. In fact, that's why we're not doing that Search. Yeah, now I understand. Okay, got it. No, we're not doing, and I'm perfectly fine with this being an SPE definitely. We're not trying to play the game of sustaining the best possible outcome.

You'll see, in fact, that the paper is written away even more because we're restricting to market strategies. Here, you're definitely gonna do better by, for example, looking for punishments that are backloaded and explaining the sense. Okay, good.

>> Speaker 7: Matteo, how many suppliers are there to start with?

>> Matteo Maggiori: For a given sector, it's exogenous.

There's a set SJ that you start with. Here, we're not doing any, it's the last step here.

>> This Ji year is exogenous. Here, we're not doing any network formation, which will be pretty interesting because you could literally buy relationships, here we're not. You start with a set of relationships and you can only go down if you stalled.

 

>> Speaker 7: It seems like you're building up to a coalitional response to an individual action.

>> Matteo Maggiori: Right.

>> Speaker 7: You can imagine there's a coalitional action response to a coalition lag.

>> Matteo Maggiori: Very good, I'm gonna be there in one slide. So market clearing, I'm gonna do it very fast. It's pretty straightforward.

There is output, the other goes into intermediate inputs or into consumption. And factors have to be used, local factors. Nothing difficult there, okay? That's what Doug was going to is. Remember, I wanted to start thinking about joint threats, which are essentially a form of coalition building or collusion.

And first I'm gonna define them and then I'm gonna start thinking the paper comes alive, where I endow one country, the hegemon, with the ability to make these threats, okay? So first, what is a joint threat? There's nothing else in some sense that acquires partition, simply saying, if I start from the previous example, I'm gonna build a joint trigger strategy where if you're still on k, I will not trust you next period.

And k will have a symmetric one. If you're still on j, I will not trust you also next period. Now, the construction is a bit more involved than that, but ultimately what does it end up with? It ends up essentially with a consolidation of ICs, right? Clearly, if I know that since these are public common knowledge, if I know that that's what I'm facing, I will never steal from j, but not from k, because I get punished from both.

And so what it really does in this particular example, it transforms the problem from having three ICs to only having the joint IC. Now, for those of you that are deep into this, you can already see how this is gonna work out in equilibrium. I'm giving you higher power incentives by telling you if you deviate on anything, I will punish you all across the board, okay?

And that's coming from a long literature before us.

>> Speaker 7: Can I just ask a question about, I guess, the modeling strategy? So you're allowing these coalitions to form, but you've taken off the table some punishment cost other than just withdrawal of my supply.

>> Matteo Maggiori: Very good.

>> Speaker 7: I don't quite see why I would realistically-

 

>> Matteo Maggiori: Why did you do it? Why did you do it this way? That's an excellent question. So I think what we try to strike is a nice middle ground between the following, particularly once you go to these games, the strategies can get very complicated and very arbitrary. So we restricted attention to one Markov to this particular set, which we found sensible.

Now, why do we find it sensible compared to something else? Well, you will see in a second that what I really wanted was China that wants to pressure Lithuania in not recognizing the Dalai Lama, but has no relationship with Lithuania. But it turns out to have a lot of relationship with Germany.

It was a major supplier to Lithuania. And so it's gonna use that indirect route to form a threat. That's why I want them. The question of why you're not allowing other things. So there are many that we don't in the interest of writing a paper, but they're easy to do.

There are a few that are actually pretty expensive to allow. So, I'm not sure if that's the one you have in mind, but the one that comes up the most in seminars is the blunt threat. Rather than doing something sophisticated that is of value to you, I could simply say, if you don't do what I say, I will not supply to you ever again, okay?

That one is an inefficient threat, why? Because, well, it's a hold up. I'm making your outside option worse off, but that worsens your incentives compared to increasing your inside option.

>> Speaker 7: A basic real world one you've taken off the table is, I hand over resources to a neutral third party that we both trust, who will adjudicate whether or not I honored the contract.

And if I'm on the losing side of that adjudication, I forfeit something.

>> Matteo Maggiori: Perfect.

>> Speaker 7: So that seems like you think that's off the table perfect.

>> Matteo Maggiori: No, no, I'm going to argue it's not, but certainly in a cheap way. Remember that we said you can't steal everything.

You can steal a fraction, and that fraction is exogenous. So I'm gonna think of some relationships where that solution is possible. Thetas are very close to 1. These are legally enforceable. This thing doesn't do anything. And I can't threaten you there, there are some, I'm building a mine in Africa.

Everybody understands that after I build it, I'll get expropriated. And there is no amount of asking somebody else for an escrow that's going to solve this expropriation problem or think of sovereign lending. So I think that's what I'm using to, and it does matter why? Because in a second, what I wanna argue is that governments always like threats.

But most threats in equilibrium make no sense for the reasons you have in mind. There are other ways around that are perfectly credible. Or I can tell you what you want, you can substitute the input somewhere else. And so I actually do want to capture that economics. I'm not gonna go for upfront contracts or collateralization.

I'm simply saying set the data to one. There's no threat sale I can make, at least individually.

>> Speaker 7: Here, the problem we're getting around is, if you cheat on Steve, I can see it, but I trust you anyway. So the government has to get together, go and enforce me to not trust you after you cheat.

 

>> Matteo Maggiori: Yes.

>> Speaker 7: That's, why don't I-

>> Matteo Maggiori: Well, look, you're JP anymore. Well, I would argue that if you look at, for example, the history of the US, there's a lot of the State Department doing that. It's telling JP Morgan you're not gonna make new loans if one of our firms got expropriated in manufacturing.

And while we're making it extreme, where it's sort of a simple assumption, I think in practice it does capture an important part of this coordination, elements that the firms themselves might not, might not do. To be clear, some of it I think actually happens at the firm level.

