If I were giving the Nobel Prize to Jean Tyrol, I would give a reputation for his game-theoretic analysis, or at least include it in the wording. It seems to me that this is the case when our intuition fits the model well, although it is difficult to test this model. This is from a series of those models that are difficult or impossible to verify and falsify. But the idea seems absolutely brilliant to me.
The Nobel Prize
The background of the premium is the final departure from the unified concept of general equilibrium as an analysis of any economic situation.
I apologize to the economists in this room, I will popularize the basics of general equilibrium theory in 20 minutes.
1950
The prevailing view is that the economic system is subject to strict laws (as physical reality is subject to Newton's laws). It was a triumph of the approach of unity of all science under a certain common roof. What does this roof look like?
There is a market. There are a certain number (n) of households, consumers of goods, those for whom the market works (goods are consumed). And a certain number (J) of the subjects of this market (producing goods). The profit of each producer is somehow divided among consumers.
There are goods 1,2…L. A commodity is something that can be consumed. If a product is physically the same, but consumed at different times or at different points in space, then these are already different goods.
Goods at the time of consumption at a given point. In particular, the goods cannot be durable. (Not cars, but rather food, and even then, not any).
So we have the space RL of production plans. L-dimensional space, each vector of which is interpreted as follows. We take the coordinates where the negative numbers are, put them in the "black box" of production, and give out the positive components of the same vector.
For example, (2,-1,3) means that from 1 unit of the second product, we can make 2 units of the first and three units of the third at the same time. If this vector belongs to the production possibilities set.
Y1, Y2… YJ are subsets in RL. Every production is a black box.
Prices (p1, p2… pL)… what are they doing? Falling from the ceiling.
You are a company manager. A firm is a set of production plans that can be implemented. What to do if you received such a signal - (p1, p2 ... pL)?
Classical economics dictates that you evaluate all the pV vectors that you are allowed to evaluate at these prices.
And we maximize pV, where V is from Yj. This is called Pj(p).
Prices are falling on you, you are told, and you must unquestioningly believe that prices will be like that. This is called "price taking behavior".
Having received a signal from the "prices", each of the firms issued P1(p), P2(p) ... PJ(p). What happens to them. The left half, consumers, each of them has initial resources w1(р), w2…wJ(р) and shares of profits in firms δ11, δ12…δ1J, which will be generated on the right.
There may be low initial ws, but there may be high shares, then the player will come from a large budget.
The consumer also has preferences א. They are predestined to him and unchanging. Preferences will allow him to compare any vectors from RL with each other, in "quality", from his point of view. Complete understanding of yourself. You've never tried "banana" (I tried it when I was 10), but you have an idea of how you would like it. A very strong informational assumption.
The consumer prices his initial supply pwi and adds profit shares:
The consumer also unquestioningly believes the prices that have arrived and evaluates his income. After that, he begins to spend it and go to the border of his financial capabilities.
The consumer maximizes his preferences. utility function. Which xi will be more useful to him. The paradigm of rational behavior.
There is complete decentralization. Prices are falling from the sky. At these prices, all firms maximize profits. All consumers receive their bills and do whatever they want with them, as they like (maximizing the utility function) spend on available goods, at available prices. There are pro-optimized Xi(p).
It is further stated that prices are equilibrium, p*, if all decisions of economic agents are coordinated with each other. What does agreed mean?
Was what? Starting stocks, each firm added its own production plan:
This is what we have. And it should be equal to what the consumers requested:
Prices p* are called equilibrium if this equality is realized. It has as many equations as there are goods.
It's 1880 it was promoted in every possible way, and for 79 years mathematicians-economists have been looking for proof that such an equilibrium vector exists. This reduced to a very difficult topology, and could not be proved until 1941, when . In 1951, the equilibrium existence theorem was completely proved.
But slowly this model has flowed into the class of the history of economic thought.
