1. Preference

so hello, everyone. My name is Nan, and I’m assistant professor and the Department of Information System and Analytics of previously. I I was here for 3 years, and then I actually work in the industry for a couple of years. Then I came back to school, computing at us. So It was fine, and I did my Phd. At Uc. Berkeley. That was many years ago before Covid and my research is on competition. And you know, a few call industry organization or using data and models. I thought, sort of to understand firm strategy, consumer preference, and also policies. I’m happy to talk more. But most of the my research is not very relevant to this class. And this class is economics of it. And AI, okay. I assume the it’s going right on. Zoom, okay, So this, master, we have a larger class than it was before, I think. Last time when I taught it was 160 students, and this time it’s almost double to 250. okay. So I made quite a lot of quite, quite a bit of adjustment. To the plan teaching plan. Okay. let’s let me 1st go through the syllabus our lecture by default is physical and in class in Lt. 19. So I do encourage you to come to our lecture, and as we move on we may have more, you know, in in class interactions, and some, you know, some, some games, and so on. And I don’t believe you know, in class is more effective way of teaching. So I use that as a strong default. I strongly encourage you to to come to the lecture hall and to really sit here. And okay, button is something going on. Okay? So I strongly encourage you to come to lecture as default. But but we will offer online sessions so as well as recording because students always, you know, feel the classes a bit difficult. So they want to watch the recording again. I think that is a valid reason. And my way of teaching this class involves a significant amount of theory by which I mean theory of economics. I think there is a comparison. We talk of mathematics, which, let’s say, at a college at a graduate level, we talk on mathematics is no longer computations, right? And so, similarly, when I think about economics. I’m going to teach you more fundamental. Foundations of the theory start from a scratch in the 1st lecture. We’ll actually study start from the concept of preference. Okay? And that involves some mathematical modeling. And we want to be very precise about what we are talking about. And we are aim to capture the fundamental element that is not moving around. That is stationary. That is what we call theory of how things works in marketing society. And that’s the fundamental idea of economic theory have very simple model that can tell the fundamental of the world and how things works. Okay, and this is my name. And the Chinese is the character actually correspond to the gender or the Chinese character. You see. restroom. Okay, that’s my name. And we have actually more. Ta, because of the size of the class. Here, at least 2 of them. I will update this slide. Update this syllabus. We will have more. We have one more, at least one more ta from the Cs Phd, and we have some excellent undergraduate from this course in previous years, and he will join us, and and they will sort of help to provide more customized or individualized attention and try to help our students in this class to to learn better. Okay, let me see if there’s anything going on in Zoom. I can’t seem to hear anything. Okay, I got this message. Is it true for everyone in Zoom? Can you hear me? Can you zoom me in place? I can hear. Okay, okay. I can hear. Okay, I, I will assume there’s something wrong with your computer. But okay. okay, so next time you can chat in the meeting. Okay? And if you can confirm with each other is something going wrong. And signal me, okay, call me okay. let me try to come back. Okay. Yeah, we will have more students more more teaching assistants. And I will definitely encourage you guys to reach out to find you theory ta or tutor. Okay, to work with them and to learn from them. I would expect they will have some free time each week, and can interact with you individually to listen to your questions and try to provide answers. and I would encourage you to use canvas discussion to post your question, and that will contribute positively to the class because others can get a question, and then some some peers classmates can help you answer the questions or the the radio question. You will also get inspired, or they can learn from the answers. Collaboratively, okay. And I think most of the time. You do not need to email me personally. And what you can do is you can email our Ta and CC, me, keep me in the loop. And, for example, most of time, I think they should be able to help you on the individual need. And okay, and there’s 1 small detail try to use email instead of canvas for emails. Okay, some of you use canvas and for for emailing. And that is slightly not convenient in my experience. Okay? So those are for communications. Okay. music screen and the the grading policy or the component of assessment is the following, and made a quite a bit of updates given. Observing the you know, the growing size of the class. We will. We, I think, most of the variation in grades or assessment come from the midterm and the final, which involve close book exam of Mccus. Okay, and our midterm exam. Will be, I think it’s the week. Oh, okay, I think those of you cannot see the screen. Okay. okay. okay, and the other component 5% for class engagement, which involves, you know, contributing to the discussion, as later on we will have the class interactions including play games, and so on. and also your contribution to canvas. But I would say that there won’t be much variation in this component. Okay, it’s more useful in a case where, when a student is at near the cutoff, for example, just below a minus or something. Then this component may help you a little bit you know, getting a or something, but most of the time the variation. Like, I will say, 99% of variation will come from the other component assignment exam assignment. There will be 3, and they are individual assignment. And they are also Mccus. Okay? And there will be 3 of them. Actually, the assignment could be, some question could be challenging, and there are 2 there are at least 2 objectives for the assignment. So 1st one is that it helps you. Apply what you learn in the lecture in broader way. Okay, so that means it involves some self learning. Okay, you will. You will sort of realize and explore the thing you learn in class, and you realize that they are applicable or they are generalizable. They can be used to analyze a lot of problems. Okay, again, what you learn in class are very fundamental, the core like methods or models. And you will learn this, and you are actually able to build your own theoretical model to analyze the phenomenon, everyday phenomenon, or it and AI phenomenon or firm strategy. Phenomenon, consumer choice. And you will see a lot okay, in this assignment those are individual assignments. You have to submit your answers. Which involve the answer of the Mcq’s and your write up like the the process of solving the problems. Okay? And the the like. The write up is in Pdf format and the the okay okay for participation. I would strongly encourage you to come to lecture. But yeah, but if it’s cool, then you can also do it on on canvas. Okay? But again, there will be no penalty for participation. It’s mainly reward base. Okay? I I really do not want to emphasize this component because it’s the most marginal component in the in the whole assessment. But yeah, okay. it’s very marginal. Okay. And the midterm. I’m not sure if I already mentioned is in class in Week 7 we will reserve 2 h in our lecture, but the actual midterm will be less than 2 h. Okay, okay, the time is our lecture time. But we definitely need a a pretty large value. Okay? And I will wait until school will the undergraduate office will have the arrangement find a valuable okay for assignment, I will exercise that. You have to submit on time. Okay, we have a timeline here sort of to keep you contract. Typically, you will have 2 to 3 weeks for each assignment. As I said, they are mcus. But you are required to input your final answer, like Ppc, right? As well as you write up, your right about will be Pdf format. You upload these 2 panels. Okay. okay? And there’s this tentative course schedule and you will see how it goes. Okay, do I have anything else? Okay, we have a textbook which I shared on canvas. Here you go this one, and there’s a book. I will call it a reference book, because you do not need to know everything in the textbook. You can just focus on the lecture slides. That will be the key reference. And as well as our assignment. Okay, assignment questions. and I will not give further a practice question for the exams, because assignment question will be sufficient. Okay, it will be will be similar style. And the difficulty I will occasionally, or as the course goes, I will add some more reference or reading to the campus website. Okay? And you can, if you have interest and you can check it out by yourself. Okay. okay. I think that is for admin of this class. Is, is there any question from students in Lecture hall or from those on zoom any questions. If no, then I will just move on to our 1st lecture. Okay? Okay. okay, so we’ll start from the most basic concept or theory in economics, which is preference preference basically measures our preference towards different situation. Different objects. And you have people have different tastes, different valuation. They derive different pleasure or cost from different things. Right? So preference will be, a function from object or situation, or contact or States right to some numerical value. And that measures you know the relative ordering of things that you like. You divide more pleasure to something. Yeah, you have a stronger preference for it. It could be positive or could be very large. Okay. so and that is important. Because you will start from this point we’ll be a very quantitative. And we’ll build everything from here. Okay, we’ll build the market and build the firm decisions. Later, we’ll build innovation competition all from this. Starting point element of preference. Okay? And in this lecture we will divide demand from profits in your earlier education. If you study some economics or you’ve heard about economics, I think the 1st and most important concept you heard is demand and supply right. And the claim is that demand is actually built or derived from this more fundamental concept called preference to this lecture, and we will derive the demand or model demand from a customer preference. and you start from the motivation. So if we want to make some judgment on something being good or bad, right from a firm perspective, we want to make money. We want to know what is the correct action that can make us. Make our business more profitable. Right? Make more money from government perspective or policy maker perspective. It cares about social welfare, right or social surplus. It cares about everyone happiness. So, and we will put some structure on awareness, or pleasure or happiness, and we will call it utility or preference, and we to drive certain functions. And that is the key sort of framework. Where we making claims, such as something is some a policy is good policy. A policy is a bad policy, a firm strategy. How much profit it makes, and they each have an idea of how consumer behave. And you can think of consumer being a function, and there will be some instrument or strategy firm can choose and consumer will respond to that. that strategy or that instrument according to some function. And that function is built based on consumer preference. Okay? And let’s say, if you’re an e-commerce company, you’re a seller on Amazon, or you or you. You are Amazon, right? You are trying to set a price for your product. You can think about your price as an instrument, as you increase your price and consumer will will respond right? And you cares about your profit, right? Your profits, price minus your cost times, the quantity or your sales. Right? Which is demand, okay? And demand will be a function of your price right? And then demand function like, how much consumer hates price. It’s an object. You cares a lot when you’re making this profitable business. You want to know that if you lower your price by 1% well, you will expect. I think most of us will expect our sale goes up right lower price, and people tend to be happier if they pay less. and that is one property of the demand function. You will learn later. but you also lose when you lower the price right, because per piece you will earn less money when you lower the price, so you can immediately say that this is not a trivial decision, or it will not predict a corner solution in price. Right? Imagine pricing is your instrument by shifting price, you will not. The optimal price will not be 0, right? Because you’ll be losing money. Even. You’ll sell a lot given for free right? So the price has to be greater than 0 and price has not, cannot be infinity. Right? So there will be some interior solution to price. And in today’s like, say, e-commerce, and that will be done in a very quantitative way by estimating something called demand function. And that demand function is based on. we observe in data how consumer. how they make choice, and we will sort of estimate some models from this review preference approach, and that is based on the theoretical model of consumer, demand consumer preference and same thing. We will, after we demand this, define this preference, we will have something called utility, function and defense curve, and the end, after we derive demand function, and we will have a concept called consumer surplus and later lecture. And we will have other components like producer surplus. And if you combine those concepts or element together, you will get something called social welfare. If you are government or you are policy maker. Who cares about the aggregation of welfare, right aggregation of profits, maximize the total pie, and this will be the objective function. If you’re a firm, you probably care less about the consumer. If you are myopic, firm and maximize. Your per period profits right, and you will maximize your your profit. But your government, you probably cares about a combination of firm profit and and consumer consumer welfare. And today’s lecture at the end, we will have a definition of consumer welfare. Okay, now, let’s get started utility function. So there are 2 term, one’s utility, meaning that you, you will get something from, let’s say, a, a product, a cell phone or computer and function means that function defined from some space, right space of of anything, a large space to a number, right? The key property of a function is it? Give you one number you give you plug, an input, right? It will return you a number so that is the structure we’re talking about here. And suppose a consumer is facing 2 goods or 2 products. When is apple and one is banana. and we can define a function. And this function will measure the amount of pleasure or utility consumer can derive from consuming the 2 alternatives, apples and bananas or cell phone, and the laptop, right or cell phone and the ipad or something. Okay? So utility here would be a function in which take 2 inputs in our simple setting. Qa and Qb, right? Quantity for the product, and it can be very general. So this looks like a toy model. But it could be very general. For example, you can have many, many imports. Right? You can have an apple banana, and you have some other fruits. Okay? And you can have apple banana on raining day. Okay, you can have apple banana, you know, conditional adding states, right? So it can be as sophisticated as you as you will. Okay? And it could be some other not even like products it could be. How much time you spend for a class, and how much time you spend on Tiktok right? Or it could be your your pleasure, the pleasure derived from a certain type of Tiktok video or Youtube videos, right? So your function can be as general as your contacts required. Okay? And and it will refer to a different combinations. So as long as you have this quantitative function, we can define a mapping from this space or product or state or anything. Okay, can be a. It could be an infinite space. It can be many, many different combinations. But mostly when we are in a particular analytical situation, we will sort of focus on certain set of alternatives or set of choice. Okay? And you will have a because it’s a function, remember, mapped. So you return your number right? And you learned, I think, since primary school, that if I give you a number 2 number, you can compare the value, your bigger one small one. Right? Okay? So that means you can compare utility. Right? So something will give you larger utility, meaning that you like more. Okay, in our 2 alternative context. Here, we can see that this Qa. Prime, Qb, prime will give you greater pleasure. Okay, you like it more than it’s Qingqb, the 2 combination. Okay? And based on our, you know, intuitive understanding, we can say something about the values. Right? For example, if if I like, if I like apple banana, I don’t hate them right. I like them, or if I, the cost of this throwing them away is 0. Okay? Meaning that having more is not worse. Right? At least weekly, better. Okay. Then I will see that you know. Qa. Prime. Qa. Prime. At least one of them will be sort of greater next year at Qb. Right? And we will have that later on. When we talk about indifference curve. So if 2 combination equally good. and you will map to a same number, right. they could both equal to 1.5, for example, right? Or they could both equal to minus 10, meaning that if we use 0 as sort of like a reference point of something you feel neutral if they have both of them to be minus, meaning that you don’t like them. Okay, like doing homework or cleaning up room cleaning, or something like that. Right. Both are painful, and they will both give you an active utility. Okay. once we have this concept, which we call utility function, that return value for any alternative, right or any context or state of the world. There are many, many element of feature you can build into a utility function, and we can have this concept called indifference curve. right in different code basically represents like a counter match where we control lines that represent specific specific level of utility. Okay, for example, the Youtube level equal to constant are all the scenarios. You are indifferent. or you are neutral, right? Like drinking green tea or something. Right? Or you are. I don’t have a preference, and there will be a set of alternative or set of situations set of states where you don’t have a preference, and that, for example, the constant equal to 0 and utility of this set of alternative equal to 0, and it could be equal to other concepts. Right, we refer to these lines as indifference curve on this indifference curve consumer do not have further preference. Okay. further ordering of the alternatives. So they are equally a pleasant, pleasant, or likable to the, to the consumer, to the choice maker. Okay. this is very similar to the counter line map where we have some visualization. You can see here. This is math, and you can see the numbers right? So this let me try if I can go. You can see the values right? 