Here, I'm gonna make the hegemonic country, but I think of large industries, for example, are doing some of these perfectly on their own.

>> So I think the point is, well think that there's some of this that will happen even in the private sector.

>> Speaker 7: With lending to third world countries, we've seen this over and over.

They default, and then the next thing some other bank goes and lends the money. But I thought we started to answer the Peter Navarro question, and this doesn't seem to have much to do with why we want to put in tariffs on electric vehicles from China.

>> Matteo Maggiori: Let me get there.

I disagree, but until we see it, for now it's a bit too early. But give me a second and we'll get into exactly that.

>> Speaker 7: Because I mean that it's not a problem of China will welch on its if they buy something on us, we'll send them a bill and they say no.

 

>> Matteo Maggiori: Let me give you the example I have in mind of China. China has built an immense amount of infrastructure in Africa in a place where enforceability has always been a problem. Now, if Italy did, we'll get expropriated the next day. So why doesn't China get expurgated? Well, because the threats that China makes are to suspend lots of other economic activities that are very valuable to that country for any one deviation.

So what they're doing is essentially cross-collateralization. They're using some relationships to back up others, and they're being big and being able to make these threats. It's quite valuable.

>> Speaker 7: That's right. I mean, that goes back to the whole history of trade. Why did trade always involve military power rather than just going to India and buying stuff?

Cuz you go in and they'll screw you.

>> Matteo Maggiori: That's what we want to capture.

>> Speaker 7: The sort of geoeconomics that sent me and Casey ballistic was not about that at all. It was about us fighting against China for who's bigger.

>> Matteo Maggiori: I'm gonna get to that too, but give me a second, cuz we haven't started that part yet, okay?

 

>> Speaker 7: Okay.

>> Matteo Maggiori: So this is just the micro part is I'm making threats, and in a second, we're gonna let the hegemon make them. Then there is this sort of, why do we fight each other on the size of semiconductor? Without macro amplification, you're not gonna get it.

And here it's all micro, everything is taken as given so far.

>> Speaker 8: Your externalities that you introduced are not on the production side.

>> Matteo Maggiori: They're not on the production side.

>> Speaker 5: Not yet.

>> Matteo Maggiori: Give me a second, just hold on.

>> Speaker 8: I'm happy that we don't have them cuz-

 

>> Matteo Maggiori: You'll see them, but that's gonna be pretty key. Okay, so first, these I'm gonna go fast cuz to most of you, this is pretty obvious. Joint threats here can be valuable. When are they valuable? Well, they're valuable because by trading you with a worst continuation value, if you deviate, they're providing high power incentives.

And that for you as a firm, it's good, okay? So that's pretty straightforward. Now what is interesting, particularly if you're in Makarov, is that these values are computable. I'll give you a very simple example. Think of a nested cs, okay? So there's lots of types of oil around the world, oil is almost a perfect substitute in production.

So the threat of withdrawing one of them, it's totally useless. Next period, I just simply increase my inputs from some other oil provider, nothing happens. The threat of shutting all oil down might be very severe cuz, for example, the basket of oil might be entering almost in Leontief in production.

If you don't have it, oil at all, you can't sustain anything, okay? So that's a very simple example of how to compute this.

>> Speaker 3: Oil is a very good example because that's what United States and Britain tried to do.

>> Matteo Maggiori: Yes, for example-

>> Speaker 3: There was a war.

 

>> Matteo Maggiori: That's exactly right.

>> Speaker 3: Now, you haven't mentioned this, but in your paper you said the prices are fixed.

>> Matteo Maggiori: No.

>> Speaker 3: No.

>> Matteo Maggiori: In the paper we said that there's an example we fixed prices to avoid terms of trade.

>> Speaker 3: Okay, this is flexible.

>> Matteo Maggiori: Yeah.

>> Speaker 3: Okay.

 

>> Matteo Maggiori: Flexible price.

>> Speaker 3: Okay.

>> Matteo Maggiori: Okay, now let's get going, okay? Now we're in business. We did the micro, now I need to make some choices, okay? And to Steve, these are gonna be pretty arbitrary, I'm gonna tell you why we make them. But these can be changed.

So a hegemon here, we already said it. It's a country that essentially can coordinate these threats. It can coordinate its domestic sectors, and it can coordinate some foreign sectors. I'm gonna have to restrict the sectors. I'm gonna assume that you have to be one step removed. So if you're downstream from one of my industries, I can use you.

But if you're two steps removed, you're too far from me. It's totally arbitrary, I could make the threats propagate farther. That's easy, okay? To fix ideas is, my firms and the firms in D, which is the downstream sectors of my firms. What do I do? Well, first we said joint threats, okay?

That's what I'm gonna offer. What am I gonna demand? I'm gonna demand transfers, okay, from the firms that I can target. These are essentials. Simplest example is monetary transfers. But this could be markups on goods that I sell, surcharges on the loans I make to you, okay? These are not revenue neutral, they come out of your profits.

But I'm also gonna allow for taxes, in particular, for revenue neutral wedges. What are these? This is where the connection with public finance and matter comes in. If you deep into those fields, you will recognize that those are the typical wedges we put in Euler equation. We love them because they can be specialized to cover tons of stuff.

In particular, they can be specialized to cover tariffs at valorem. They can be specialized to do quantity restrictions. In this particular paper, the cleanest thing is to think about quantity restrictions, okay? So I'm asking you to reduce your input, your usage of a particular input, okay? That can implement it with a tariff or with a quantity restriction.

I'm gonna do local rejection of the contract. And because this is Markov, I'm gonna keep assuming that the hegemon can commit to future contacts, okay?

>> Speaker 5: Can?

>> Matteo Maggiori: Cannot, cannot.