You have to go all the way yourself and study outdated models. Analyze why they didn't work. Where were the objections? Then you will have an experience, a good historical digression.
The history of economics should study the above model in detail, because all modern market models grow from here.
Objections
1. All products are described in an extremely abstract way. The structure of consumption of these goods and durable goods are not taken into account.
2. Each production, firm is a “black box”. It is described purely axiomatically. A set of vectors is taken and declared admissible.
3. "The invisible hand of the market"prices are dropping from the ceiling.
4. Firms stupidly maximize profit P.
5. Mechanism of coming to equilibrium. (Any physicist here starts to laugh: how to "feel" him?). How to prove its uniqueness and stability (at least).
6. Non-falsifiability of the model.
Falsifiability. I have a model and according to it I say that such and such scenarios cannot happen in life. Such people can, but such people never, because my model guarantees that there can be no equilibrium in that class. If you present a counterexample, I will say - this is the limit of applicability, my model is lame in this place for one reason or another. With the theory of general equilibrium, this cannot be done, and here's why.
Because... What determines the behavior of an economic system out of equilibrium? For some "r"? You can build an excess of demand over supply.
We throw prices from the ceiling and know exactly which goods will be in short supply and which will be in excess. About this vector, we can say for sure (1970 theorem) that if the trivial properties are met, then it is always possible to build such an economic system (indicate the initial data) in which this particular function will be a function of excess demand. For any given prices, exactly this value of the excess vector will be returned. You can simulate absolutely any reasonable observable behavior using the general equilibrium model. Thus, this model is not falsifiable. She can predict any behavior, this lowers her practical meaning.
In two places, the general equilibrium model continues to work explicitly. There are computable general equilibrium models that consider the macroeconomics of countries at a high level of aggregation. Maybe bad, but they count.
Second, there is a very nice little spec where the production part changes and the consumer part stays pretty much the same. These are models of monopolistic competition. Instead of a “black box”, a formula appears for how production works, and instead of an “invisible hand of the market”, it appears that each firm has some kind of monopoly power. The main part of the world market is monopolistic.
It is important to note that claims are made to the economy in a tough way: “The model should predict what will happen tomorrow” and “What should be done if the situation is bad”. These questions are absolutely meaningless within the framework of the theory of general equilibrium. There is a theorem (the first welfare theorem): "General equilibrium is always Pareto efficient." It means that it is impossible to improve the situation in this system for everyone at once. If you improve someone, it is done at someone's expense.
This theorem is in sharp contrast to what we see around us, including the seventh point:
7. “The goods are all private and there are no externalities”.
In reality, a huge number of goods are "tied" to each other. There are a lot of examples when economic activities influence each other (discharge of waste into the river, etc.). Intervention can bring improvement for all participants in the interaction.
Tyrol's main book: "The Theory of Industrial Organization"
It is impossible to expect that the markets will interact effectively, produce an effective outcome, we see it all around.
The question is this: How to intervene to correct the situation? Why not make it worse?
It happens that, theoretically, it is necessary to intervene, but, practically:
8. The information needed to intervene correctly is not enough.
In the general equilibrium model, it is complete.
I already said that this is about people's preferences. When intervening, you need to know these preferences of people. Imagine that you are intervening in some situation, you will begin to “improve” it. You need to know information about who and how will “suffer” from this. It is probably understandable that economic agents that will suffer slightly will say that they will suffer very much. And those who win a little - that they will win very much. If we do not have the opportunity to check this, get into the head of a person and find out what his utility function is.