50, 40, 30, 20, or 10. Those are correspond to our utility. Preference framework. Those are the strengths of the preference. Right? It can be high or low. Okay, and if you draw this vertical line, you can see that certain points at certain points give you similar utility as you have the think, matter of 30, 40, or 50 right? And this is another way of drawing the the in utility framework is the indifferent curve. Right? All those points on this level 20 will give you similar pleasure or or utility. Okay. I was getting a more real one. and we can visualize in the utility of preference complex and visualize these curves, and we call the indifference curve, and on this map, you can see XY, which is Qe and Qb. Remember, they are quantity of apple and quantity of banana. Right? Okay. Qa is quantity of apple. And Qb is quantity of banana. Okay? Sorry. Okay, so please do not direct email me? You can email the I’m sorry. Direct message, me, you can message the whole class. And there are other students who are co-hosts, and they can let you in. Let your friend in. Okay. look at these curves. So the direction that that increase your utility. North, east, right, north, east. Okay. As you move northeast, you will have a higher utility, and that is based on assumption that more banana is at least weekly. Better, right, or no more. Apple is weekly better. If they have no better, I can throw them away. I don’t need to pay the cost. Okay, to clean up things. I just throw them away. So at least more is weakly. Better right? And it to the extent I actually likes a banana apple, or is strictly better, or to the extent I can sell them to someone who like them, apple and banana. So as you move upwards, you will have increasing utility. and you can see that a few points here, e, 1 and and E. 4 is on the same curve, same indifferent curve, meaning that you are indifferent between e. 1 and E 4. Right. So when you move along this indifference curve. So you are not happier or less happier. Okay, you’re just basically the same. You are indifferent. You don’t feel happier or or less happy. Okay, so e, 1 e, 4. So if if I ask you to make a decision between the 2, you will see that I I don’t care either one. Okay. so e, 1 involves 2, but 2 apple and one banana, and E 4 involves one apple and 2 bananas. Okay? And you can see that E 3 is a dominated option. because E. 3 is one apple and one banana, and it’s below. It’s lower than the indifference curve right? It’s strictly dominated. and you can. You can sort of argue. It’s because you can compare e 1 with E. 3. Well, the only difference. You are having same number of bananas. Only difference is you one offer you one more apple, so that is strictly better right, because one apple will offer some marginal utility, or give you some marginal pleasure the E. 3 and same argument follow for E. 4. Right? If you compare the 2,000 e. 3 e. 4, and you. You are actually getting the same number of apple, but one extra banana. Okay? And and you can see, as we move up, you’re having larger quantity of the goods. You’re having more apple and bananas at same time, and you can see E. 2 is strictly is to the northeast, right? And it’s dominating both. Ui, 4. So those those are better options. Okay. so this is the concept of indifference curve. And you can take many different shape depending on consumer preference. Right? So look at here. Here. I think you could argue that consumer have a somewhat of substitution, a preference over upon a banana such that if you, if you take away one of my right, if you take away one of my apple and you have to give me one more banana right? That is the trade that is happening happening between you and E. 4. Right? If you sort of want to take away an apple from me right along. Then I will have one apple. so I’m on this line right. and to keep me on this indifferent curve, you have to give me one actual banana, right? So that this is the movement that that is happening to keep me on the indifferent curve. So there’s some sort of trade going on right internally when to make when you want to make the consumer or the decision maker to be different. Okay? And you see that the indifferent curve is downward, sloping right downward, sloping in this way, and the reason is that if you add more quantity of one product to keep the consumer is different. You have to subtract some other good right? If you don’t subtract, just add on one you just give consumer more banana. She will be strictly better off right happier. So you have to take away something to keep at the consumer on the different curve. And you can also observe something called convex shape. The convex shape basically is this is a concept of convex right? Convexity, basically look like a quadratic function. Right? So you can look at here. It’s convex. And what does convex mean? Actually, here? Well, it actually have empirical meaning. And it is a typical feature of our preference. And it’s based on the law of diminishing return or diminishing marginal utility. Imagine your utility of water right in the beginning. Your utility is very high, you need water, but as as you drink one more at the quantity grows your marginal utility of water decrease? Right? So that’s why I imagine I’m on this curve of indifference. Curve right. You can see that if you take away the last apple I have. If you take away last apple I have, meaning that I’m here right. I’m at you for what Apple? If you take it away, then you have to pay me a lot of bananas. So the curve becomes so bending right? So that is, the convexity come from. So that particular curvature or feature of the different curve actually carry empirical meaning. which is that I want to have a combination of consumption. Right? I want to have a variety. I cannot eat noodles every day. I need to have some time. Have noodles sometimes have rice. If you want to push me to the extreme. give me only Noodles, right? You have to give me a lot of noodles to make me give to make me give up my, the limited amount of rice I have. Right? So that is a feature driven by the utility function. So I really prefer to have a nice mixture of things. Right? I want to sort of on Tiktok. I I watch cat videos. I also watch some funny videos. Right? So I need to have a combination. I do not want to only watch funny videos. I need to have a mixture of things. Okay. and if you have closer substitute and different curve will be fatter. Or why is the case? And you can think about things being very close. Substitute. I think one example would be currency. Let’s see us dollar versus Singapore dollar. Right? So at given point in time, there is an exchange rate that is fixed or given by the market right as given by the market. and I do not really cares about how much. Well, if you consider that fluctuations of this currency, you may have preference. Let’s not just assume that the exchange rate is given, and there’s no uncertainty. There’s no risk preference, for example, and then I do not really have a preference because they have very fixed exchange rate. Right? They are all. They’re all just currencies, right for so I do not have further preference to them, so they will be linear right, and the slope will be determined by the exchange rate. Right? How much I want to trade for Singapore dollar versus us. Usc, right? So they that would be. That will be linearly determined by the by, the exchange rate given the market. But again, you can think about consumers start having risk preference, because this exchange rate may fluctuate right once they have risk preference, they may want to have a mixture of things so sort of to hedge against the risk, right? Like usd getting too expensive or Std get too expensive. I want to sort of hedge over the the risk. Okay. But if there’s no risk you can imagine the slope or indifference curve will be will be linear, determined by market. Okay. a second structure we will impose I will set up to derive. The demand is well, we will have some constraint. Why? Well, we all have config. We have a limited amount time every day we have 24 h right? And we have to sleep and eat. So we have limited time to watch, tick, tock to use twitter or or X, okay, or go to class or study doing homework. Right? We have something called budget constraint right? And we have only 4 years for for college right on average. And we have we have a fixed budget. because every month my mom give me money that that is a fixed number I cannot over spend so I have to. Sum up the price. Times committees. Okay, like, how much noodles I do, how much apples I do. And at the end. They have to be smaller than my budget constraint. Okay, for it could be my monthly income or could be in a company. It could be a budget that given by the management team right? Like your advertisement budget. And if you have advertisement budget, you have to spend across different channels. How much promotion to do on Tiktok? right, and you have to do the calculation, and at the end your your total spending will be smaller than the budget constraint. That is important, because that is the that is very realistic. Imagine you don’t have any constraint. Right? So if you ask me how much banana I want to have. Well, I want to have as many as right. I can sell it even. I cannot use so many bad bananas, I can sell it so my preference is always more right, and my preference is always more so. This is actually, you can think about the greedy element of human nature. Right? We just want more. If I cannot eat, I can sell it. So if I don’t give you any constraint. Well, so that will not give an interior solution. Right? You cannot arrive at a decision. I I can. I can. I can. You know, if I don’t have a time. Constraint, I will, you know, do infinite time on Tiktok and things right? So I have constraint. So that actually drive a decision because of because my resource is bounded. Okay? And I have to do optimization. I will have to choose something like a finance, an interior solution. Right? So. okay? And it’s very simple in our example. But in real, in realistic world it can be very complicated. You can have dynamic budget. You can have soft budget right soft budget. For example, I’m not exactly targeting a number. Right? I’m targeting some sort of soft budget. It that will be more complicated will introduce a lot of incentives, interactions. And you know. for example, in political economics introduce corruptions and things. But we are not. We are not there. So we’ll consider a simple situation where you have a hard budget. Okay, hard budget constraint. So your spending will be smaller. The sum of your spending is smaller than your we call it income just for convenience. Let it be. Why, okay? And this budget constraint will imply that a relation between the 2 quantity right? And we later on, we will prove a point, or we will make a claim that. given this setup, you will always use all the money you have. Okay? Because you know in this setup. It’s a hot budget. If you don’t use the money, the money will be gone, will not be in your pocket in your personal situation probably is more complicated. You are solving dynamic problems. If you don’t use the money today, you can have some saving for tomorrow. Right? You can carry over state, right, carry over time, carry over state of the world. But now we are not that complicated. We’re just seeing the most elegant and simple toy model world, where. if I don’t use the money, use all my income, they will be wasted, and I do want to have more right. Imagine and recall. Our agent has preference and want to have more right the utility function, the shape, and in different curve. So our agent want to have more. So. You can show that. You can argue that, you know. She will spend all the money she has. and so that means the optimal spending eventually will be will make this constraint binding. Okay, you will be exactly on the budget line. I will show you in a moment. Okay, now just assume it’s bonding, and you can rearrange your terms. Right? You can divide the 2 sides by PA. Times. Pb. Sorry. Divide the 2 sides by PA. Okay, and you can rearrange the terms, and you get this line right? So try to recall we are looking at a space of 2 coordinates. QA. And Qb, right? So this will be a lot. This line will define a relation between Qn and Qb, okay. it’s very simple. Right? It’s like middle school math. Okay, so even simpler. Let’s just look at this line, this line, this bold solid line. Here is the budget line. the budget line, by which I mean that you will make my budget binding. I will use all my resource, all my income to consume this these 2 products of apple and the bananas right. And the red region is the feasible region, right? Something I can consume or I can enjoy, because it’s below my budget line, and the white region is beyond my budget because they will involve, they will require income. That is more than I have. Okay, and you can do something called. There’s a jargon for comparative static. And you can. You can think about what if the parameters change? What? What will change your budget line? Look at this line is is a relation. It’s a functional relation between QA. And Qb, right? But the quantity of the 2 items okay. because the price are fixed, all these wire pap are fixed. They are numbers given by the market given by my mom. Okay, my wife, for example. And Papb is determined by the store owner, who’s selling these fruits right, and that that those when people make their decisions, I will be facing this line. and I can choose either below this line or on this line. I cannot choose a point outside this line, the white region. Okay, so this is simple. So next, what if my mother give me more money. Now, why increase? Right? Okay, my mom, give me more. Why more income? And then what happened to this budget line? I can’t afford more. I can’t afford more. Right? So all this right region will still be affordable to me. And I can, even, you know, spend a little bit more to get more apple and bananas. Okay, so that is what happens here. And look at the relation between a white line and the blue line. So increased income will push your budget line upward. Right? It allows a larger, feasible region to for your consumption. Okay, so that is increasing. Why and what happens if your income decrease and you can reason corresponding. What happens if the price change? Well, if the price change, let’s see if there’s a change in price of bananas. What happened to my budget line on my Facebook consumption. Okay. my claim is that you will become it will become this green, this green line. Okay, that is my claim. You see. Why? How do I make it disappear? Okay, okay, just to see why will look at this budget line. So this point. Well, it’s linear, right, because it’s a linear relation between the 2 quantities. And this point is determined by what this point determined by my income and the the price of apple right? Because at this point I decide to not buy any banana. right? So I don’t care about price of banana. So this point is fixed. Okay. this point is fixed because I don’t care. There’s increase of banana price, because I’m using all my income to buy apple. That is this point, right? Which means that when there is an increase in banana price my budget line will still stay on this point. Okay, but my budget line for inter intersect with Qb. Will be different because price of banana increase right? So the amount of banana I can afford. If I spend all my money on banana will move to the left. So this port. So this point. when I spend all my money on banana, will move to the left, to to here, right, because price of banana increase. And this point doesn’t move, because at this point I’m spending all my money on Apple. I don’t care about another price. That’s why the budget line is moving from the right to the green line. and you can do this exercise. And to do this called comparative statistics. When the world, the state of the world, changing like prices, changing what happened to my budget line? Okay? When my mother, feeling happier, give me more money than what happened to my budget line. Okay? Hanged up. Okay? And recall that. Recall that if we move to the northeast, move up right? And we are having enjoying a larger quantity of apple banana, and we are happier. So that is the direction of increasing utility. And on this slide our model of choice will conclude, okay, we’ll come up with a solution, with a prediction of choice of an individual consumer’s choice. Remember, we already have a thing called preference, which is a function or assigned value to choice, to alternatives, right? A set of choice, a set of alternatives. and I have something called budget constraint, which means I don’t have infinite resource. I have a finite resource. And and I need to allocate to maximize my pleasure. Right? Okay? So now we arrive at a prediction which is called a choice. Okay? So we will use our indifference curve to argue that this QA. Star and Qb. Star will be your optimal choice, your optimal consumption in this scenario. Okay? And so we can draw this in different curve and draw a lot right? So there will be downward, sloping, somewhat convex and bending in this way, right? And we can sort of, and we already established. If you move to the move up to the northeast, your pleasure, your utility will increase right? So what I’m going to do is I’m going to start from very low and then start moving upwards bye, start from very low indifference, curve and then start, move upwards and and see at which point I can do no better than I’m doing right. I can. I can find out something called optimal solution or my optimal choice. My optimal choice will be a prediction that and maximize my benefit. I cannot do any better than that. Okay, so let’s do it step by step, and see how we get there. So let’s start from a very humble deep, indifferent curve. Well, I’m not taking a lot right? I’m just taking as something I’m definitely affordable. So that is the very light a blue line. So the 1st line here the lowest line. So the claim is that there is a segment on this light blue line that I’m I can afford, which is from here to here. Right? So this segment. It’s affordable to me, right? Because it’s below the red line, which is my budget line. Right? Given the money, I have. Okay, I can easily choose any anything on this in different curve. Okay, right? But I cannot. I actually cannot choose this one. Right? And I cannot choose this point. Okay, but I can choose anything on this the indifference curve. I can choose some sort of combination of the consumptions along this segment, but this is not the end, because I know that I can do better by moving my different curve up. For example, this point, this point in the empty space white space is affordable to me right? Because, remember our logic, this