>> Speaker 5: Cannot.

>> Matteo Maggiori: What I want to avoid, this is the discussion we were having before. I could try to do better by, for example, backloading the punishment and things like this.

Here the government can't commit to the contract, it can do only one period contracts. Okay, good. This one I skip. This one I'm gonna skip too. If you're confused, I will do it again. This one I'm gonna also skip, except one thing. What's special about this environment compared to domestic macro policy is that very often here, I'm contracting with entities that are foreign.

I'm gonna ask a foreign firm to do something or a foreign government to do something. I can do it by legislating on them. This is not domestic taxation, so they have to voluntarily participate in my contract. I'm offering a threat and I'm asking for actions. So I'm gonna have to have a participation constraint.

And this is what I was saying here, what we're avoiding is the blunt threat. Here what I want to do is I want to give you something of value, which is the commitment that comes from the fact that I can threaten you a lot. But then I'm gonna ask you for costly actions.

Now there is a limit to my power. If my threats are invaluable, I can't ask you to do anything cuz you're gonna simply reject the content. So the participation constraint here tells me how much power do I have over you at the beginning. In some sense, this is micro power.

This starts to get to where we're going with macro. This is saying, take as given all aggregate prices and quantities, I have some power over you, I can ask you to take actions. That power is being tracked by, how much luck do I have in your participation constraint?

If I'm supplying rare earths and without them you're screwed, I have a lot of power over you. There's a lot that I can ask out of you in equilibrium. If I'm supplying water in a Nordic country, I have very little power over you. You can substitute me away with anybody else, okay?

Now, macro kicks in when I start thinking about, what do I want to do as a hegemon? Why, because while I can manipulate, I understand that I'm not contracting with each firm only, I'm contracting with a whole set of them. And that's gonna affect equilibrium prices and quantities.

The firms in accepting the contract are only thinking about their private costs, but I'm thinking about changing the world equilibrium. Now, this is crucial, cuz if you think you're national security, that's a clear problem. A clear problem is every firm that is thinking about accepting a contact from China is thinking about, given the equilibrium, what happens if I accept?

The government is thinking, what happens if the entire eastern seaboard of the United States, all the parts were controlled from China? So it's pretty important to distinguish this two level. And I'm gonna go there. So what is the hegemon doing? This is straightforward. It's maximizing its representative consumer welfare.

Those are the two terms from an hour ago which I definitely don't remember. These are the indirect utility from the consumer. And these are potentially political preferences that we mentioned. What is the wealth of my consumer? Where is the profits of the domestic firm, but also the transfers that I get from the foreign firms, okay?

So transfers domestically are a wash. They get paid out of firms to consumers that own the firm. But transfers out of foreign entities are not a wash because I don't own those profits, okay? Good, so I mentioned that we rigged the problem to have cheap threats. And what do I mean?

I mean that, in fact, in equilibrium, I will do all the threats that I have on my disposal as an hegemon, okay? This, you can modify. You can make some threats expensive to make in equilibrium, but here, the threats are actually cheap. Because in some sense, we've assumed each of the buyers is mole and there is a market out there.

If instead you're thinking of a world where if I don't sell to you, whatever I produce goes to waste, that would be a very expensive threat to make, okay? We're not doing that. So that's the cheap part, okay.

>> Speaker 3: Each of the firms in my coalition promise to believe to the end of time because they will never get paid for any of this.

 

>> Matteo Maggiori: Yes.

>> Speaker 3: And the fact there's no learning like there's an extra, this is totally static. These guys don't really care. And they're gonna have these beliefs that are not.

>> Matteo Maggiori: And it's totally self fulfilling. So you can see that the model is very static. In fact, if you had seen this presentation of Chicago a month ago, this would have been a two period model.

It turned out that the two period model was actually a pain because the problem with the two. I mean, this is an aside, but the problem with the two period model was the following. Because you're not doing markup, the continuation values are not taken as given. And so guaranteeing that some of the optimizations were convex problems was a bit more difficult.

This one actually turned out to make it easier. But fundamentally, it's a two-period model. There's really no real dynamics here, okay? We started with a two-period model. In fact, for those of you that don't like longer models, to me, my preferred version of this model is two periods.

The second period, there is no incentive problems. Everything is essentially an IO matrix, exogenous. You're solving the one time, the ex ante period. That would be, by far, my preference that if it is, we'll hate it. So we can do infinite horizon, repeat it. But turns out to be that.

 

>> Speaker 3: In both theory and in reality, one period of punishment and then forgive and go back seems to be robustly what you do in repeated games. And it seems to be what people do in reality. You don't punish forever.

>> Matteo Maggiori: Yeah, so, okay, so there's lots of things that you can relax.

With these problems, it's very easy to go crazy in terms of difficulty. One thing that is certainly possible to do, I've done it in a previous paper. I need to be careful and think whether it works here. But I think if we did exogenous reentry, a persona rival or changes of beliefs will be just fine.

Because all it does is it scales down the continuation value by the export, the punishment by the probability that we reset. I have to think carefully about those. In general, those are possible to do in this class of models.

>> Speaker 3: I think the fancy word is subbed in perfect, but I may be wrong.

If I could punish you one period and then you agree. I'm sorry, I won't cheat on you again, that's better. It's better from my point of view to go back to testing than it is to punish you forever, just cuz I said I would.

>> Matteo Maggiori: No, to be clear, so this is where you play with these models.

Here, given that everybody else doesn't trust you and we're doing Markov, that I think in the future I'm not gonna trust you. I don't do anything better from trusting you this period. So in that sense here, the threats are credible, there's no credibility issues. At most, you're indifferent, they're fully credible.