In the "invisible hand of the market" there is no pricing mechanism, and in
9. Perfect competition.
The modern approach to where prices come from, the most popular, is that someone who organizes the market announces the prices. A fairly large percentage of modern transactions are transactions that go through auctions. A very good alternative to this model, in terms of distrust of the invisible hand of the market, is the theory of auctions. And the key point here is information. What information does the auctioneer have? I am studying now, I am an official opponent on one of the dissertations, which is done in Yandex. Yandex conducts advertising auctions. You are being "pushed". In Yandex, work is underway on how best to “push” you. The dissertation is absolutely brilliant, one of the conclusions is completely unexpected: "It is critical to know for sure that there is a player with a very large stake." Not averaged (there are 30% of advertisers with an extremely strong position, requests), then this information is nothing compared to the fact that you know that one has definitely entered the market and is now trying to insert this advertisement. This additional information allows you to significantly change the threshold for participation, significantly increasing the revenue from the sale of advertising space, it's amazing. I didn’t think about it at all, but when they explained the mechanism to me and showed me the mathematics, I had to admit that it was so. Yandex implemented and really received an increase in profits.
If you're intervening in the market, you need to understand who's preferences are. It becomes not obvious that it is necessary to intervene.
There is also a superficial understanding that can be completely wrong. For example, a superficial understanding of a monopoly is that it is better to regulate a monopoly, for example, break it up into two, three, or four firms, an oligopoly will arise and public welfare will increase. This is typical textbook information. But it depends on the circumstances. If you have durable goods, then this model of behavior for the state can be completely detrimental. No. 0 years ago in reality there was an example.
They began to release records "Rock Encyclopedia". At our school, some copies were running around, which said that they were limited edition and sold for 40 rubles. 2 months passed and all the counters were littered with these records and they cost 3 rubles. These people tried to mystify the public that this is a perfect exclusive. A monopolist, if he produces a durable product, he begins to compete with himself tomorrow. If he tries to sell for a high price today, tomorrow this something can be resold/repurchased. He has a hard time convincing today's buyers not to wait until tomorrow. Prices are lower than usual. It was .
There is the Coase Hypothesis, which is that a monopolist with a durable good, which quite often revises its pricing policy, completely loses monopoly power. Subsequently, this was rigorously proven on the basis of game theory.
Suppose you do not know these results and decide to share such a monopoly. An oligopoly of durable goods emerged. It must be dynamically modeled. As a result, they support the monopoly price! It's the other way around. A detailed market analysis is extremely important.
10. Demand
There are millions of consumers in the country, aggregation will be performed in the model. Instead of a huge number of small consumers, there will be an aggregated consumer. There are many problems of both theoretical and practical importance associated with this.
Aggregation is in conflict with preferences and utility functions. (Borman, 1953). You can aggregate the same ones with very simple preferences. The model will be lossy.
In the aggregate model, demand is a black box.
There was some airline. She had one flight a day to Yekaterinburg. And then there were two. And one of them leaves at 6 am from Moscow. For what?
You fragment the market, and for the "rich" who do not want to fly early, set the price higher.
There is another objection to rationality. That people behave irrationally. But on large numbers, a rational form gradually emerges.
If you want to study economics, study the general model first. Then “begin to doubt” and examine each objection. A whole science starts from each of them! If you study all these “chapters”, you will become a very competent economist.
Tyrol lit up in the study of several "objections". But I wouldn't give him the Nobel Prize for that.
How to build a reputation
I suggest you think about these stories. and when I tell you about the reputation, we'll discuss it.
In 2005, an unprecedented reform was carried out in Georgia. ALL militia of the country was fired. This is the first story.
Second story. After the dispersal of rallies in 11-12 years in Moscow, all policemen received sleeve numbers and stripes with surnames.
These are two different approaches to the same problem. How to deal with a country or a group of people with an extremely negative reputation of some community within?
“Fire everyone and hire new ones” or “personify violence.”
I affirm and will refer to Tyrol that we have taken a more competent path.
I give you three models of reputation. Two were known before the Tyrol, and he invented the third.
What is reputation? There is a dentist that you go to and recommend this doctor to other people. This is his personal reputation, he created it for himself. We will consider the collective reputation.
There is a community - millionaires, businessmen, nationality, race (the West does not like to discuss some terms).