point is below the red line, which is my budget line, which means that I have enough money to afford that. Okay, meaning that I can move up further. Right? I can move the curve up further. Or I can choose this point. Actually, there will be a region. I can draw the indictator for this point. And they’re actually a large set of consumption that is available to me right at this level of utility, this level of pleasure. I cannot only choose this point. and I can also choose other points along this indifference curve as long as they are under my budget line. meaning that it’s affordable to me. Right? So you can. You can use this argument and move the your indifference. Curve up, up, up continuously right until you reach your point here until you reach a point where you, your indifferent curve at only one intersection with the budget line, and the claim is that that point will be your consumption, the optimal consumption. optimal choice of QA. And Qb. Number of apple and bananas. Right there. They can be very sophisticated and rigorous, solid proof using topology or real analysis. We don’t do that. Okay, we just use graphs. But you can get the idea that right? So in this toy model world, we can derive a solution of choice. We can predict that what this agent or this consumer, this student. whatever this manager, will do right. So so this point will be our prediction. The point is defined as intersection of in the only one intersection of the indifferent curve, and the budget line. We have not conclude the proof. Okay, I just, I just showed that or argued that for any literature lower than this line. Okay, lower than this indiffer curve. It’s not as good right? Because I already argue that this point is affordable right? This point is affordable. Okay? And I still have to argue that I cannot do better than that right to do better than that, that I have to move upward even further. I cannot do that, because if I move up a little bit further I will be detached to my budget line, meaning that all the points on Difference curve are not affordable to me right. So this conclude the proof that I cannot do any better. So this is my choice. Okay, the argument goes the following, I start from very humble allocation right, and try to move up, increase my preference level. basically playing with indifference. Curve right? I move up a little bit a little bit. I’m still happy and feasible. I have tons of choice on the indifference curve until I reach your point. There’s only 1 point, and that is my choice, and I cannot move any further, because if I move further up I will. You know I don’t have enough money. Okay, so this is the process of choice in our toy model world. Again, we have set up something called a preference, meaning that I have. I have preference over the events or the arrangement right, or the set of States, the set of alternatives. Right? I have some ordering of these things right? I like something I don’t like something I like something more right when I said this, I’m meaning that a number, a real number, right? A number measure the strength of my preference or my taste, right or something, utility to me. Right, given that we can define in different curve, and we have a a agent or consumer or a manager that has a budget constraint. and that sort of imply. If we make the reasoning and we can derive a final solution which is called a choice. Okay. let me clean it up and move on. Okay, now we have choice. How do I cleaned up? Okay. So there is a proof behind this graph that saying that is that says that. How do we derive? How do we identify the optimal choice? And we can make stronger claim or predictions? What what if the world is changing. And what happens to that decision? Right? Okay? So we visualize, we visualize the decision process by drawing the the budget line in defense curve, and we find the tangent points. And now, what what if there’s a change in price? Well, we already learned that if there’s a change of price of bananas. What happened to? Well, well, that will not affect my preference right in this toy model world. But you realize you can think of a scenario where the price change my preference, change like my preference for Lv. Right for some luxury product. If you change the price, I change my preference. But now we are playing with a simple setting where my preference is primitive, it’s a fundamental in this world will not change. Okay, my preference will not change. It’s given structure. But if you change the price, okay, so my indifferent curve will be the same same set of indifference. Curve right. You change the price. Well, that only affect my budget line, and it will, showing that your budget line will rotate. And this point will now change, and this point will shift to the left, because price of banana increased. Right? Okay? So this. okay, I guess the right line is the default. And and okay depends. Because here, says an increase in pb, so the the green line is the default. Am I having some message here? Sorry. Okay, no. I’m good. Okay, right? So an increase in price will shift, will rotate. Actually, in this case, rotate my budget line from the green to the red. I’m not going to repeat the logic, work it out by yourself. Okay? And you can even play with it by arguing. What happens when income increase? Right? What what happens when income increase? What happens to the optimus consumption. We’ll just repeat the exercise and logic and find the optimal points. So here the optimal points. we’ll move from this point to this point. Right? Okay? And this is the income effect. I would escape because it’s not that complicated. And the consumer can choose the combination on the budget set that hates the highest indifferent curve and optimal choice is on the budget line. I have made some arguments to support this claim, which is that in our setting. The the consumer likes the product right in different curve, going upwards. If you give the, if you give the consumer the agent more, she will be happier. So given the budget and given assumption the setup that the money on the table is with this, so she will spend all the money right, because, imagine you still have 1 to purchase the product. You can easily do better by purchasing more product, a right, more apple. But apple is not discrete. So if you play with the integer problem, that discrete problem. It will be more complicated here for simplicity. I would just assume we’re in a world that quantity is continuous. Okay. the number of apple consume depends on. We already play with the change of the state of the world like price, like the money my mom gave me. So that means that you know, we already show that optimal solution or the choice will depend on context, right? The choice depends on price in the market. It depends on my resource. Okay? And in general, you know, this is a very simple model. I hope it gives you the, you know, really the fundamentals of the world, but you can easily add more element to it. You can add the ringing day right? You can add a interaction. Right? You can add more complicated preference function. You can add state dependency. You can edit stochastic States or world randomly through model. But this is the simplest and most elegant model that gives you the insight of how we make choice. Why, there is a choice that is a finite choice. Okay, there is a finite, a choice that is not infinite, but it’s certain number, and this is the structure or the setup we will use to derive a individual choice. And after we have individual choice. We are not done yet. Well. imagine you are a firm right. What you cares about not. You may not care that much about individual consumer. You cares about your sales, your everyday sales. And we need to do aggregation over all the consumers. Okay? And if you are, do, if you are like Tiktok video platform. You may cares about an individual consumer taste over videos. Right? Okay? And you can measure that. But here, let’s say you are aggregate. You have to aggregate, and you cares about the aggregate measure. And if you are, government also cares about this aggregation, a total quantity when you aggregate over individuals because we typically many policy content also don’t target individual person. Okay. But you know, even, let’s say, in in a more individualized context, you may still cares about a type of person like all the 30 year years old male, for example, right? Or in us year 2, undergrad, something like that. Okay, the type of people may have some share similarity in in preference. You can do aggregation. You can derive a demand curve for a given type of people. You can do the aggregation conditional? Some observable attributes? Right? Okay. and demand curve, because we already know an individual consumer, individual agent. How to make? How does she make the choice. And we can do aggregation over this individual. So we get an aggregate quantity which we call demand. Okay? Or in a business contact. You can think of it as sales. Right? Total sales on my product. You can think about it as a schedule of It’s a schedule. Okay? And the slope will depend on the curvature that will show you demand curve. But I think most of you know what is demand. Curve right? You may say it on newspaper. Yeah. So it’s very, most fundamental concept in economics. I will show you next slides. It has curvature, and that curvature in our class is a theoretical concept you follow, you know, in some functional form, right? I will not tell you. Teach you how to get the functional form, and that is an empirical question. That is a measurement or estimation problem that you can learn in some other class that teaches you big data. Okay? And in that kind of a class, you will treat consumer preference and demand curve as an empirical object, and you will recover that from data okay, that is called demand estimation. Okay, some of the economists are doing that, as in Amazon, for example. Okay, they try to estimate preference from the data by observing how people make choice, and they treat it as empirical problem and try to recover the preference, which is some latent functional form that we do not observe. But we can ask me from observing, for consumer make decisions observing their choice. But that process is called revealed preference. Okay, that process that is not in this class. Okay, that is in some other class that teach you how to play with data. That concept called review preference is by using choice data using behavioral data to recover preference. Because your profit, I observed. Let’s see. this set of students choose a Us. Versus ntu, and I can ask me their preference. Right? Observe, they make the choice. and I can sort of estimate how much better they feel about Ntu. They feel about nus versus ntu. Okay. So by observing, you make the choice, and I can inference. I can identify your preference. That is a different treatment or technical problem, and I will not if you need reference, and I can show you, but that is not central to this particular class in this particular class I will use. I’ll do it more theoretical way or analytical way. and I will use this contract to derive predictions and to explain how things works. Okay and demand curve normally downward sloping. And you can find counter examples. Okay, I think in the next lecture I may give you some example, and you can also think about example yourself, but mostly the demand curve by default. Most of the demand curve is downward. Sloping. A demand curve is a function. right? Or it’s a relation between quantity and the price. Okay? So giving a price demand function, return quantity, a sale right? So your manager at Amazon will will ask you what is optimal price? And well, then. when asked that question. I know that, or I expect the problem to be of such structure that give you a price, and there will be a quantity right, and the profit will be quantity times, price minus some cost. Right? Okay? And I’m interested in the functional relation between quantity and price that you can ask me. But the theoretically speaking, they will be downward, sloping for most of the product. Because just think about your your ipad. If apple increase the price of ipad, what do you do? Well on average consumer will purchase less right, and I will show the demand curve because there will always be a marketer consumer, that is, who we find who finds apple or who find ipads no longer attractive. If you increase price and we will show that repeatedly in in later lectures. Okay. and demand curve. Have price on on the vertical X, as inverse demand. Curve is, what would you be willing to pay for one product? 2 product 3rd product, a 3rd apple. Okay? So that is well, that is actually an individual level demand curve. Okay? But if you add to an aggregation you’ll have market level demand curve. Okay, we will end in 10 min. I try to finish. I give you some extra time, so I will finish today’s lecture in 10 min, and if I cannot finish the content, I will leave to the next lecture. Okay, and a lot of demand basis, as the quantity demanded falls or decrease as price price. So this is like a conventional wisdom, or is some empirical pattern, that is, that holds in most of the context. But again, it’s not always okay and demand curve will depend on availability of substitute. What does that mean? Well, it means that demand for apple product like iphone, which depends on availability. If it’s a new version of Samsung, right? Because Samsung cell phone is a a substitute right? And even you know, the demand for iphone 14 will depend on the price or the attractiveness or the design of iphone 15. They are also actually substitute. Even they owned by the same firm. So in that sort of context, from firm perspective, you will face pricing on multi product that are complements or substitutes. Okay, okay, I’m a little bit ahead of me, and we will talk about this concept in the next lecture and demand will be affected by budget right? That is like general economic condition, like people’s average income that will affect that will affect the demand. And we already argue that at individual consumer level. Okay. alright. Let me just move on. I can give you some demand curve, for example. So this on the left, you see a linear demand curve. Well, the this is a straight line, right as you as you decrease price. Right? You move downward and you will be able to sell a larger quantity. Right? Your decrease price, you’ll be able to sell larger quantity. and on the right you see something called constant elasticity demand. Again, I have not defined the term called elasticity. So this will come in the next lecture. It’s a little bit counterintuitive, because it says constant elasticity, but it’s not constant. It actually looks very has a huge curvature here. Okay, we’ll define elasticity in the next lecture. Okay, the demand function depends on a lot of factors depends on prices and availability of substitutes or complements depends on the population. We are talking about their age, gender and their income. It depends on advertising. Advertising may provide information of the product or may change, actually change the preference. That’s also possible. And a lot of other things right? Because when we define the utility function of preference. We made a point that the utility function can be very, very sophisticated. I think it depends on a ton of stuff. Okay, but we. We do not write down everything. But potentially, if your situation requires, you can write down a lot of other factors into your preference for demand function. and it’s important to distinguish between movement along the demand curve which is changing price or shifts of demand, curve upward or downward. I think I will come back to this point again in future lectures. Okay, here I have an example, Coke, or passing 2 drinks and your exchange rate, or a different curve between the 2 products. And I think this is not very difficult. And basically e 1 and the E 2 are on the same indifference curve right? And so mid, which means that. oh, sorry. This is demand, curve. Okay. So e, 1 to E, 2, okay, e, 1 to e, 2 is increasing the price of coke. Okay, so this is making the point here. The shift along the demand curve versus the shift of the demand curve itself from e 1 to E. 2. And this is increasing the price of coke right? So that is a movement along the demand curve. But if they’re decreasing price of competitor or substitute right and demand curve will shift the price of right. So the demand curve will shift. Okay. this is the price and quantity right for Coke. So this is a demand curve. Okay. I will briefly talk about this, and then I will leave the most important content to the next lecture. So why do we do this much of a modeling setup? Well, it will lead to very a lot of analysis when we also set up the supply setup. the firm side of the model, the problem, and we will have something called the equilibrium, and we will be able to analyze the total welfare or goodness, right goodness of a system, goodness of condition. But now, even if we only finish the demand model, we’re able to come up with this this concept of consumer surplus which measure the welfare of the demand side. How happy consumers are. And if you have the demand curve, okay, and you’re already able to compute the consumer surplus at given market price. And I will leave that to the next lecture. Okay? So I think I will. Stop here, I think there’s question on for those who are not in Lt, how do they have candy? You have to participate in cover discussion to get 5%. Okay? So I think I answered this question, there won’t. There won’t be much variation in the participating score. It’s a reward based, and the variation will be very small. It only matters when you are on the margin. and that. yeah, okay. And yes, if you participate in canvas discussion, you will. You will also count and get reward. Yeah, but I think most of the condition. It won’t matter much. Will there be group work in this module? no, not explicitly, because assignment is individual assignment. But you can work with your classmate. But just make sure the right app is your own. Okay? And you have to submit your own answers, okay, your own answer. But I cannot control you. You can discuss with your classmate you call your teammates. That’s okay. But at the end of the day you have to submit your own assignments. Okay? And you have. And you have to write it up by yourself. Okay? And that is helpful for your exams as well. Can I upload all the slides? I I may still make some changes to some side. I do have future slides. I? Okay, I’ll try to upload a few of them. Okay, but keep in mind that. I will still update and make changes to the slides. Okay, you will refer to the final version of the slides. Okay. okay. So I think that is for for today forget it. I will stop the recording. Okay? So I think I answer all the answers in zoom. if you feel I still miss anything. So yeah, you can. Please let me know. and I appreciate, if you can repeat your question, if, in case I miss any of them. But I sorry would it be possible for all slides. Oh, I think I answered, this right.