Now that's a technical answer. That's not an answer about. But what if we did expedite those exclusions and you can play those games here. I really wanted to make it two periods. So in my mind I keep thinking of this model as tomorrow the world ends. But there is lots of good things that come from simplification, that come from the full mark of structure.

Otherwise we wouldn't have done it up to a month ago. But we didn't do it that way. But it turned out to be actually stranger and less standard. And so it was easier to recast it. Okay, now let's start looking at the optimal fun. This is the fun part when we do the application.

So first, and this goes to what you were asking me about prices, I want to start thinking about the simplest possible environment. So what I'm gonna do is shut off general equilibrium prices. The way I'm gonna do it is pretty simple. I'm gonna assume that the preferences of the consumers are quasi linear and they're gonna be marginal in all the goods, okay?

And I'm gonna assume that the production functions don't depend on any aggregates, okay? Now that's a very simple environment, why? Because there's essentially no macro amplification. So what does the hegemon want to do there? He wants to impose no wedges. He has no reason to alter your privately optimal allocations.

Okay, so no wedges. But on foreign firms, I want to extract transfers, so what do I do? I don't extract any transfers from my domestic firms, why? Well, transfers are awash in terms of profits. They come out of firm profits and go to my consumer because they own them.

But because they come out of firm profits, they're worth an incentives. They tighten the ICs. So transfers here are purely distortionary. So on the domestic firms, I don't do them. On the foreign firms, I do want to do them, why? Because I don't care about their profits. I don't own them, but I do get the transfers.

They're not a wash for me as a hegemon. So formally, what is happening, which we will do when we think about efficiency? Here, if you go back to the original motivation of what's good and bad about these powers, the power that comes from being able to make the threats, it's positive sum globally, why?

Because it relaxes constraints. It expands the production frontier of the world economy. And so it describes a new pareto frontier that is better than the old one. But then the problem is, once I have that power over you, what the hegemon does is he picks a contract that is to the inside of the global frontier because he tries to redistribute well towards itself.

Even if he knows his distorting incentives, he doesn't care about the foreigners, okay? So this gives you a very simple baseline to think about what happens when you have macro amplifications. In particular here, you can see I only have micropower. All of the aggregates are not affecting anything cuz prices are constant.

And here, if you think about the word that people use, strategic. In what sense is a sector strategic here? Well, a sector is strategic if he lets the hegemon make valuable threats, okay? This goes back to Schelling in the 50s. In fact, recently I have written this beautiful page of the paper that it was all about how strategic has to be assessed in equilibrium.

And it did turn out that Schelling had written the same page in 52. So now it's out of the paper, but it's still true, which is strategic cannot be. I know the parameters of a CS function, which very often, we think about that informally in economics that way.

Well, that's one input, but strategic is an equilibrium, is the set of threats that I control. The typical example is controlling one variety of oil is useless. Controlling all of them is very valuable. That sort of starts giving you this baseline. Now, let me switch on macro. Okay, so how does macro amplification work out in this model?

Let's start think about a world Where I remove the assumptions that lead to constant prices, where prices are moving around. If you remember your Leontief inverse from grad school, what's happening is, as a side pillar, if I produce more, that changes the price of that output in equilibrium, but that means that it's changing the price of that one as an input to everybody else.

They'll respond, and so on and so on, and will get amplified until we reach an equilibrium. That's one form of amplification that comes to prices. The other one that you're familiar with is from aggregate, sorry, external economies of scale. If I produce more as a sector, that might make some other sector more productive, and that in turn will make some other sector change our productivity.

And we're getting an aggregate propagation that comes from external economies of scale, which are an externality. It turns out, and I believe that actually we're the first to do it in generality, but it's 15 lines of algebra. But it's kinda nice that you can put all of these things into a simple demand system and do the entire endogenous amplification in pretty much the same way you do Leontief inverse.

You can see it here. I'm not gonna go through the entire notation, but for any shock, the aggregate response of the vector of all outputs has two components. It has the direct component multiplied by a matrix that looks like a Leontief inverse. And then it has the effect on prices and the effect that those prices will have on output, okay?

So when you start thinking about macro, you have to think about, okay, if I'm the hegemon, I can manipulate the system in my favorite. I'm not gonna choose a contract for the world economy not thinking of what happens. Now, almost all of international macro and trade have been about terms of trade manipulation, essentially all of the WTO literature.

All of that sort of motivation comes from manipulating relative prices. Here we're gonna also allow to manipulate production externalities, okay, and you'll see an application to national security in a sec. Now, if you have that one, the contract becomes more interesting and more complicated, and I'm gonna try to unpack it.

Well, let me actually not go through this, which is long and complex, and directly do the foremost. So let me fix for a second a foreign sector with binding constraints, okay? What is the tax formula? What do I want to manipulate, and why? So the first part looks pretty standard from macro problems.

This epsilon here is capturing the marginal benefit to the hegemon of altering the activity that I'm imposing the tax on, okay? So terms of trade, input, output, amplification, all of that is inside this epsilon. What do I trade it off against is the fact that because I'm making you deviate from your private optimal decisions, I'm worsening your participation constraints and your incentive constraints.

So the tighter the constraints, the less I will ask of you. So you can immediately see that one of the things that power does for me is because I can offer you something very valuable, I'm creating slack in the constraints, I can then ask you to deviate from your privately optimal decisions.

So I'm asking Nvidia not to sell to China, or I'm asking European firms not to use Huawei technology. To each of them, that ask is privately costly. They comply because I'm offering to withdraw lots of other things in support if they don't comply. So that's what I'm constantly trading off, is how much can I ask out of you depends on how much power I have over you in the first place.

And now we're gonna go for, okay, but even if I can ask you, what exactly am I asking you to do, and why?

>> Speaker 3: What is the lambda bar?