Model 1
There is a team. Inside which each participant is written "on the forehead." Coming out of there, he is already with some kind of sign. But you can't tell by a person from this group whether he is or not. For example, when students from NES are accepted for Phd programs in the USA.
In general, America despises the rest of the world. If there are no missiles, then he despises; if there are missiles, he despises and is afraid. She treats the world in such a way and at the same time, like a fisherman throws a fishing rod ... Oh, good fish! You will become an American fish. This country is built not on the original fascist principles, but on created ones. We will bring together all the best and therefore we are the best.
Someone from the "third world" comes to America and it turns out that he graduated from NES. And then something lights up in the eyes of employers. The grade for an exam is less important than the fact that it came from NES.
This is a very superficial model.
Model 2
Not politically correct at all.
Reputation as an institutional trap.
Here is a Negro comes to work. (In America) You are an employer, look at him: “Yeah, he's a black man, I basically have nothing against blacks, I'm not a racist. But they are, in general, just stupid. So I won't take it." And you become a racist "by actions", not by ideas.
“I don’t know if you’re smart guy, but on average, people like you are dumb. So, just in case, I will refuse you.”
What is the institutional trap. 10 years ago this guy went to school. And he thinks: “Will I be as good at studying as my white classmate? What for? All the same, they will take only low-skilled work. Even if I work hard and get a diploma, I won't be able to prove anything to anyone. I know how everything works - they will see my black face and think that I am the same as everyone else in my group. It's such a bad balance. Blacks don't study because they don't get hired, and they don't get hired because they don't study. Stable combination of strategies of all players.
Model 3
There is some interaction. Which happens between a randomly selected person from this population (the people) and (the police). Or businessmen-customs.
I have a businessman friend who often communicates with customs, he confirms this model.
You have a need / desire of a person (from the people / businessman) to contact (the police / customs) and give him some kind of “task”. Understand the situation, smuggle the goods. And he expresses, thus, an act of trust. And the person on the spot makes the decision. He has no forehead seal (model 1), no decision to invest in himself (model 2), nothing that predetermines how he will work today. There is only his present good will.
Let's analyze what this choice depends on and where the trap occurs?
The man looks at the official. Tyrol suggested only one thing, a thing dubious in its meaning. But she explains everything. He suggested that it is unreliable known about this official that he had done before. In other words, there is a story about everyone. Here about this militiaman, in principle, it can become known that he used to extort money for the performance of his work. We heard stories about this customs officer, how he delays cargo. But you may not have heard.
There is a theta parameter from 0 to 1, that if it is closer to zero, then you get away with it. Roughly speaking, if a policeman does not have any license plates, he can beat anyone up, no one will know about it and nothing will happen to him. And if there is a license plate, then theta is close to one. He will bear a heavy cost.
In Georgia, they decided to chop off the complete lack of faith with an ax. They have recruited new police officers and think that the old reputation will die. Tyrol argues that what kind of dynamic equilibria exist here ...
How are balances arranged? If an official is approached, then they consider him honest. A person can act really honestly, or act badly. This will partially determine my "credit history". Tomorrow they will not contact me if they find out that I have behaved dishonestly. The average faith towards nameless officials is very low. The next day, there is a small chance that you will be contacted. If you have already applied, then this is a rarity and you need to squeeze out the “maximum” and rob. We are all thieves and crooks here, and no one will turn to us anyway. We will continue to be thieves and crooks.
Another kind of dynamic equilibrium is that people believe that officials behave well and are approached a lot. So tomorrow, if you have a clean reputation, you will have many offers. And if you piss yourself off, then the number of appeals to you personally decreases. And this is an important aspect. If you have such a belief, you lose a lot from bad behavior.
Tyrol shows that in dynamics, which equilibrium develops depends critically on theta, and not on initial conditions.
By introducing theta, you increase the individual's personal responsibility. If he does well, it will be recorded for him, he will be addressed, even if they will not be addressed to others.
Source: habr.com