2. Elasticity

today is the second lecture for our class. That’s 2, 2, 3, 8, which is economics of it. And AI, let me repeat some of the admin and cost requirements we discussed in previous lecture. We will have a large team of ta, I added. Quite a few we have 3 50 students from information system analytics and Cs and Ecom will help us on the course of work answering our questions. We also have some very excellent undergrads who were from this class previously. Did very, very good job, and I hope they can also assist our students this semester to you know, I understand some difficult part of the class, and we also share the experience. Of taking this class. So I hope we can help help you guys in case you need any help. And I have seen some. I have seen some questions on discussion Forum which I really like. I think this is great. I really appreciate, if you can. You know post your questions on the discussion forum to share with your classmates, and also appreciate those who helped, you know. Answer this question. I have worked out a procedure for our Ta to assist you in the learning and in the discussion process. So my procedure, my proposal, was for them to wait a little bit to 1st to allow peer classmates to jump in to help or to discuss. So typically we will have. We will have one teaching assistant each of the working day. 5 days of the week, and each one of them will help answer questions. But typically I recommend them to wait for like, let’s say, half day, 12 h to give more time for other classmates to think about the question, or to to jump in and discuss before they will, if the question not addressed, and I will ask them to help with the the question. So discussions are really helpful. So I I encourage you to ask or to engage more in this part of the class. and I also get questions regarding the assessment. One thing I want to clarify is that we will not have group group projects that this year we, our assignments are there will be 3 assignments all of them are individual assignment, which means that you have to submit your individual answers to campus. Okay? And I already, actually, I already up uploaded assignment one, as you can see. They are all Mcq’s okay. And also, as I mentioned in previous lecture, they will be complementary to our lecture content. Some, maybe, you know, explore alternatives or other models that we do not discuss or explore extensively in the lecture. Some will help you understand the concepts and method or theory you learn in class. Okay? And it it’s set in the assignment that you know. You will also be required to upload a separate Pdf or work file, or any file that it sort of documents your the to showcase you actually finish or work on that assignment independently. Okay. so this is for the assignment one which is due on February 14.th So you have a little bit more than I think. 2 weeks to work on the 1st assignment. okay. A midterm is, I think is the week, the midterm week which is the week 7, the week after recess week. and we are still in the process of reserving a venue for the midterm. I will update you guys once I heard back from once I hear back from the ut office. I’ll get you. And the exams, as I mentioned, will fall. Follow similar style as the assignment you see, and basically, they will be Mcq’s okay. And on canvas you can see a timeline for the class. And you can also find a a reference book which helps you if you need to learn more about the topic, and you know more grab the framework. And this is a standard textbook. Okay. what else? Let me check assignment lectures. Okay? One more thing in the last lecture, I think some of you on Zoom asked for asked for, you know. like all the lecture slides which I uploaded past year lecture slides here, if you really want to, you know, learn in advance, and you can take a look. and there’s also a set of slide which I call optional they are adjacent to what you learn. Some are deeper, some are, you know, if you you want to explore a bit more, and those are the options. And for all the slide you can find, actually find the corresponding part and in in the textbook, okay. okay, okay. And for our class, we have a recording. And you can find the lecture recording here. And I will post on the recording on this page, and you can view it. And you can also download the recording. I assume you can download if you cannot. Okay, let me know. Okay, And from today’s lecture, actually, we’ll try the AI function of zoom and see how it works. Betty, see what’s in the chat. Yeah. So there is a question on- on the for a group project. We don’t have a group project. Okay? So there will be 3 assignments each account for 10%. Okay, let me see. Yeah, okay, I think that is for the admin. Now, I will move on to our lecture content. So recall in the 1st lecture we were discussing a concept called Preference. and based on which we assume there exists a utility function, right? A function which is a mapping from alternatives, options, or states or situations. Context, right products. I choose some continuous numbers like 1, 2 5, or could be minus as well, which we call utility. and we argue that this preference is important. We will build all our demand, and later we build supply all from here, from some consumer that has a preference. things they like and things they don’t like. Okay. And we actually have a quantitative measure how much they like about a particular product policy, design, technology. weather, country, anything. Okay? And we will derive a concept called demand demand function from individual consumer choice. And later we will have a supply set. that is, you know, reacting to the demand function or letting try to take in demand into account. And after we introduce supply side. We actually have both demand and supply. We have a market, a model of market and how it functions. Okay? And last lecture, we start from the concept of preference and utility function. We describe a framework called Indifference Curve, which captures the level of utility consumer derived from a bundle of products, right, a pair of apple and bananas. And in reality we discussed, utility can be very complicated, can define, be defined on very large complex space of bundles of products. Okay? And the important thing is, we did we talk about some theoretical property of the indifference curve, shape or indifference curve. And we set up a choice contact where preference, under some constraint which we call budget constraint, right, will lead to a choice, lead to a final decision. Okay? And if you aggregate the decision of individual consumer, you will get something called demand, which is aggregation across many, many consumers right in a geographic market or in a society in a state, in a country, for example, right? So that is the idea of demand is based on aggregation over many, many consumers in some model. Those consumer can be homogeneous, or which means they can be similar, or one representative consumer, or they could be something we call heterogeneous. They could be different from each other. You can have different taste. the some like noodles, some like rice, right, some like leisure, some like more like like working. They can have different preferences, and you aggregate over their individual choices. You will get something called aggregate demand function. Okay? So we are. I think we discussed about the demand curve we talk about demand function. What it depends on. I think we already discussed about this graph. A shift along the demand curve versus a shape, a shift of the demand curve. Okay? So the most important concept we want to derive here is consumer surplus. and the idea of consumer surplus is a to measure. How happy consumer are, you know. You know the welfare you may call it welfare right, or the happiness amount of pleasure that we aim to quantify, and from policy maker perspective, you cares about the aggregation of welfare like aggregately, we see, like these countries, people live very well, they are very happy. What do we mean? Well, we actually, in a very precise definition that they define based on choice. Okay, that means that this, this society has higher consumer surplus plus something we will introduce later, when we introduce the supply side called a producer surplus, a supply side, a surplus the surplus of firms which is profit. Here we, we 1st introduce a demand side consumer surplus. We can intuitively think think of it as profit to consumers. Okay, what do we mean? Well. because consumers are making decision? Right? So you observe in the market that what we call posting posted pricing mechanism. Where you go to a supermarket, you see a posted price, and you make an active choice. If you like the product, you buy the product right. You buy the product, which means you will pay the posted price as a 10. That’s why no one force you and you make the decision to buy the product right? By observing that you buy these products, I can learn that you review preference preference, reviewed from your action, suggests that you like this product more than 10 product right? And based on that, you can a lot of consumer in your market. You observe them making that choice right? And you treat this problem as empirical work, and you can ask me their preference from the data. And that process demand estimation. Okay? And after you do the demand estimation, you are able to have the demand curve. Okay? And then you can do very quantitative analysis and study empirically the impact of policy. Right? For example, you can estimate, if you’re open. AI, you can estimate your consumers willing to pay, or their preference for you know, for the subscription. and after you empirically estimate the curve, the distribution of Willings. You pay right, and you can. You can treat it as a known function with all the parameters. Even the shape of demand curve is known, and you can go on and and study how to set optimal price, and that that part, the theory part will be covered. In the following lecture we introduce a supply set actually, in some of assignment, you you’ll be asked to think about optimal pricing problems. Okay? so that is to answer some of the question. I will use some classic examples to visualize. What do we mean by consumer surplus? Imagine a very simple contact. You are drinking water, or you are eating or buying pizza. You’ll have valuations or utility over pizza right? But your utility, your preference, is not constant right your valuation for pizza. Suppose you’re not thinking about selling the pizza just, you know you have to finish. You have to eat all the pizza. Then you know your marginal utility of pizza. It will be decreasing in many context, like similar to drinking water. Right as you drink more water, your marginal utility of the water will decrease. You will feel very full right in this context of pizza. Suppose your the 1st slice 1st slice of the pizza worth 3 to you. and you, after eating the 1st slice you will feel like less hungry and marginally you’re willing to pay for the next one will be lower. I think this is probably 1.5 or something. Right? Some you have a lower willing to pay. That is the marginal utility decreasing. Okay? And if I I sell you. I try to sell you a 3rd slice, and you’re willing to pay even lower. If, for example, you can. You can only eat 2 slice right. The 3rd one is like. you can see as showing this graph that you’re willing to pay is very low is 0 point 2. Right? Is that like, I’m 80% full, right? I don’t want to eat an actual slice of pizza anymore. This is very simple example. Well, now, think about if there is a price for pizza. Okay? The restaurant is setting a price for pizza. In a later lecture, we will set up a model that describe the process of pricing how the supply side set the price. Okay. But now just the prices are given given to consumers. We are not concerned about where this price come from, the price will be determined in our next lecture. Okay. Now. given this price of 2, this blue region. Right? It’s a difference between my winnings. You pay and the price actually pay, which is 2 here, and if I buy a second slice well my ways, you pay decrease by still above the price, so I will purchase 2 slice right, and I will still earn a little bit of consumer surplus. But you can. You can argue that in this example, suppose Price is at 2. So if I’m I don’t want to spend 2 to me. Right? So I will have negative surplus. So this is the, you know, the process of a decision. It’s very classic setup economics. When we think about this marginal benefit, marginal return, so I will keep increasing my consumption. But remember my marginal utility decreasing so. and to your point a decrease to the level of the price, and I will not continue expanding my consumption anymore, because if I purchase, I consume even more, and I will be worse off. Okay, so this is the very simple example. But you can think about how price is determined in this model. Right? So this is a uniform price. If you think about restaurant pricing based on quantity right? As well like I can. Price. So think about if you are a restaurant. How do you set Price? Well, I can set 4, 2 slice. Will the consumer buy it? Go ahead, Andy. The answer is, yes, okay. But you can charge like slightly below 4 for 2 slice right? And then I’m extracting more consumer surplus right? As a firm right. I can charge a higher price. If I’m willing to sort of bundle I can. I can sell 2 slides at a time, right? 2 slices at at a time. Okay, so you can realize that there’s very rich space of pricing strategy the firm can can utilize to extract consumer surplus. because from a firm from managerial perspective. Consumer surplus meaning means the money on the table. Right means money on the table. You can always increase your price a little bit more. and to get in a higher profit right? Because your your marginal cost will be the same. By serving one slice of pizza you can increase your price a little bit more, and extract a little bit of consumer surprise and translate that into your profit. But from a government perspective, what what does this mean? Well, from government perspective, this practice. It’s purely redistributional, right? If it doesn’t affect the the amount of trade we’ll talk about 1st welfare theorem later, in later lecture, more holistic view of the social welfare, but from government perspective. If the firm is earning a little bit more money. But that is coming from consumer surplus, then it’s what we call redistributional. It’s some redistributional surplus from consumer to the firm. Okay? And you will be asked a lot in in assignment and in later lectures to think about the welfare, social welfare, implication, consumer welfare, implication, and the implication to firms profit. Okay. but this is a very simple example, and consumer surplus is the blue region above the market price. P. Star. Again, I haven’t specified how P. Star. The equilibrium price is determined. It will be determined by cost and competition that we specify in later lecture. Okay, and below the price. And you can see that this part, P. Star times, Q. Star, is the expenditure. Right? That is the money that goes to that goes through the supplier right? Because supplier will be selling Q. Star, which is a quantity, and making a money of P. Star. So this expenditure is actually a transfer of money or profit from consumer to the firms to sellers. Okay, we will be very specific, and we will discuss much deeper in our later lecture, and we’ll also will. I will tell you that. And there’s also a factor called cost right, because not all the expenditure are welfare, because some are cost. I need to spend electricity or spend some resource money to produce this product right? So, and if you take the whole region minus the cost at the end, you will get social welfare. Okay? But that is in the later lecture. So this graph, what it shows is that it’s a smoother graph of the demand curve where you have P. Price as a vertical axis, and you have quantity as the X, and you can think about aggregation over many, many consumers, right? And you can think about here. Of of course, we are talking about a continuous function as an approximation or theoretical model where quantity is continuous. Right? That’s why this graph looks continuous. But in our toy model example, like Pisa, perhaps, is not this continuous and price? Well, you like the price? Is not not that continuous? Because because it’s 1 cent, right? It’s not like a 0 point 0 1 cent. So essentially. But the continuous model is being thought of as approximation. You can think about. You can. Sorry. Let me exit you can think about. There are many, many. Consumer right has value of has valuation of peace, and you aggregate over all the consumers. Individual decision problem. They have their preference. They have their budget constraint right? And you aggregate across many, many consumer. Let’s say. And you as students, and you get a demand curve for us students. Right? Okay? But I think we can expand it little bit. And this works smaller. Okay? And you can compare the amount of consumer surplus in different market for water. Somehow, price is very low, right? And that is actually a classic theory. Economics. That’s because there’s a lot of water, and the supply is very, very large, and that drives down the price right? Think about that. Many people or firm can sell water right? It’s not a capacity constraint. so that competition may drive down the price of the water to be very low, and consumer surplus is very high for for water and relative to the consumer surplus. Well, the firm is not able to extract a lot of surplus, a profit right, and that is because the price is being very low and close to the cost marginal cost of of water, and we’ll talk about in our next lecture when we talk about a supply side model, and as a comparison for diamond, the price is very, very high. that is because supply is very limited. Is capacity, constraint in terms of the amount of diamond we can produce. That is constrained. You can think about a bitcoin. Right? Price is very high. Consumer of diamond has to pay a lot. Okay, and amount of surplus is very small relative to the expenditure. Okay. I will conclude our discussion on the demand demand side of the market Demand Center market will be half of the whole market. I don’t literally means half, but it’s, you know conceptually, is one side of the market, and we we define everything from a concept called preference customer, like something don’t like something like like apple, more than bananas. and they have some constraint like finance resources, and they would, will derive a choice and aggregate them. We get demand and get demand, curve, demand, function, and that concept also allow us to study how happy consumer are which is measured by consumer surplus. Well, and in the current context, a very interesting question people are asking and thinking about is that as we have larger data and better models that can learn and predict consumer preference. We we start seeing situations where models actually are able to understand very well what consumer wants. Okay? And the point here we want to make is that it’s unclear how, whether or not it’s good or bad, it depends on a set of conditions. It depends, for example, competition and incentives we’ll talk about. You can think about, you know, like Tiktok or Youtube, right? Who have data on your prior consumption, not only your consumption, but also other consumer consumption, right? And by, you know, model to learn the the behavior pattern of consumers. they can uncover some very deep insights about a consumer that the consumer may not know by herself. Okay. It happens quite often, for example, to me that I will say some very interesting video in the recommendation. I actually don’t know their existence right? So that has welfare implications. I go to Amazon. I, I purchase a product. Sometimes there’s frequently bought together right, or some other consumer who view this also view that buy this also, like that kind of recommendation that is very smart in predicting or understanding what I like. Right? So what is the implications to to this? Well. it has some good good good effect. Well, in lower lower search cost? Right? So if the algorithm very smart, right? Imagine you’re buying on Amazon, you’re trying to search for some products right the well, the the recommendation and search ranking can help you. Right? The algorithm. Yeah, they are very smart. They help you. They understand you very well, right? So that may save your time right? And sometimes they can help you identify options that you didn’t realize without. If you you if you don’t have access to the technology, the search technology or recommendation technology. Okay, so that is one aspect of the recommendation, personalized recommendation, it reduce the information friction in the market. Right? Because, consumer does not have, like a all the information of all the options they can, they can that exist that is available to them. Right? So that’s why, in the early days