>> Matteo Maggiori: These are the Lagrange multipliers on the participation constraint. And the sum combination of the ics is that the ics are entering the price I made.

 

>> Speaker 3: What makes you think that they're unique?

>> Matteo Maggiori: What makes you think that they're unique? These are only necessary conditions, so they're definitely not unique. These propositions here.

>> Speaker 3: Okay, so that formula could imply a variety of.

>> Matteo Maggiori: Yeah, these are only necessary conditions which i is saying, an optimal contract rather than the optimal contract.

The and there matters is if on an equilibrium, this will be the contract, it's definitely not the only one. Okay, now let me unpack it, cuz that's the interesting part. So if you take the benefits to the hegemon, what do they come from? They come from the direct impact.

This is like, I move that activity. Suppose, for example, I don't like it for a reason or another, that's a direct impact. But there is an indirect impact. There's a whole part of the world economy that I don't control. I don't have any threats on them, I don't even trade with them.

But they're gonna respond in equilibrium to the actions I asked out of the part that I control. In principle, I might care more about them than the ones I control. So think about thick market externalities. I might only control a small part of the world, but if I can get those people to be enough to move the equilibrium, what I enjoy is an overall outcome, okay?

And of course, I care about the impact on prices. So I'm gonna give you an example where you have national security that looks like this. But what I want to go through is, once you're thinking about macro, sectors that are strategic are now about something else. They're about their influence on the world economy.

I don't care so much about telling Nvidia not to sell to a particular firm in China. What I care about is that if they did sell, that firm in turn will use the chips to produce military technology. That military technology will be installed on ships. The ships might threaten my shipping lanes.

And in equilibrium, I have to go around, I don't know, some shipping route that before was open and now it's not. What I care about is essentially the full propagation. And I'm trying, in this sense, sectors that are strategic are those that have very high indirect influence. I'm twisting a few levers and letting the world propagate, which is a very different definition of strategic from the micro perspective.

Also, this one is one that you can take to a data. I mean, while these equations might look very fancy, this is what we always do when we take input-output matrix to the data. They can be measured with things like sale shares and domata weights and things like that.

 

>> Speaker 3: Where do I see in this formula the impact of my decisions about which sectors to penalize or not on the decisions of other hegemons that's somehow showing up here?

>> Matteo Maggiori: You wanna know, very good. No, no, that's not showing up.

>> Speaker 1: It's not showing up.

>> Matteo Maggiori: No, no, yeah, it's all right.

For now, I'm thinking about a single hegemon offering a contract. Everybody else is reacting, but everybody else is not offering contracts.

>> Speaker 3: Okay, thank you, I got it. Okay.

>> Matteo Maggiori: At the very end of the paper, we do a couple of-

>> Speaker 1: So there's only one hegemon?

>> Matteo Maggiori: Yes.

 

>> Speaker 1: No, not a multi.

>> Matteo Maggiori: We have a separate paper that hopefully we'll write at some point thinking about multiple hegemons. It's actually interesting, but there the strategic interactions get a lot more complicated. So at the end of the current paper, we have three pages on this, or maybe actually now that during the appendix.

But they're really just proving a couple of setups, like examples. We haven't characterized it generally with multiple Hegemons, that's a.

>> Speaker 3: So this is more like a dynamic principle agent problem.

>> Matteo Maggiori: Yes, but when you have multiple hegemons-

>> Speaker 3: All these little countries.

>> Matteo Maggiori: When you have multiple hegemons, you have a lot of other things.

 

>> Speaker 3: That is the game.

>> Matteo Maggiori: Yeah, exactly. Which is why it's interesting, it's also why it belongs to a different paper. Here the other thing that is important is the target countries, in some sense, take it on the chain. They're given this contract, there is a very clear rule here for anticoercial policy.

Suppose that you're the government of a small open economy being targeted by the US or China, what you might want to do is make them feel that the outside option for accepting the context is better than it really is to induce them to not give up so much surplus.

So that's yet another paper that we're working on, but they're not here. Here is a single hegemon. The rest of the world is small.

>> Speaker 5: So Matteo, the single hegemon, the industry structure is fixed because I'm thinking about it, I want to make it so that I increase my strategic path.

 

>> Matteo Maggiori: Absolutely, so the answer is yes, it's fixed. And the second part is, like you, I find it very interesting to think about endogenous formation of the industries. Now let me be clear. What's fixed here is the sectors you have to begin with and who this apply to and the rest of the world.

 

>> Speaker 5: Yeah.

>> Matteo Maggiori: So what I cannot do here is say I wanna build new relationships. What I can do, which is what this tax formula is showing you. If I have a sector that is strategic because it has a lot of influence on the rest of the world, and it's an activity that I want to grow, I would want to crank it up.

So I can make it bigger in the intensive margin, as in I can subsidize activities, which is what this formula is telling you. Think of, for example, my domestic semiconductor, is a sector that has a huge impact on the world equilibrium. I want to subsidize it a lot, given the constraints.

That is fully solved. The fact that I have that sector to begin with here is exogenous. And I think it would be interesting to relax it because it might lead, for example, to increasing returns to scale as a hegemon. Here, I'm not talking about what made you an hegemon to begin with, but very clearly, there is a sense of being big and having many sectors is kind of volatile, but it's too informal.

 

>> Speaker 8: Sort of fixed over time, right? It's like, it's not like I cut them out of all the semiconductors. They can never get semiconductors and they can't form their own semiconductor industry.

>> Matteo Maggiori: Yeah, that's exactly right.

>> Speaker 8: If you get too big on this, then you're gonna conclude, you get enough people.

 

>> Matteo Maggiori: So that's exactly right, so that's what the participation constraint does. If you try to ask too much out of them, they're gonna simply reject the contract. That's what I like about this paper is telling you, look, this isn't, I'm dictating rules to domestic firms, here everything is willing participation.

But what I'm not allowing is for a new sector that was never there to emerge or for new links to fall.

>> Speaker 8: If I control Swift, I can kick them out of Swift, but they can't kick them out of payments forever.

>> Matteo Maggiori: That's exactly right.

>> Speaker 8: Here you can kick them out.

 

>> Matteo Maggiori: Here you can. So that's a great example. So let's think about Swift. So Swift is a wonderful, for those of you that I'm sure everybody knows, but it's a payment system. So the reason why we care about them is you can see it here, they're essentially, they have very high external scale properties.

I like being on Swift because everybody else is on Swift. Now those are the typical things you use in threats, why? Because in equilibrium, everything else is an extremely poor substitute. They're the typical national monopolies where one system governs almost everything. Okay, so the US likes to use them in threats because that's a pretty meaningful threat to provide to other countries to kick you off Swift.

Now what am I allowing here? I'm allowing the US to subsidize Swift activities to make it even bigger because he values it. What I'm not allowing is either for the Chinese government to respond. That's the question from Steve. Here the rest of the world is mall, we're not considering the full game of them responding.

Second, I'm not allowing for somebody else to create a sector in their country that does this if it wasn't there to begin with. Those are given. They're all things that we would like to work on, but the first paper was fix the structure of the economy. Look at all the intensive margins of how you would like to manipulate the activities.

That's exactly right.

>> Speaker 5: Matthew, what I had in mind was also taking the example of swift. Swift becomes very important if first the biggest countries joined Swift. So in some way there is an optimal sequences, on how I allow my supplier to serve certain activity, and that could give me more power.

 

>> Matteo Maggiori: Yes and no in this paper. So in this paper, what you can definitely capture is through the external economies of scale is that having big players join your system impacts the value of the system to everybody else much more than small players. All of that we can allow, we have no dynamics.

So here there is no sense in which you build this slowly over time. We have a separate paper that has little to do with this, but it's about building that kind of dynamic construction. Here it's totally absent, what's present is just the idea of externalities impacting.

>> Speaker 8: So is this intuition, right, that if I think about a strategic sector, not in the hegemons country, but some little countries, think about oil in the 70s, right, was mainly in the Middle East, I think that's further right.

These contracts are less likely to solve that problem because everybody needs it and I don't control it.

>> Matteo Maggiori: What I might want to do as the US is I might want, if oil is, for example, controlled by three countries, I might want to find very powerful threats over those countries so that they get to do my bidding.

 

>> Speaker 8: Right, but it's harder to find an economic threat because no one's gonna go along with not buying oil because they all need it. So this tells me why we have military basis.

>> Matteo Maggiori: It also tells you that, yeah, if you make the oil supply very dispersed, for example, now the US is a big supplier, that minimizes the power that you get out of OPEC, because there is another willing supplier where it tells you the current situation, where it's very difficult to punish Russia if China and India are willing to buy.

This is a bit different from this because it's a supply on selling rather than buying, but it's relatively similar. We're done with the difficult part, so let me try to just do examples and applications, okay? First, I wanted to get back to what are strategic sectors in this model and how do you judge them?

So I wanted to distinguish micro from macro. So we mentioned micro is essentially the ability to give threats, stake prices and quantities as given. If I have valuable threats over particular sectors, that tells me the micropower I have over them, and a sector is strategic if it lets me make these threats.

Now once I have micropower over you, the actions I ask you to take might have to do with me exerting micropower. I might simply ask you to give me money, or I might ask you to modify the way you produce because that impacts somebody else that I care about and so on.

So market power here is measured very differently, is measured by. By looking at sectors with high indirect influence in a leontier sense. Now, the origin for macro-power is micro-power. So I can't have one without the other, but they're exerted for very different reasons. And you will measure them in the data very differently.

For everybody after I have questions about this, we kind of started thinking about this is a more general problem for example, think about regulatory capture. It's a little bit like this, I'm essentially a big player. And what am I doing? I'm asking people to do individual actions because I have some power over them.

They accept to do my bidding, but they think that the equilibrium is constant. But the reason why I'm actually asking them is I'm changing the equilibrium and they're not noticing that, that's my source of ultimate power. So it seems like a very general thing like all general things, I'm assuming some previous economists, defunct or otherwise, have probably done it, and I'm rediscovering it.

 

>> Speaker 8: But has the flavor of some antitrust issue.

>> Matteo Maggiori: Yes, he has a flavor of exactly that, and I would like to get deeper into that. So I would love to know more, because it seems that the modeling tools, these concepts are very general. And here I picked a country, but it could have been an industry.

It could have been like the financial industry doing this.

>> Speaker 8: The leading firm in an industry.

>> Matteo Maggiori: Yes, exactly, you could be a big player in an industry. It's coordinating this, and I quite like it, and it's measurable. So I love to know what people's reactions about this.

But then let me finish with this paper. So, how much time do we have left? Okay, 20 minutes, perfect. So, first I told you that how much you want to influence things, but not the direction. And this goes to the idea of political enemies and political friends. So there's a very easy way to see who are your friends and enemies in this equilibrium.

Pick anybody, like a country, a sector, a firm around the world. Look at how they affect your value function. So a friend is nothing else, that is somebody where all their externalities on you are positive. A neutral is somebody that has no externalities on you. And an unfriendly is somebody that has negative externalities on you.

And the sign of these objects tell you what you want to do with them. You want to tax unfriendly sectors, untax the neutral zero wedges and subsidize the friendly. Now, the nice thing is that these concepts aren't based on who you are, what you do, they're based on how you influence the equilibrium.

So Nvidia is unfriendly to the US for its activity of selling to China, despite the fact that overall we might want Nvidia to be very successful and big in the US. So they're about how you influence the equilibrium rather than simply some example characteristics, okay? Genetic inefficiency stuff.