of Internet, when I think some of you are not even born, I guess When we we have e-commerce in like more than 2 decades ago, there is predictions that price will converge to like a single price. Right? It’s called one price. Okay. Law of one price, right? If the product is the same, right? So later we’ll talk about supply side, because your competition, and if the market. If the market is frictionless, there’s no friction at all. How could there be 2 price for the same product? Right? I’m thinking about? I’m talking about exactly the same product, like same brand, right and same quality. There’s no quality, uncertainty, and like same shipping fee or same, you know, delivery time. Everything is the same right? But you could argue that there’s always marginal differentiations of the product. Right? But let’s have a thought experiment. Suppose they are exactly the same identical. All dimensions. Right? Will there be multiple price like one charge for 8? Okay, I can tell you that in the early days, like 2 decades ago, economists are predicting that there will be only one price. But there’s many, many empirical evidence suggesting that it didn’t happen. Okay, you can still find price that are very different for very similar products. Even, you know, nearly identical or identical. You, you agree, if they exist, but identical products. And this is called price dispersion. Okay, price dispersion price, you know, even for same products. Right? So the price dispersion cannot be explained totally by you know, either quality or other factors. So that is that suggests there’s still friction in the early Internet period. But it’s still there’s still friction up to today. Right? You can see a shopping comparison site. Try to help you compare the product right? Because there’s another side of story from the supply side. Actually, the sellers are trying to do something we call obfuscation, which is, they try to increase the search cost right? Imagine our website on shopping. Sometimes they. They try to make their product appear different, right? Even if it’s the same product, you know. Seller may have a different dimension of of the of the product consumer, not certain if they are the same or not, and all these of frustrations they try to increase the search cost to consumers and to to increase the differentiation and to earn more profit. Why? Because if if the product is I indeed identical to consumers, right? And let’s say, on Shopee. How could there be 2 price? Okay? So there’s still there’s other series called Consideration Set, which is very closely related to search cost consideration set says I have. I’m a consumer. I have limited time, I have limited bandwidth. I cannot consider all the possibilities or the product, so the consideration set is sometimes can be very small. Okay? So those sort of imperfection, those are all you can call frictions market friction. A lot sort of can sustain a price dispersion for the same product. But that is hard to explain, using very classic, nearly perfect models. Okay? And the better prediction technology they can understand your need, and they can read your mind. They can, you know, based on your behavior patterns, your some demographic information or some other insight they may have from you. From all sorts of data. They can lower your search cost. They can expand your choice set choice sets. Okay? You can think about in our model. Right? Let’s assume there’s no friction like limited attention. For example, limited attention can predict that. You know. I, even you, give me a very large choice set right. I would not pay full attention to all of them. I would just look at a few of them. I’m tired. I will make a decision. Right? Okay, suppose there’s no this computational cost for consumers. Mental cost. Okay, you expand the choice sets. Well, consumer have utility over all the alternatives right to take a maximum over the alternative. If the like choice is larger, one is strictly larger than the other. Well, larger choice that will give you higher high higher quality decision. Right? Because you are considering or you’re maximizing over a larger set. Right? So that said will give you a larger maximum. Okay? So that is sort of efficiency again, from better technology and understanding consumer preference. So they will sort of be help consumer, like our assistance or collaborator, right or agent, whatever you call it, help consumer to make decision, make better decision. But the the other side, where, when the the technology is not fully incentive aligned with the consumer. And this could happen, for example, if the technology is more leveraged and exploited by the supply side, what will happen. Well, there could be a possibility of the perfect price discrimination. Where I think you will see this kind of example in assignment question from can charge individualized price to consumers, and that will allow firm to attract more consumer surplus. Right? The pizza restaurant owner can charge different price if they can read your mind, and they can understand how much you want to pay for pizza. Right, read your mind and charge your price, so what will happen? So she will give you an offer that you’re just being indifferent between taking or leave it right to give you. The the seller Amazon. They can understand your winnings. You pay for this back handbag right? And it will give you. Let’s see, Amazon can read. Your mind knows that you want to pay 8.5 5. Right? So by doing this right? So you can push and you track all the consumer surplus. Right? Okay. And the second one, you will see more example later on when we talk about supply side. And this will actually also interact with competition. Okay, what about? There are 2 Amazon 2 platform. Both can break my mind. Would that be good or bad for me because they will compete on price. Right? Both of them know that. They have this technology, right? Okay? And we also talk about in the second half of the class about a concept of principal agent problem where there’s vertical incentive conflict. I’ve for example, when Platform gave you a recommendation. Well. it does does not. It will not fully internalize the benefit right? You can think about the platform as an agent. And you are principle. Okay? And you will give recommendation, because the agent who, making a decision, or who is very smart, may have their own private incentive. Okay? And Tiktok incentives to keep you there as long as possible. Right? And then Tiktok can make money based on your time. Your attention right? Advertisement. Right so. But your incentive is different. You have family. You have study, you have work right? So Tiktok doesn’t care that right? So which means it’s not fully incentive aligned with you. So when it is very smart. Well, it will mislead you or manipulate you right to a decision that maximize the agent. Profit but not yours. So this we will talk about in the second half. Another example is that Amazon, for example? If suppose Amazon knows what you like, right well, it may still have their as an agent. You are the principal. You’re you’re Amazon boss. Right? Consumer is your boss or your god, but they will not serve you 100%. Because Emily essentially cares about profit. Right? So what about this this product? Actually, I know you like that one better. But this one, you know, I make more money. So what do I do? Right? So that is the principal agent problem incentive conflict in this kind of context. So that will introduce problems. Okay, some message. Okay. let me conclude our discussion on consumer preference or the demand side the most fundamental setup. Well. so, consumer, have preference in our world, this toy model world or theoretical world. This preference will define as as good as you know, math right like a function which we can assign values to items, to good, to good, right, to alternative, to options. Right? If we have this preference, we have some construct or models that can analyze consumer decision based on utility and budget side right? And the individual consumer making decision based on their preference, based on their constraint, will lead to an aggregate function called demand function that characterize the behavior of the behavior of the market demand side behavior. And we still have the supplies that missing. We’ll talk about later. And most importantly, to allow us to make comment. Okay, and to make judgment on what is good and what is bad. We have come up with a concept called consumer surplus, which is not that familiar. It captures like us people in the country. Suppose all of them are consumers. How happy they are aggregately right? So that is less familiar. But we are probably very familiar with the supply side, like your Gdp. Right, or the the like the profit of the firms. Right? Okay. But like I said, consumer surplus will be very important, depending on your perspective. If you’re a scholar, right? You study, you make the independent study you cares about aggregate welfare because you are independent, you do not take a side with the firm. If a government who cares about consumer side consumer surplus, especially because nowadays firm are huge, right firm, has very strong political power and bargaining power. Economic power. You know. You see they have a very large influence, and they will bargain with the Government, with the Regulators. Right? But consumers are very small. That’s why, from government perspective, they may, cares about consumer surplus a little bit more also because you know, some of the consumers. Yeah. relatively disadvantaged people or community in a society right? They may be low income people especially vulnerable to hardship, to economic, to bad economic conditions. So that’s why, as a policymaker, we may put extra attention to consumer surplus. And then we need to have that analytical tool or framework to understand. What do we mean by customer surplus? And this will be the starting point. Okay. any assignment questions. And in our future lectures I will ask a lot about consumer surplus, but from a firm perspective. And we’ll introduce in next lecture that what you care is about is profit. Okay, firm will maximize profit. And but to maximize profit you need to have an understanding on what consumer likes right? That is demand function. Okay, once we have this demand function, we know how consumer behave and respond to our product design. Okay, respond to our marketing effort like promotions. Right? Advertising, how consumer respond to our price? Right? All the marketing instruments or product product design. And then we can we can. making all sorts of strategies, that maximize our profit. Okay, that that will be in our next lecture. But next I’m going to talk about next, I’m going to be a little bit more formal and quantitative about the demand function we have set up. We’ll talk about concept of demand elasticity. But like I just said firms, right? So they want to maximize profit. So they they need to have an understanding on how the consumers respond react to all sorts of instrument. They have right. And in practice they can do marketing research. They can use big data and try to estimate empirically about consumer preference or this demand function. But. theoretically speaking, we have, we, we, we will define a concept called elasticity that exactly captures some sort of sensitivities or the causal impact of consumers to the instrument available to firm. We focus on one particular called demand elasticity to price. Okay? But you can. You can have elasticity to advertisement. Right? Like. For example, when I was in e-commerce company one very important matrix is to study the elasticity of app download to my advertising, spending right, and I have to empirically estimate the elasticity in all the Channel, like, for example, every day we spend like 10 million doing which is Chinese? Tiktok. How much of this money, how how many download is generated by my advertisement spending in this particular channel? Right? How many new download is generated by my advertising, spending in a search engine like Baidu right? Or in the app store, or in Android store, right or so that is called the estimation of this elasticity. There’s many related method and topic like the one I mentioned previously called demand estimation. There’s also a whole field actually teach another course for level called the causal analytics. And that is related to this causal analytics, which is basically to estimate return on investment. The causal impact of consumers or or the other party. How how do they respond to your your action? Your strategy? Okay? And our discussion here is mostly radical. So I will introduce a concept and make sure you understand what they are and what they intend to measure. And if you want to know actually how to how to estimate them, and there will be many, many other modules you can take like applying metrics, not sure if you hear about, and causal inference and the general prediction models machine learning. And it can also be helpful and causal machine learning. Also important. Okay, so we’ll focus on 2 concepts, one called demand elasticity, the second called cross elasticity. And here all these elasticity is defined as quantity, elasticity. With respect to price. Okay, as I said, the elasticity can be very, very general. Okay, it can be like investment elasticity to tax policy, right or investment elasticity to like firm. How much like Singapore, for example, if I’m setting the inflation target certain level, suppose it’s something that a decision maker can control. Right? I want to understand elasticity of like next year’s investment. national investment from the private sector. How do they? How are they sensitive to my action? Right? So if I understand the necessity, and I can, you know, set my decision correctly to induce the outcome that I desire. Okay, but we’ll we will only focus on demand here. Just want to highlight these concepts. Very, very general, okay. cross elasticity measures. How is my my quantity and my sales sensitive to my competitors? Price. in the context, in many contexts called differentiated product market, which means that firms are selling a non identical, different branded right apples. And so they are different differentiated product. But they are also substitutes because they serve roughly similar. I use term roughly, because well, you can measure how similar they are that is again called demand estimation. You can estimate the level across elasticity based on the data. Okay? And these serve roughly similar. Need, right? Like, playing games, right watching videos or calling friends talking chat with my friends. Okay, that’s for for example, cell phone so the the economic principle is that sometimes reducing price will attract more customer. That is, when your elasticity demand analysis. Elastic. Okay, when it’s elastic, means that right? It’s very intuitive means that consumer response more dramatically. Okay, my, for example, lowering my price by a little bit will induce a very large increase in sales. Okay, so that is what we call elastic, right? What the other way. Is that why cut my price a little bit? The customer doesn’t respond that much that is called inelastic, and you can see that immediately. They will, they will have implications on my strategy, my pricing strategy. Right? So that will be covered in our next lecture on pricing. But 1st today we will. First, st lay a solid foundation on this concept called elasticity. Okay, how consumer respond and will be very important when we talk about a pricing strategy. Okay, or the supply side of the market. Okay? But the 1st problem we encounter when we try to make a claim on how sensitive is the uni dependency. What do I mean? Well. you can think about examples when you talk about like oil demand to oil price. Well. in 2 theory, I want to have some quantitative measurement on on sensitivity. Right? But it depends actually, in practice, it will depend on the units. How do you want to measure the quantity right? I can measure the quantity in 1 million units or 1,000 units right? That will imply different quantity, difference, level of sensitivity. So that’s the jazz. you know. Dy dx okay, like. Well, exchange how much y change. The most simple and direct measure may not be sufficient because it’s unidependent. Right? You can also compare across the 2 scenario examples. and the unit measurement can be very, very different. They are. You are comparing apple with bananas. Right? So for this measure to be sensible that can that is generalizable and hold meaning across context, and is comparable within the same context across applications. We have to have some measurement that is very generalizable. That is not unidependent. Okay, so what do we do? Well, we will normalize the quantity and price. And that is the concept called demand elasticity. How is it defined? Well. I have this term. You can call it E, or absolute. I will call it absolute, which is diagnosticity. Is Dq divided by Q, then divided by Dp, divided by P. Okay, so what it is? Well. so well, it’s a ratio of 2 quantity. Right? Okay, it’s racial to quantity. Well, I think I can rewrite. Rearrange the 2 quantity into 2 term. Right? I split the Dq divided by Q is just a fraction. Right? Okay, I split that. You can see this Dq divided by Q. Right? Using a different color. Oh, okay, then, this is Dp, divided by P. Right? I just played the term. and the 1st part is the Dq. Divided by Dp, the change d means change. Okay, change in Q divided change in price, and I have highlighted in the previous slides. There is issue with this measure, which is that it depends on the unit right? It depends on. If I measure price in dollar, or 1,000. But Dp divided changing price divided by price will cancel out the units. The measurement I’m using right? So that is that that sort of normalization make my measurement independent. Right? Make my my elasticity number independent of the scale of the I’m using. Okay? So that that’s very simple. So I just normalize the changing. So this this is called a percentage change. Okay? Because you change value by the value, okay, and changing price divided by price. So it’s also called percentage change. Right? Okay, percentage change. Okay? And if you know how to take derivative with log quantity. okay, log. Q, well, log. Q, take derivative log. Q is basically Dq, divided by by. Q. Okay, so that’s this part, the second equality. You can see that the elasticity is what it’s a change. D means change. Right? D means change. The change in log quantity divided by the change in log price. Okay, this this equality. You need a little bit of calculus, right? The second equal this one, the 3rd equality. You need a little bit of a calculus. But after this sort of very simple translation you realize that there are 2 ways of representing what we are talking about. One Dq divided by Q normalized changing quantity divided by normalized changing price. Okay, I call it percentage change. And it actually equals to the change in log quantity divided by changing log price. Okay? And you’ll have some implication later, when we interpret the numbers, and I will go back to these slides again. Okay, let me sort of try to move on. I think I have more formula here. Right? So it’s a percentage change in quantity divide a percentage change in price. Okay, so the D is a continuous measure. So like a very small change. But you can use a delta means sort of like a larger change, and we will not be very rigorous about the notation. The main idea is to give you intuitions. Okay, I will not, you know. I will not make like, you know, the journey has to be continuous, like your change in quantity has to be continuous, like a very, very small right? Approaching 0 or by 10 you cut a little bit what happened only to change it from 0 to a finite number. right? And quantity at 0. This is like, has infinite slope. Okay? So it’s infinitely elastic at the this corner. Okay? And it’s very elastic here, because you exhaust all the demand. If you cut your price further, there will be no response or quantity anymore. So elasticity is 0. Okay? And suddenly it will change continuously from very elastic elastic demand to elastic demand. You will have, we’ll have definition later. So when this greater than one. We call it elastic. When the absolute value is more than one, we’ll call it inelastic, and somewhere is like the threshold, right? The the point where the 1% percentage change in price. We introduce one percentage change in quantity and you will have some implications to revenue. We’ll talk about it later. Okay, so this is to make give you a very strong impression about the difference between, like the curve, the the slope of demand curve versus elasticity. Okay? And elasticity will be the term that we use a lot when we talk about firm strategy. Okay, firms, pricing decisions. And this graph shows you constant elasticity demand curve. Okay, let’s look at a few examples here. So let’s evaluate this. This. This price point. Okay. and this is absolute equals minus infinity. Okay, what does it mean? Well, this is a horizontal curve, right? And it only has quantity at this price. Okay, if you undercut. if you charge a higher price, you will lose all the demand. Okay, you will see this curve when we have the supply set at very competitive market. The infinite amount of firm. Okay, the infinite amount of firms are competing on homogeneous or same product. Right? Then the demand elasticity can be infinitely elastic, meaning that suppose there are many, many people selling selling ice cream exactly the same ice cream, right? All at 10, and no one is able to increase to 10. I can now increase my price by a little bit. Okay, so this is the marriage right? This line is marriage. Okay, quantity equal to one. You are married right, no matter what the price is. Okay on this line. And we can compare this 2 lines. This is what we call elastic. Okay. absolute is greater. Absolute value is greater than one. This is called inelastic. Well, it’s a little bit, a little bit continual, because you may, you may, you may, you may. You may think this line is actually flatter right, or or less less steep. Right? But this, remember, this is price. Okay, this is price. Let’s look at this, what we call very elastic line. What happens if I change price right from here to hear what happens. There is a very large change in quantity. Right? How do I visualize that? Well. and I’m going to use the this color. Okay, if I change price from here. Sorry. This is one price level, right? Price should be mapped to the Y, okay, price should be mapped to the Y and quantity should be mapped to the X, right? Okay? Oh, I’m so bad at drawing this. Okay, so this is quantity. Right? This is price. Okay, if I drop my price, what does that mean? That means that I’m moving downward right from this point of price to this point. And remember, we talked about. I’m not shifting demand curve. It’s changing price along the demand curve. Right? So. And the changing quantity will be this difference right? There’s a difference. Okay? And you can see that a flatter slope here means that a change in price imply a larger quantity change than this one. Right? So let’s see, I’m using a different color for the other one. Which color at this. this. Okay, same thing. Okay, I’m that’s looking at this price level. Oh. okay, going to be very, very high. Okay, this right? And this price level translate to a quantity here for this demand curve right? And you can see that the changing quantity is smaller when the demand curve is deeper. Okay, when demand curve is deeper. Okay, that means that demand is not very elastic. Basically, there are not that many consumer leave within this region. Right? Think about what is demand curve actually measuring, because this is aggregation over a lot of consumers. Right? When I talk about changing quantity. Actually, you can implicitly think about consumer switching away. Individual consumers are starting to exiting or choosing to not purchase. That’s when demand curve. That’s when quantity sort of decrease. Okay, quantity decrease. Okay? So if it decrease very fast in 2 price levels, right? So from a high price to a lower price. If quantity decrease a lot, what does that mean? That means there’s a lot of consumers leaving this region. The winnings you pay is in between the 2 numbers, right? The willing to pay is between the 2 numbers. That’s why, when I cut my let’s increase my price they they left, or my decrease the price they start buying my product right? Think about it, think, think, sort of recall what we talked about in customer choice right? When you change your price you’ll be consumer switching, switching out right? Why are they switching in, switching out? Because they are the marginal consumer who are just being indifferent, indifferent between my offer right? If they’re just being in almost indifferent between buying and not buying, I cut my price a little bit, you’ll be happy, and there will be more consumer, marginal customer coming in. If I increase my price. Your margin of the marginal customer who already bought my product will exit or choose to not buy anymore. Okay, think about that exercise. Okay, so there’s a lot of things going on. We talk about the changing along the demand curve. Okay. let’s move on. And I mentioned, why doing cares about this? What useful result you can derive from this elasticity is basically linking the elasticity of demand to revenue, which is something that firm cares about. This point will come up again in our future lecture. You can do a little bit of math. So the change. So delta means the change. The change in revenue equals to the change in price, times, quantity, right, because revenue is price times quantity. Okay? And there’s something missing here which we will introduce in future lecture called marginal cost. Because suppose I’m making 10. It doesn’t mean I make 10 right now. I cut my price to 1 right that is known, but that is what that is the Delta P. Right delta P. Because my cut of my price. So this is the cut of my price effect, right? Like per unit per per unit per queue. I will lose 1 I will earn more dollar queue will be greater than 0. I will, I will sell more right, I will sell more, and that times price is the incremental revenue generated by market market expansion, right by market expansion driven by price cuts. and you can see that well, if my demand is very elastic, what does that mean? That means cutting price is very effective in gaining demand, right increase or expanding market. Right? So that is the case where my revenue will increase when demand is very elastic. Okay? And this is the algebra, and you can visualize it. Using graph. Right? You see, in the next slide. Okay? And I will skip this slide. Just use the previous previous elasticity. You’ve seen in earlier slides, and you and apply to the formula you’ve seen in the last slide. You’ll get the numbers. Okay? So this is a visualization. Okay, so this is a visualization of the 2 effects I mentioned in the equation. Okay, exactly. The 2 equation. The one is when I cut my price per unit. Revenue decrease right? Because I cut my price from 9 per unit. Revenue decreased by $1 right and second effect is market expansion effect. Because of my I cut my price. Quantity, expand right quantity expand, meaning I have a higher revenue. That’s the 2 region, and that is being colored in right and the blue. So what is this one? Delta? P. Well. we are looking at a price decrease? Right? So what does that mean? That means price. Move from e, 1 to e, 2, right from e, 1 to E. 2. That that is a price decrease right? Because this is p, right? So it’s lower the price in this graph lower the price. Okay, so this is the effect of Delta P, right? Because I lower my price by this amount. Right? And this is a quantity. Delta. P. Times. Q. That is a. The pink region, right? That is. that that is Delta. P. Times 2. This this one, right? Okay. And the Blue region. Delta. Q. Because I cut my price, my quantity increase right? I sell more. So that is a game, right? That’s the other. That is the game. So I have to. Here it goes where I did versus my loss. Poon Yi Ming Poon Yi Ming 01:27:32 Thank you. Chen Nan Chen Nan 01:27:36 Okay, I hear someone talking. I assume it’s accident. Move up. Okay, okay? So you’re comparing 2 quantity, 1 1 is last l. From which is the the red region from decreased margin and call it decreased margin marginal return per quantity. Right? Okay. So that is the the last l, this is L L can be used a different color. This is L, okay, and this is G, this is game. Okay? Again, it’s from market expansion from the quantity. So eventually my revenue will be determined by the 2 region, and you can see that that determined by, if the value is greater or smaller, that minus one. Okay, so that is exactly what the algebra is telling us. Okay. I can. You know, I can repeat this part again our next lecture, if you find the part a bit difficult, but it’s basically repeat the same logic or argument, using both algebra and and try to visualize it. Later we’ll introduce a lot of concept, for example, cross elasticity. How your price respond to a competitor and all your one product, if you are multi product firm, you’re actually competing, or you or consumer considering between your product line, right? If you, for example, apple iphone and the ipad. Right? Are they substitutes or compliment? That’s actually empirical question. Okay? Oh. okay, cross price elasticity. I will talk about this in the next lecture. But those concepts are relatively straightforward. I encourage you to take a look. Many of the concept has been discussed in today’s lecture. They are very straightforward. Okay, and this example is an empirical calculation. Or to take the concept and theory elasticity into the field using data. Okay. lastly, I will have a discussion about the switching cost we talked about. And how does that relate to demand acidity and processity? And what is the implication to consumer and to firms or to market participants. Okay, so that is, I will leave to our next lecture, and I will stop here, and I wish you a happy lunar New Year and enjoy your vacation and see you next week. We will still have class next Friday, because it’s not a official holiday. Okay? But I will still keep everything online and recorded. okay, thank you for today. That that’s all I’m seeing. Okay, yeah. Happy lunar New Year to everyone in our class. I hope you take a good vacation. Enjoy yourself. But assignment wise online. Okay. But you have a lot of time. You don’t need to do it during the holiday. Enjoy yourself. Okay? Regarding what a diamond example in the previous slide, you explain what is expenditure? Expenditure means twice times quantity. Okay? are you still there? Yeah, I I basically price. Okay, if you still got any question, please hold on canvas. I’m going to switch my attention to our students in the lecture hall. So for all all those on zoom, thank you. And okay, see? You see you next week. Okay, I’m going to talk about going to talk to our student in class, and I will stop recording. Now let me. is this stop recording.

3. Cost

This is our 3rd lecture, and let me 1st start from some course administrative. so again, we have this discussion for so you’re welcome and encouraged to post questions and also help each other to answer any question that is related to our lecture or course material, and we have the lecture recording posted here, and you can click on the link and take a look on our course, recording. The one more thing is that I uploaded the 1st assignment. there are, I think, 25 questions, and you can submit your answer, and once you finish, you can submit your answer to this assignment, and I think, probably already solved by some issues pretty fast. Didn’t enjoy your New Year holiday. Right? Okay? So let me start our lecture last. In the 1st lecture we introduced concept called Preference, a utility function, and related to that we introduce a indifference curve which is used to analyze the shape of the margin substitutions of preference of consumers, and we introduce a consumer that face some budget constraint with limited resource, and try to make a trade-off and make a decision on how much to consume for a pair of product. See apple and the bananas and the framework can be very general. You can have a much larger set of alternatives, and you can ask consumer to make any choice, given any constraint. So it’s a constraint, optimization problem, and then we derived individual level consumer choice. And we introduced a concept called consumer surplus, which is which can be thought of as the profit enjoyed by the consumer customer willing to pay 3. Then there is 3 of enjoyment or utility which we call consumer surplus, and if we aggregate over consumers, we will get a important construct or curve called demand curve demand curve basically tells us a relation between prices and quantity demanded in a market. And we and related to that. We introduced some important measurement to measure the the property of the demand curve, which is called demand acidity. and importantly, as we explained in previous lecture demand. Elasticity is free from the measurement unit because we are going to normalize the change. So we call it the percentage change in quantity divided by the percentage change in price. Right? So because we are taking this ratio into full screen mode, try, okay? Because we normalize the quantity, right? So we were computing the change, the percentage change in quantity divided by the percentage in price. So it doesn't matter if I measure the pricing dollars or cents or 1 million dollars or 1 billion dollars. So the at the end of the day, the measurement of elasticity is free from the the exact measurement units we use. and we show some equivalence. So these elasticity can also be thought of as the change in log quantity divided by the changing and price. This will be a useful later, when I introduce a particularly pre code application. Okay? And we are done with this. We can use this elasticity concept. We introduced to study the implications of a price change to my revenue. Okay? And this will be important in our next lecture. We talk about pricing decisions. Okay? So you can think about a managerial decision of selling price. Say, we start from a level price if we increase the price, and there will be 2 consequence. The 1st one is because I I changed my price, my per unit profit change right? Suppose I'm still selling out the same amount of product, but because my change in my price there is a price change effect, right? So my per unit profit change. The second effect is like an indirect effect, because I change my price, my demand will change, the quantity will change right. Suppose I cut my price, lower my price. I expect to have a higher demand or greater sales, but there is a quantity effect. so we can decompose the effect of of price on revenue to 2 component one is the quantity. The other is the the change in per unit profit. Okay? And last lecture, we went through this equation of decomposition, and at the end we show that? Or we, we claim that the change in revenue can be directly linked to the demand elasticity. So if your demand is more elastic, right? And suppose price falls, and there will be a large quantity effect, because your demand is very elastic, right? So there will be a large quantity effect, and then the quantity effect is positive, right? Because we are talking about the price part right? Lower price. The quantity fact would be positive in a sense that I'm going to sell more products. So in that sense, if the demand is very elastic, that means the demand or the quantity fact will dominate right? So I will say a positive quantity fact that the dominating and or compensate the per unit natural effect of price on price right? So that means, if my demand is very elastic and epsilon, which is a measurement of elasticity in most of the cases we talk about in the lecture, you will be negative. So in that case, epsilon, the absolute value is greater than one which we use as an indicator or cut off for elastic demand. If Epsilon is greater than absolute value is greater than one. That means a one percentage change in price will produce or generate a larger response. greater than 1% change in quantity, right? So price fall if demand is last. But if demand is demand elasticity equal to minus one okay, or absolute value of demand, elasticity equal to one. Then a price ball locally will, you know, will now change the revenue. Okay, so last lecture, I also introduce this graph illustration as a more intuitive way of decomposing the effect. Suppose we still, you know we are looking at the price decrease. Okay. Price, decrease the price from decrease from e, 1 to E, 2. Right? And again, there are 2 effects. One is the direct price effect. Okay, because price decrease right? But has decreased. So the price effect is negative, which is the right one. So this one is negative. Okay, so that is the loss. Okay, loss equals to Q, which is a quantity times minus delta. P. Delta. P. Is a change in price. Right? Okay. How do you? Well, why, why, the red region is the loss in revenue. Well, okay. So look at a point E, 1 point e. 1. The revenue is P. Times. Q. Right? So it's this, try this square or rectangular. Right? So this one, this part. Okay, if but if I decrease my price from e 1 to E. 2, you're looking at this red rectangular right? So the wider one. Okay, right? Because price decrease. So you can see that in the 1st rectangular this region is your revenue, and under the second, the lower price. So you have this wider right rectangular right, this wider one, and the white region, the white area is like overlapping region, right? So that is like the same amount of revenue under to price regime. Okay? But then, right? Why is the previous profit? Sorry previous revenue? Right? That disappeared? Because I decrease my price. but I also have a like a game right for a positive effect. Which is this the Blue region? Right? If I change my price. I got this wider shape right? And I have this actual blue region, which is positive, which is the demand effect or the sales effect right market expansion because I decreased my price. Okay, and market expansion equals to well a price times the expanded market or the increased sales. Right? So the overall effect would depend on the size of the 2 color region. So if this red region it sorry the P. Times. Delta. QP. Times, Delta, QP. Times. Delta Q is the blue region. The blue region is greater than the red region. and the gain is greater than the loss, so I will make more money, or I will collect a larger revenue. Okay, we haven't taken into account marginal cost, or you know, this is not the final profit. This is a revenue effect. Okay. you can do some simple rearrangement. And you prove the logic that you know the overall revenue effect is, if and only the elasticity is great is more than minus one. Okay, then you will. Your revenue will increase. Okay. this is a slow motion of the the effect of a price decrease on revenue change. I mean. let's move on. So this is a recap of what we learned in a previous lecture, and you can define a lot of very convenient concept to describe the behavior of demand for the shape of the demand curve. Why is cross price, elasticity, cross price. Elasticity is very important in competition, right? So one firm is concerned about a competitor's product offering, if competitor cut their price, what happened to my demand right? So if the 2 product is comp a comp a substitute, what happened? Well, substitutes, meaning that consumer substitute between the 2 products? Right? So if my, if the competitor cut their price and compares competitors product offering becomes more attractive or competitive, then some of my consumer will switch to competitors. Product. Right? Okay? So in that case I will lose demand. And you can define as a formal way of measuring. This is called cross price elasticity. And you can see that this absolute now has 2 subscripts. Right? I and J, okay, I means we are. Look, we're talking about demand for product. I. So I is the focal product. the quantity I'm examining now. So that is product I, and probably J is another competing on competing product. If they are substitutes. Okay, so we are talking about 2 product I and J. If J. Price change will happen to ice quantity. But the the previous most standard elasticity we defined in that last lecture is called own price elasticity many times we call call it price elasticity or elasticity, for short. But it's actually own price elasticity, meaning that I equals to J. So, my, what? How would my my quantity change if I change my own price? Right? But if other change their price, my quantity may actually respond. And there are few cases. So if absolute ij, the cross elasticity is greater than 0. 0. Okay? And we see I and J, the 2 products are substitutes. We can think about many examples of substitutes. For example, 2 cell phone, right? Samsung and apple iphone, they can be substitute, because consumer will be choosing between the 2 of them. They will. Many consumer will choose one of them right either. So either Samsung galaxy or iphone. 