So this is pretty intuitive, and I'm only gonna do it intuitively, even in the general sense of the full contract. In general, having an hegemon is inefficient, why? It's a bit of the same as the simpler version I gave you before. What an hegemon is doing here is moving up the Pareto frontier.

It's supplying very powerful threats. The fact that it's this big police watchdog around the world is making the world work better. It's providing more enforceability. But by the same token, that country now has power. And what it's gonna do is, it's gonna choose a contract that moves to the inside of the world Pareto frontier to maximize his own welfare.

It does it in two ways. First, it extracts markups and transfers or loans at higher rates because he can. Second, it imposes wedges that don't maximize world welfare, that maximize the country welfare. The typical example is China having a large semiconductor industry might be unfriendly to the US, might be very friendly for the rest of Asia.

And a world partner would have chosen a different set of instruments than the hegemon does. So this clarifies why I think the sort of newspaper version of what's positive and what's negative some doesn't work that well.

>> Speaker 8: So your way of thinking about this, how would you characterize on the first day Biden initiative an executive order saying that all benefit cost analyses in the federal government have to take account of global benefits?

 

>> Matteo Maggiori: Yeah, I don't think we ever do, I mean we care about our national benefits.

>> Speaker 3: Yeah, but he issued an executive order forcing agencies to take out of global benefits, not just-

>> Matteo Maggiori: Then they should, but the global benefits.

>> Speaker 3: I mean, so he's acting in a, is acting in a counter to your way of thinking about.

 

>> Matteo Maggiori: Let me put it this way.

>> Speaker 3: I'm not saying political, but that's an-

>> Matteo Maggiori: Yeah, no, no, I mean. It's pretty obvious that if you're acting in the global benefits-

>> Speaker 8: That only counted if it was global carbon reduction benefits, other political benefits, I don't think they were interested in.

 

>> Speaker 3: Well, maybe they weren't, but the executive order was for all cost benefits.

>> Speaker 7: They also said you have to count reductions in inequality as a benefit.

>> Matteo Maggiori: Let me stay on this for a second. Okay, let me do the examples. So I promise a national security example.

So let me give you the construction. This boils down the model to literally three or four sectors in total. So I'm gonna think of a country like the US that has some unspecified sectors. I don't care what they are, okay? But it has some threat on some firms in some sectors in the rest of the world, okay?

Potentially there are many in the rest of the world. I'm gonna assume just two, okay? And I'm gonna assume external economies of scale. So if i is using a technology from China, j finds on the margin that technology more productive and vice versa, okay?

>> Speaker 5: I and j are in the same country.

 

>> Matteo Maggiori: No, they could be in any genetic rest of the world, okay? China, I'm gonna assume, has a single sector, Huawei, that is selling technology to the rest of the world firms, okay? And this goes to the question of liking or not liking activity. This is the only time I'm gonna really use heavily the term in the utility function.

I'm gonna assume very directly that the US suffers a utility loss if the rest of the world is using this chinese technology. Now, you can write full models of that. I like to spy on enemies and friends, and the only way I can do it is if I control the infrastructure of telecommunication like they did for T&T downtown.

 

>> Speaker 7: We suffer utility loss if our electric cars aren't built by union labor in the US.

>> Matteo Maggiori: Yes, whatever it is, and that's why I said, I want to claim that we don't have a model of that. I want to claim that we have a simple model to capture it in the way that most of the time we do in economics, we determine the utility function, okay?

Now see what happens in this one, if you think about, okay, so sorry. First, the only thing I'm gonna focus on is that the US has some powerful threat over a part of the world economy. So we can ask them for some actions. In the particular example, Huawei, think of this as Western Europe, okay?

We can ask some governments in Western Europe to reduce their usage of telecommunication technology from China. The question is, how much do I ask them to do it and why? This is kind of interesting. So, this goes to the idea of why the externalities matter so much. Well, first I want to ask the government that I control I, to you to use it less.

That's a direct benefit from me, because they're not using it, I don't like them using it. That's a direct benefit. Second, I know. That the other governments or the other sectors that I had no relationship with, I couldn't ask them to do anything. Because of externals economies of scale Germany will stop using the technology if England is not on it, because the technology all of a sudden isn't so valuable to Germany on the margin.

Now I'm fully incorporating that. So what that means is that I'm gonna really push hard on the government that I do control. I'm gonna ask them to do more than I would have otherwise asked them because I want to use their indirect influence. Also this last term, which is the indirect effect on the participation constraint, tells me that to the extent that I'm moving the equilibrium in my favor, it makes it easier for people to accept, why?

Think about the beginning the first time, everything else is given, I ask England to not use this technology, they'll complain a lot. They'll perceive it as very expensive cuz they're thinking of a world where France and Germany are on it. But as this amplifies and France and Germany are using it less, England itself is gonna perceive it to be less valuable to be on it.

So I'm tightening their participation constraint less. So what I'm really doing is I'm totally manipulating this in my favor through both the effect on others, and now it filters back on the people that I do control. And ultimately I might get to switch the equilibrium even if I control only a small part of the world economy.

So that's a pretty good example of the macro aspects. Now it also tells something else, which we're not doing in this paper, it's a separate paper. Suppose that you're a government being targeted from a foreign hegemon. You can be sure that the hegemon isn't picking at random. Here the hegemon love strategic sectors, why?

From a macro perspective, because they have the largest gap between the private cost to all of the firms in your country participating and the social benefit. The hegemony is essentially distorting their margin against you. So they're guaranteed to pick or to want to use the most sectors that have very high indirect influence because everybody's gonna participate, because they're not taking into account the overall matter effect.

Now, as a government, that gives you a rationale for doing anticoagulant policy is to make every sector or every firm in your country understand that they shouldn't participate so willingly because they're changing the overall equilibrium, okay? We're not studying those policies here, but I would like to study them.

Okay, last example and then we're done, is I'm going to think about the China Belt and Road Initiative and I'm gonna make it super simple so you all know what it is approximately. But in this particular framework I'm going to think of it as I have two sectors as China, so my hedge one is now China.

I can lend, so think of it as making loans, this could be sovereign lending, there could be official loans, and I'm providing some manufacturing input, okay? To affirm here, I'm gonna pick the target country to be say a small open economy in Africa, okay? And this actually connects to Jeremy's wonderful work we can on the bularog of results.

So here I'm gonna start with, okay, I'm making you a loan each period. It's a fixed interest rate. If you default, I'm not gonna trust you. I'm gonna assume that there is no legal enforceability, so the counter is totally unenforceable, okay? But I also have a valuable manufacturing activity and by converse I'm gonna say that the manufacturing activity is fully enforceable.

There is nothing that you can do about stealing it. So think about what happens. First, if you do isolated threats, you can only sustain a small amount of lending and this actually goes to your work. Here, in fact, we're rigging it because if we allowed upfront contracts and savings, we will get back to your result of you can't sustain any lending at all.

Here the only reason you can sustain some is I'm not allowing you to write up front contract. But the second part is really a suggestion from their paper where they say, look, one thing that I can threaten you with if you default on me on a loan is with other relationships that have nothing to do with the loan that are themselves very valuable, like trade.

That relationship is gonna essentially act as a sort of collateral for the loan. And here you can see that what it does is that it shifts up the maximum amount that you can borrow. Now from the perspective of these countries, it tells you though, something else, there's a large empirical literature trying to measure, is China getting a good or a bad deal on the loans?

Was the Marshall plan or the US a good or a bad idea? This is telling you to be very careful with the analysis, why? Because both the sustainability and the actual money that you, the actual profits you make out of this might not come from the interest rate on the loans.

For all we know, China might be making horrible returns on the loans, but if it's getting political recognitions or to use naval bases that might be worth a small fortune, or it's securing critical mineral supplies. But it's not just the outcome, it's also that the fact that the activities are sustainable in the first place is because they're jointly coordinated.

So it gives you a very different way or a cautionary tale of how to assess these projects. Well, I think a lot of the sort of literature on China has been on the interest rates on the loan, but that's like a thermal, like another. It's part of a big threat with tons of other things, and where you see the profits or the losses might be in seemingly very unrelated activity the moment you think of it as a hegemonic power.

Okay, I'm gonna largely stop here. Lots of things came up during the talk. I mentioned that there's a few that we're working on, so I wanted to give you a sense. So first we're working on the empirics. A lot of this framework, despite the fact that it looks, at least to me, rather complex, it's pretty easy to, it's pretty measurable, at least by the standard or macro models.

Second, we mentioned geopolitical competition. What if you have multiple hegemons and you're studying a full game where they're both offering contracts? So we've only done minor sketches of that, but we'd like to do a lot more. Similarly, we mentioned that even if you're not a hegemon, you're a small country in this game.

We're not studying your optimal defensive policy. Here, your sectors are behaving, sometimes competitively, and accepting contracts. Right now a lot of governments, including the governments of western Europe, are passing what is called anti-coercion policies. It's unclear, what are you trying to correct. Why are you doing them? There's a nice set of policy tools that can be studied in this framework that we haven't introduced yet.

That's it. Hopefully, this gave you at least my rendition of what I think is a very large topic that had been out of economics for many, many years and that I think we can treat with modern tools quite efficiently. I'm sure people will have different ideas about to do it, better ways to do it, but I think it shouldn't be a topic that we don't deal with.

In fact, I'm gonna argue that there's lots of good standard theory tools n empirics that can be brought to bear on these questions. And there seems to be a large policy relevance, and I think collectively as a field, we should have good things to say about this. And right now it's pretty massively understudied.

Thank you.

>> Speaker 1: Thank you.

>> Speaker 1: One question.

>> Matteo Maggiori: Please.

>> Speaker 3: Can I give a summary, you tell me where it went off. So already we had a monopsony power and international markets in these arguments. Already we had domestic increase in returns. But you seem to be telling us, well, the value of those things is more than the sum of the individuals.

 

>> Matteo Maggiori: That's correct.

>> Matteo Maggiori: No, no, no, I don't think you're wrong, I think here the notion of power is much more general cuz it's not just market power. In some sense, markups are one of the things I might ask you, but my power to ask you to do costly actions.

But also it's different from the typical micropower cuz some of the actions are driven from changing the equilibrium and not from simply extracting surplus out of you. In fact, I like those that extract minimal surplus out of you and affect the equilibrium a lot because I can convince you to do a lot of those for me.

That's exactly right.

>> Speaker 5: The mechanism is different but also in the future work, you have the potential of the creation of coalition because even a small country, I mean, the Huawei story is very interesting because at one point, for example, Italy positioned to acquire it. And they were playing a game, obviously, and this is a small country, but in coordination with some of the other countries we're actually trying to position toward one edge of versus the other one.

And also with China relationship between Italy, and the-

>> Matteo Maggiori: I think there's a lot of scope for that. I mean, here a very obvious coalition or cartel is OPEC. Each of the country would have little power for the threats, but threats for more of them, since they're a large part of the world's supply of oil are very valuable.

But I think what you have in mind is a little bit more of like if I have two hegemons, how do I play against each other?

>> Speaker 5: Yeah, that's what I'm thinking.

>> Matteo Maggiori: Yes, absolutely. So I'm very interested in that problem, we've only started sketching it.

>> Speaker 1: Thank you.