15, right? Okay? And you can actually have substitutes substituting product within a single firm product line. I see, apple right? Apple may have different version of iphone different version of iphone. And those are sub. Those are substitutes. Okay? Okay, because you can think about older generation of iphones actually competing with the newer generation. Okay, this is like some pricing problem or doable good pricing where? a seller actually competing with yourself? Okay. So if I, if the if or if there's a new version of iphone that is coming out or introduced to a market and demand for the older generation will decrease because consumer will substitute a newer generation and same thing. You can see like there are different AI models. Right? So they are also considered substitute. and you could also have compliments where the consumption of one product will increase with the consumption of the other. When you consume them together. The utility is actually high. Okay, you can think about, for example, many of the apple product actually are compliments like, like your ipad, or you're using Mac. And also you use iphone. They could be complements in a sense that they share same operating system. Right? So they are, you can. If you if you own a a whole set of apple products right, the join. Utility will be higher than you own them independently. So loose. Loosely speaking. you can think about one plus one greater than 2 because of the the joint, the utility of owning more than one apple device. Okay? And of course, a 2 product can be independent if there's no mutual influence between them. Right? If you change price of one product, the other doesn't respond. Okay. another. Another useful definition is called income elasticity, which describe how sensitivity demand of your product consumer income at a high level. If you look across years right, the economic condition may fluctuate or change may may be improving over time. Okay? Or for individual consumer. There could be some negative shock like, get unemployed right for some reason. And then there's a negative income shock so that will affect consumer budget constraint, right? Because earn less money and the choice will change right choice or the consumption pattern will change. And accordingly we can define something called income elasticity. So when my income, we use the letter Y change by a given percentage, how my consumption for given product change and depending on the sign of the income elasticity you can have inferior good normal goods. So most most of the product. You can consider as normal good in a sense that if a consumer get richer, which means why increase income increase right and the consumer will choose to consume more right? Like, maybe by we'll change and change or buy new cell phone faster, or even on multiple devices, right? Multiple laptops. Okay, so normal, good. If consumer get richer or more money, they will purchase more. But if you're good, you can think about as some in product. That is not that attractive consumer purchasing the product because she's not rich enough, but at the moment she becomes richer, earn more money, and she will choose to consume switch to something more expensive like upgrade her consumption. Right? Okay? And similarly, you can have a luxury product where our income increase and consume actually will be very will all of a sudden switch and consume much more demand. Very elastic to to income. Okay? So those are some useful definitions. I'll give you an empirical example of elastic elasticity and try to asthmaid a model or demand of gasoline demand okay, in the United States. So we use historical data from 1953 to 2,004. So, as I mentioned before, we do not teach in this class how to empirically estimate demand curve. It can be as simple as the OS linear regression. Okay? But if you want to measure, let's say, demand elasticity without bias. Okay? Then it can get quite complicated because of something called hemogeneity problem. If you study like applied econometrics, because essentially, we are doing causal inference, we want to estimate elasticity, which means one exchange. how much y will change? Right? So that is actually something called a causal impact. And it can be very complicated. But for this class I will not dig into that. But I just assume you can just run simple linear regression, and your estimate is unbiased. Causal effect. Okay, what you can do is you can regress log quantity of log price. And as we defined previously, the slope. okay, the slope of a log- log regression. It's actually elasticity. If you go back, many slides to the point when we introduce this elasticity definition, look at this quantity is change in log log. Q, divided by changing log. P, right? That is electricity. Okay. so assume your linear regression is correct, measuring the causal impact. Then the slope parameter. Here it will mean data system. Okay, the slope parameter, which is a minus 0 point 0 4 2. Okay? So because that is when price a lot price change by 1%. How much lot quantity will change right when you change by one, and log quantity will change by minus 0 point 0 4, 2 times one and interpretation. Remember, if you if you already take, I don't know, like any regression classes, we teach you how to interpret these coefficients. Well, it's holding everything else constant, right? Holding everything else constant. So the the intercept the same and income is the same to 1% change. Okay, one unit. Changing log price will lead to 0 point 0 0 4, 2 unit changing quantity. Okay, and same thing. You can. This coefficient measures the income elasticity. And you can plug in the values. Suppose from 1953 to 2,000 to 2,004, and price has increased by annual rate of increase by 3.9, and income is by 3.4. If you compute the growth of the change in quantity implied by your model estimates. Okay, this you plug in the value right? 3.9 and 3.4, and you use the elasticity you estimate. In your linear regression or as model and you will get the predicted change in quantity. Which is this one. Okay? And the actual growth rate actually, 2.7%. So there is some prediction error. But roughly, you know, not too different. 1.6 predictive value and actual value, 2.7%. Okay? And this is a very, very simple example of how do you ask me? This city using the simplest OS regression by regress, log quantity and log price. Okay? And if you do is like a real world setting in, let's say you are, you are business analyst or your data scientist Amazon that is studying how to set price. And you want to estimate demand curve, and that you will use more sophisticated techniques. Okay, something like demand estimation based on, you know, discrete choice models, and so on. And you have also to take into account inogeneity issues, and some of the discussion is in the optional material on campus, which you can take a look. Many of the I say e-commerce platform, or I see tick tock right when they try to ask me this elasticity. They can use experiment. Okay, they can use these experience. For example. If I'm Amazon, I'm an airline company, right? I can randomize price and change my price by 1%, or maybe 10%. And then I can observe, if I do A B testing, I have treatment group and control group. I can quantify and measure how much of my quantity change. And it can compute the elasticity using a B testing. Okay. so that is for the series on demand demand elasticities. I will conclude this lecture by discussing the important concept called switching cost that has limitations on consumer demand, and, as we discussed will also impact competition policy computation strategy. So what is switch switching cost? Well, switching causes. In reality, when consumer change their decision? Okay, they change from a default to some other or their current choice switch to something else. And behavior. Economics have shown that. And psychology research have shown that consumer somehow seem to experience some cost. You can call it inertia, you know, switching, or you can call it switching cost. Sometimes it's purely behavioral, like, I'm just, you know, I just good with my parent option, and it's costly for me to think about. Other alternative sometimes may involve, you know, time or effort, cost or financial cost. In in certain context. say, if you're using apple right? Or if you are upgrading or you're you're buying a new cell phone. You can. You can feel that right? So you have to migrate every your password and your document, your your photos to a new phone, right? So you clear that switching cost switching to a new cell phone. Okay? And if you pay attention, if you use ipod device, they have the they make it simple. They try to make it simple for you to switch within the apple ecosystem right? They won't make it easy for you to switch to Samsung. It actually will make it hard for you to switch to Samsung. okay, or to some other android cell phone, right? So because that is good for them, and that is sort of incentive conflict with you, because you want to have the lowest switching cost, and you want to switch to whatever you likes right, but for apple well, it has incentive or to keep you with within their ecosystem. They are okay, or they make it easy for you to switch to a new phone, because that means you are. You're paying them. You're buying more of their product right? But they will make it very costly for you to switch to. Samsung. You can also think about this AI platforms if if let's say open AI using let's say Gpu from Nvidia. Right so. And then what will you do if you are Nvidia, CEO and Jason Huang? Well, you will. You will try to make all your customers. You will try to increase all their switching cost. Right? Okay? So a simple cloud computation. Okay, so you if you use aws right you're also trying to increase your switching cost. Because by increasing your switching cloud, right? So their demand they cross the Scc. Will be lower, will be lower if let's say if you you if you you all your infrastructure is in aws right? So assume some other cloud platform. run promotion and decrease price and try to attract you. But still you're locked into this platform. And so that is the impact of switching costs and switching costs will make the market less competitive. Right? You can lock in the consumer and make the demand less less elastic. so consumer will be hesitant to switch, even if there is some attractive offer but in the long run, learning curve, learning curve can flatten. Okay, another issue is learning curve. Because if you switch to a new as a platform, you have to learn how things works right. You have to learn their Api functions, and you have to. adapt have to face some uncertainty as well. You don't know what will happen. How reliable your services. Okay, but that will smooth smoothing out in the long run, alternative solution may come out. Okay. Switching costs may drop. So in the long run, demand becomes more elastic. I think we mentioned this point in our previous lecture. Okay? But given firm. Let's see. Google, okay, if you if you notice if you use Ios device like apple like Mac or iphone. The default search engine is Google. right? Is Google. So you can. I don't know if you realize that or not, you buy new iphone. And you want to search for something. The default is Google. okay. It's actually very impactful, the default choice. It shouldn't right because you are so smart your Soc undergrad. And you know it's not difficult for you to switch to a different search engine. Right? But most of us don't do that. That's the default. Okay? And I think I have some comments in next set of next lecture slides. Actually, Google pay Apple a lot. Many billions. I think tens of billions of Usd to be the default search engine. So you know how powerful the default is and default gives you market power. Okay? And that to the extent Google is willing to pay. I think 20 billion Usd per year to apple just to be the default search engine right? If everybody is so smart. It doesn't really matter right, or everyone has 0 switching costs to do. Make the decision their search engine decision perfectly right. So it doesn't matter which one you give me as a default if you give me Google. But I like something else. I will switch right. But it's actually not the case in reality, because people people are the limited attention or they are lazy, they don't pay for attention. They just don't want to. Don't want to switch so that creates inertia, and that that makes the market less competitive. So it will be very hard to have alternative. A different search engine at the end of day, right? And then Google is actually willing to pay 20 billion for to be the default search engine. So that tells you how profitable it is to be the default search engine, right? That is very surprising. But it also tells you how much Apple can charge Google right? Because well, Google can see. No, I don't want to. I pay you right. But Google in that pay apple a lot of money to be the default search engine that tells you like, who is the real boss? Right? Apple apple is the gate gatekeeper in this game, right in this context, because Apple is able to ask Google to pay 20 billion just to be the default search engine. You guys are are very smart. You know how cost the marginal cost of 2 of letting Google be the default search engine is almost 0 for apple. Right? But you can. You can. You can earn a lot of money. And I think our comparison number is how much Google spend. or it's R&D total. So that actually, the the default search engine part is, I think it's like 20 or 30% of the total budget for Google's R&D, so you can see that that is called market power. Okay, the ability to charge a high, very high price. Okay? So that was some comments. the default effect I encourage you to check. There is actually the lawsuit going on regarding this so called gatekeeper or default search engine case. and you can see, like a lot of the platforms. What they are doing is that they're trying to be your default. Okay, once they they are, they succeed in doing that. They make huge profits. Okay? In last, in the next lecture we'll talk about scale economy. That it. It pays you to be big. Okay? And it pays for you to be the default or to be the gatekeeper. Okay, apple is your apple is the gatekeeper for the whole Ios system. Google is the gatekeeper for search right? And the AI platform. They want to be the gatekeeper, for in the future, for all the you know the the intelligence, then that is, that pays very, very well, and that gives them, you know, monopoly power or market power ability to charge very high price. And when we talk about supply side, we'll introduce a concept called perfect competitive market, and that is a world where you can. You don't have market power. But in the real business setting, especially relate to innovation. Okay, like firms want to be the first, st and firms want to be the only one as such that they can charge a monopoly Prof. Price and earn actual profit. Okay. I think we already discussed the implication of high switching costs which reduce cross elasticity and give firm market power and consumer less likely to switch or to jump to the substitute if they are locked in. Okay? And that will impact investment incentives and affect market competition as well as pricing strategy because of the existing switching cost. I think there's a related assignment question because of switching costs for me. Use some, you know, pricing strategy that is more dynamic, right? So they try to attract consumers. Once you switch to the platform, well, they will increase the price, or you can think about the other ways. They can give you a a new user discount. Right? New user discount. You try to compensate the switching cost. Okay? And you also related to bundling of hardware and software. And you will see some examples in the assignment questions. I will. Just, I will move on to our next set of topic. which is the supply side. Okay? So we'll have, at least 2 lectures on the supply side. So at this moment we already finished discussion on the consumer side. Recall what we have. We have a consumer who has preference and have budget constraint and make choice to maximize utility right and then aggregate. We can have demand curves. But then we define, demand electricity across electricity and consumer surplus all those important concepts for us to to understand the welfare of consumers and the decision process of consumers. But from now we will move up to the supply set. So imagine, imagine your marketplace on your society. They are probably speaking 2 side when it's the consumer side. where is the supply side or firm side production side? Okay? So going to to talk about production function, cost structure and supply function. You can see some very similar but structure us in the demand side. because, essentially, what a supply side does is also to make decisions right supply side, also making decisions, making decisions about the input of many human labor to hire, how many compute to purchase right? How many data to purchase? How many production inputs? Or which kind of inputs to buy and depending on the technology different firm can be, have different productivity. So we define a function called production function, which is Co which correspond to the utility function, reserve in the consumer side we call the utility function is a measure. How happy consumer are given a bundle of products, right? Or giving a set of alternatives or a particular consumption. Consumption bundle. Okay, and production function defines the level of productivity or the quantity of firm can produce. Given a set of inputs. Okay, so it's very, very flexible. And using this function. F, for now, without specifying the functional form, it can be very, very flexible. Okay? And it can apply to any setting. Okay, you can think about production of intelligence applied production of ice cream. And we'll talk about production of T-shirt. Okay? And also some inputs, we will use 2 of them like the classic textbook example, P. And LK stands for capital. L stands for labor. Okay. this is like the the old school textbook style. But now I think if you think about like a AI setting, and you can think about very important elements or inputs like data, right? Gpu and talent people like Cs graduates, okay. we can define similar, like the what we have defined in the consumer side, where we have a difference curve, you can have something called iso clunks or control lines that connect points of the same level of productivity. Okay, if you keep the same level of output, right? So this particular line. a set of points right in the space of P and L, like a plan. P. And L. Right, it will give you same level of output. Okay, so correspond to the indifference. Curve. same thing for simplicity where we only, you know, have these 2 inputs, you can have a larger dimension of inputs. Okay? But if you have only 2 inputs, then the the sort of the counter line. Isopaths will be a curve, a line right in this on this plan, but if you have a larger dimension, then it will be a something like a set right, a set of points, okay? And generally similar to the indifference curve. The isotrons are generally convex, which capture idea of diminishing marginal return. Where? Which means that you don't want to have. You don't want to only have like data, right? No, Gpu, you don't want to have only have Gpu. But no talent people, right? So you want to have sort of a mixture of inputs to maintain a high level of productivity. So that's the idea of diminishing marginal return. If you have too many Gpu chances are you? You won't be able. Your other inputs are not enough. Okay, so that will sort of have a diminished marginal value to your production. Okay? And the more convex. the more complementary the inputs. But the flatter by flatter, we mean is closer to a, to a, to a straight line, the the the closest, the substitutes for the 3 inputs. Okay, we can visualize this, using a simple example, plans and pilots on the left, and 2 type of beef in the right figure. Okay, if you look at a beef example on the right of the figure. They are flat, right? They are 3 lines. So this correspond to our the different curve of Usd and Sgd, right? 2 currency. You don't have really have a preference on them, because, you know, they are just. They can be exchanged freely in the market. And here I see the 2 beef somehow. They are close. Substitute right? You can marginally, you can substitute the 2 beef to type of beef at a constant rate. Okay, but on the left it looks very different. It's very convex to the region, to our region. Okay. it's very, very complex. Why? Because well, it's a very extreme case. Think about airplane right? And think about pilots. So assume you have one only have one airplane. Okay. it doesn't matter how many, how many pilots you have right? Because you only have one airplane. So they gives you the same level of productivity. Okay, so the blue line here is the Iso count or the level of productivity. Okay, it doesn't matter. Say, you can have assume you have only one pilot, right? It doesn't really matter how many plans you have, because you only have one labor to drive the to fly the the jets right? Same thing for 2. If if you have 2 pilots, and you need 2 airplane. But if you have 3 airplanes, doesn't really matter right? Because you only have 2 palace. Okay? So this is a very extreme case. The 2 resource are paired right? Exactly paired. It's a for really just for your illustration, and you can think about like you have programmers and compute right? So they are roughly, in some sense, paired, but not exactly right, because especially when we have this. In the age of AI, you can have one programmer which can use like 5 5,000 gpu, right? So it's not exactly paired as extreme as this case. Okay? And same idea you can. You can have a budget constraint for the firm. Right? So remember, in the consumer case, consumer has a budget constraint. Right? Try to split the money on apple and bananas. Here the firm can split the money on labor. And you know, equipment. Right? Okay? So the right line is a budget line and the optimal point optimal production will happens at the the tangent point here, when K. Star equals 1.6 and L. Star equals 2.5. That is the lowest cost to produce 2 units right? Because this Q equals to 2 right? So this is the actual count of Q equals. Q, 2 produce 2 units. And this is budget line. And this is tangent point. Okay? And you can apply the logic we had in our consumer decision problem. If I I cannot reduce the money. Reduce your budget anymore. Because if I reduce your budget a little bit and your budget your budget line will go lower and you cannot reach this. Q equals. Q right production line. Okay. so that is the the production function and the cost minimization that leads to a certain level of production critical for giving input prices are for for capital or for technology and w for wage of labor right? And you want to achieve a given level of target output level queue. And you can find the optimal input mix of P and L that can achieve the level of output and minimize the cost. Okay, so that is idea in this graph. this graph here, I want to achieve this level of production. And the the optimal mixture of inputs will be K, star equals to 1.6 and L. Star equals 2.5. Okay. okay, this should be L. Star, okay. But you can use a notation. Ld, if you like. Okay, if the demand for labor. how does the optimal solution or the optimal amount of labor change one wage or or interest rate change well. when wage increase, you can see that the the red line will become the red line becomes deeper. Right? Okay. okay, so you will have something like this. Right? Assume, you have the same amount of same budget. Okay? And you can, you can immediately realize that the the 2 equal Q is no longer achievable to you. If if R. If W increase, okay, which increase. okay. and what happens to your optimal. What happened to your optimal allocation of resource? Well, and in principle, when wage becomes more expensive, you will hire fewer people right? So that is what happened in the market, and when I say Gpu becomes more expensive, you will. You will buy that less Gpu, and you will hire more people. Okay. But if Gpu becomes very cheap, then you will hire that. You will hire less people and sort of substitute to technology. Okay? And you can think about many examples, and I will skip the the example. A particular, you know. Widely used production function is called Copeda glass function, which takes the following functional form, where qi is level output equal to Omega i Omega I is the total factor. Productivity. Tfp and times key, a functional key and times a function of L, okay, and it's a power term, Alpha and Beta, which will be parameters. Okay for given empirical context. Okay. and labor productivity is the output divided by the the total labor you have, like per labor per per unit per human capital. Okay, how much? How much is your output? And I think this number the labor productivity. You can measure it. If you look at the the company like open AI. I think my impression at least one years ago was, they have less than 1,000 people, and you can measure the the queue or the output by, you know their their the market value of the company or the amount of customer deserve. So this number is very, very high nowadays, because of digitization. Okay, one firm can serve like many, many people. Okay, even billions of people. But that is not the case. Let's see, 50 years ago, okay. that will create a a lot of problem about distribution. Okay. hmm. okay, for giving input, price, r, and WR is the price on K, okay, and W is the price on L, okay, you can think of K, as I say, Gpu L. As programmers, you have and determine the cost, the total, the cost will be well, your cost on the 2 input components, right? And Cq is the minimum cost required to achieve an output level queue. Okay, Cq is a minimum cost required to achieve a local output level queue. We would introduce more definitions for cost. And those definition will be important for our decisions. Business decisions. Okay. the computational profit quantity to produce. And whether or not we want to enter a particular market. Okay, now, let's give you most specific and useful definition for cost. The 1st one is fixed cost, and it's very intuitive. It's a cost that does not depend on the output. Right? So it's C, 0, okay, the cost when I don't produce anything. So unfortunately, this is the case for many business setting. Okay, you need to pay a very high cost, or at least a 9 0 amount of fixed cost, even though you don't serve any consumer. Imagine you open a noodle shop right? Right? Noodle restaurant. Right? So you have to hire people pay salary, or you have to rent a a place without serving any consumer. You already have to pay it. So some of the money like like you, you sign a lease, sign the place right? And do you? Rent? Could be considered fixed cost. Okay, that is independent from any number consumer you serve. And if you start an AI company right, you have to buy Gpu's hardware even. you know, without training any model. But in some cases you don't need to right. You can use more flexible what they call on demand compute. Right? Aw aws has on demand compute which translate a fixed cost to something we call variable cost. Right? Okay? You can. It's very flexible. Okay? The idea of cloud computing is to make things very scalable or very flexible, very elastic. Okay, suppose you you you have some cost. That will depends on how many consumer you serve. Right? You can install your program on on the cloud, and you can scale it up, depending on how much consumer you are serving. You can do that with some elastic compute so that could be variable. Cost the cost that that would be 0 if if you output level or 0. Right? You can. You can, because it's on demand. Right? You can scale it up. You can use more server. If you have a if you market expanded right? If you have a large consumer base. Okay? So that you call variable cost, and we can define average cost, which is the total cost. Okay, that is important is the total cost. Cq, plus. Q, okay. total cost include both fixed costs and variable cost. Okay, so every cost is you can is per unit cost, right? So the total cost divided by output cost. Sorry. The total cost divided by the output level. Okay, that is average cost another concept of marginal cost. This one is a little bit more tricky because we don't use this term in daily life, but but it's very important for economic analysis. So remember, in the previous lecture, when we talk about consumer choosing how many pizza to consume, we have the concept of marginal utility. Right? So when my utility from an additional piece of pizza right? Additional unit of pizza, so same thing for the supply side when firm, make decision. Marginal cost, as I will show, you will also be very important. Okay? Because marginal cost will drive certain decisions. Okay? And marginal cost is the unit cost of a small increase in output. or you can think of it as one unit increase of output. Okay, will cost me like 3 more. Then the marginal cost will be 20 per week. Let’s assume we’ll be looking at a time frame of one week. So 1 per hour on weekday, and our worker works 8 h per day for 5 days a week, the 5 weekdays. So that that is up to 40 h on the weekday right? And for the weekday we pay like a normal hour, 2, but if we ask our worker to work on Sunday, we have to pay even more like 20 fixed cost right for the week. But for normal weekday I have. I still have to pay 2 for 8 h for up to 8 h. Okay, but if we ask our worker to work on Sunday, we have to pay up to. We have to pay 20 right? We have to rent the machine for production. That is 20. Here, what is a variable cost? Well, we call our definition of variable cost. Variable cost is V is Cq, minus C, 0, right. cq, minus C, 0, c. 0, which is the cost of producing 40 units miners. The Ccu. The fixed cost, the fix are 1, that is 60, right? Because that because it’s 40 of variable cost. Right? So 20 is C, 0 and 40 is. see 40 minus c, 0, right? The variable cost of the 40 units of T-shirts. Okay, so that is that that means the average cost is 40, divided by 40 units and every cost 2 for the additional unit, because that is the wage for Sunday. Right? Okay? So that implies the marginal cost at 40 units to expand your production by one additional unit. You have to pay 2. Okay. so the marginal cost is 1, because at 20 units you can still ask your worker to produce one more unit without paying the Saturday wage, because still you can pay the weekday wage right? Which is 1 per hour. Okay. what about at output level of, let’s say 50, 15. Okay, that means you are on Sunday. You have to ask your worker to put to work on Sunday, right? Because for the whole weekday she will produce 40 units, and for the Sunday, Saturday she’ll produce additional 48 units right? So if you are at 50 unit, meaning that you have to ask the worker to work on Sunday. Right? So that is 3. Okay, so try to work out this offline by yourself. Now, we can visualize the marginal cost and the average cost by this graph. Okay? So first, st let’s look at the the blue segments. Those are marginal cost. and it’s very intuitive, because if I’m below 40, the quantity Q below Q. Below 40 means that my worker is still on the weekday. Right? So I do. The wage I pay is 1 per unit. But do I have any message? No, okay. But once I cross this 40 units, I need to ask my worker to work on Saturday. I have to pay 2. Suppose I exceed 48 right, and I have to ask my worker to work on Sunday, and I have to pay 3 per unit. So that is the meaning of the the blue line segment. Okay? And the the right line is the the red line is the average cost. Remember, the average cost is the total cost divided by the total quantity. Okay. and the total cost in the average cost initially decreasing because of well. because big house. The average cost take into account the fake cost right the machine. I pay 20 plus 40, divided by 40, that is, at output level of 40, and at the output level of 40. My every class is 1.5, right? So this is 1 point file. right? That is at the level of 40 output. Every cost is this point right? Okay? And if I expand my production a little bit more beyond 40 units marginally, I will pay 1.5. Actually, we’ll pay 1.8. Okay, for any number of T-shirt. So our T-shirt factory is a price taker. Okay, it takes a price as given. Okay? But in reality you could have firms bargain about the price level. Okay? And in our next lecture we’ll have a price generating mechanism. The price was generated by market competition. and in the second half we will have some other oligopoly competition that generate the price. Okay. But now for simplicity, we will focus on the T-shirt factory decision production decision. And without worrying about the how the price is generated, we’ll we will discuss the price generating mechanism later on. Now, pricing is at 1.8 per t-shirt. how should a factory produce. How? What’s the quantity? Optimal quantity? Well. should the factory increase output beyond 40 units. therefore, operating on Saturdays? Well, now, let’s look at this setup market price said, 2 margin cost is 1.8. That is, price given by the market and per unit. I’m paying 1.5 cost. That is my average cost. So I’m making money. Okay, I’m making money. But profit will be. what about if I increase my production? Well, the claim is that profit will be lower if we produce more, and reason is because price is smaller than marginal cost. because price at the 1.8 and per unit. My margin cost is 2. Because we discussed previously at level 40. If I want to produce one actual T-shirt, I have to ask my worker to work on Saturday, and I have to pay 2, but they are 1.3 in the previous slide. We have a price of 1.3? Well. So in this case. T-shirt may not be profitable anymore. So because we can, we can do the following computation, we can compare price, which 1.5. Right? Okay? Every cost is 1.3 is lower than my average cost. So this business is not profitable at all. Okay? So we consider 2 scenarios. The 1st scenario are using marginal cost to make a decision. and we use marginal cost to decide how much to produce, how many units of T-shirt we want to produce. Okay, that is conditional on the business, profitable right? That condition on the average, cost is lower than the market price, that is, conditional market price given at 1 per unit hourly wage right from 40 to 48. I asked my worker to work on Saturday I pay 3. That is, the marginal cost a 3 horizontal segments, the wage for weekday at 2 and Sunday at 1 and market price is 1.6 right. Wow. So that argument suggests, if I’m at here to the left of 40 to a level 40. My masterclass is always, why don’t and monthly price given at 1. I pay a weekday wage to my worker, and I claim 2 because I need to ask my worker to work on Saturday. I pay 4 right? Because I pay 1.6, so I should not expand further. Right? So that is also the insight we had previously, which says that I’m going to make my decision based on my marginal cost. Right? My marginal cost for this quantity. 40 to 48, 1, but beyond 40 units my market becomes becomes 3, well, now, weekday is profitable. Saturday is also profitable, so my supply is 48. So I will ask my worker to work on weekend plus Saturday. What if marketplace greater than what color greater than 1.5? That’s why you want to enter this business right to enter. You are given a market price higher than 1.5. That’s why you want to enter. Okay. Now, suppose you given any number between 1.5 and 2, say 1.7 market price is higher than the marginal cost of 1. Right? Okay? So I will. I’ll explain this again next week. Don’t worry the price goes up above 3, so go, let’s go back to our let’s go back to our which schedule here. If it’s about 3, it’s profitable or is good for you to ask your worker to work on Sunday as well. Right? So you will exhaust all the capacity you can have given this worker. and from 3. Yeah, from 3. What does that mean? Well, it means Sunday is not profitable, but Saturday is right, but the fact that Saturday is profitable. You will ask your worker to work on weekday, which is 40 h plus 8 h of Saturday. So that’s 48 h right? But you won’t. You won’t expand further, because Sunday is 3. Okay. okay, don’t worry, and I will. I will explain this again in our next lecture. It’s a little bit complex, because you are essentially doing a very rigorous proof. Using graph. I will stop recording now.

4. Price

Okay, thank you. Everyone for coming to the class today is our 4th lecture. And in the 1st lecture we discussed consumer preference which you define based on ordering of products, alternatives, or states. We introduced the indifference occur to describe the shape of behavior, consumer preference and consumer. Given some budget constraint and make a choice, or make trade off to maximize their utility functions. If we aggregate across, you know many, many consumers, we can have different preferences, and we can get an aggregate function or a curve. What we call demand curve, which describe the the shape or response of consumer behavior to the instrument, the policy makers or the firm has. typically, we discuss or we use price as example where consumer respond to price. And we care about a concept we call elasticity, which measure the sensitivity consumers are to the instrument, the firm or the policymaker has, and we also have a measure called consumer surplus, which quantify the happiness or pleasure consumer derive from the consumption. and after that we move to the supply side of the market, which is, you know, at the other side, in addition to the consumer side or the demand side, and on the supply side. firms are going to make decision. They are the decision makers and their decision variables can be. You know how many advertisements or what kind of inputs to use, how many people to hire, and what kind of product to produce, and how how much, how many of the product do I want to produce? And we will we use a very simple example of a T-shirt factory that is making decision about, you know, buying the the impulse right to the investment and decide whether or not you enter into the market, and producing T-shirt, and if if we indeed choose to enter the market, and how many T-shirts are the factory want to produce. And that was our last lecture, and we highlight the key insight of firms. Decision making. 1st of all, 1st decision problem has many similarities. Right? With the consumer decision problem, both face constraint to consumer. You know, you have a fixed budget. right? Fixed resources, and you want to consume different you know, combinations or different products or goods. But for firms they also have a constraint right? They have to borrow money from the the market capital market, and they have to hire people and pay the money, and they have a fixed budget frequently. For example, you want to achieve certain level of productivity. using as little money as possible. Right? Okay, there’s many similarities between the consumer side and the supply side in terms of the decision problem. Both are constrained constraint, optimization problem giving limited resource auto, make the decision to maximize either my utility or my profit. And we, on the firm side we introduce, you know. There’s a very simple problem. A T-shirt factory the T-shirt factory has to lease some equipment which is a fixed investment, right fixed cost, and has to decide how many units of T-shirt you want to produce to maximize profit right? And we highlight that the firm should use marginal cost to decide how much to produce and use average cost, to decide whether or not to enter into the market, or to consider this business at all. Okay? And this has some similarity, especially the marginal cost which you will, we will discuss again today in today’s lecture. They share some similarity with the consumer side problem. Where, when we use the piece of problem we, you know, discuss how the consumer making the consumption. Decision is actually also the marginal rate of substitution right along demand curve. For you know, when you consume the pizza, what is the additional or incremental utility from actual slice of pizza versus the price or the the cost for the actual slice of pizza, so that intuition still hold here in the supply side. And we sorry. Let me and the Mo. The most important thing we derive is something called a supply function, which describe a a price taking firm. Okay, the firm now is price taking okay. Later on we will actually allow Price to be chosen by the firm. And and after that, when we talk about competition, we use a perfectly competitive market, and that there you you will. You understand the price generating mechanism? Okay. so right after this discussion of the cost and supply function. I’m going to talk about actually, exactly pricing problem, how price is generated by a monopoly firm or firm that has some degree of market power. Okay, now, firm is price taking meaning that price is given by someone. It’s a government or the market. The firm does not control the price, so it only control whether or not I want to enter into this market. And how if I enter, and what is the quantity like my supply schedule. Okay, we call it supply function. So we discuss how to interpret this supply function, which is in the right color, right? It’s in the right fake line segment variable. And let’s just recall what this graph is about. So there are a few variables here when it’s average cost which we define previously is the total cost divided by the total quantity right. This is decreasing because by decreasing initially, because we have to pay a lump sum fixed cost in the beginning. So that makes the average cost very high. If I were to only produce like one unit or 2 units right? Because in that scenario, when I produce very little quantity. Q is small, the average cost taking into account that long, some very big fixed cost. Okay? So that that is why, when I just start increasing my output. Okay, my marginal cost is my average cost is decreasing. Be, that is because all my output queue is sharing the initial lump sum fixed cost. But you need I? I eventually will increase as some as I, exhausting my capacity. And in our T-shirt factory example. average cost increasing because of labor cost right, because the marginal cost of the hiring the water and asking my water to work weekend is more costly. So that is the second variable. Here we have called marginal cost. Okay, the margin request is a stamp function. and it has 3 levels in our toy model. The 1st level is. let me use a different color. Let me use this. Okay, the 1st Level is 1 1 meaning that I’m asking my worker to work on weekday right? The weekday rate, which is 1 per unit. Right? That is our definition of marginal cost, which is an incremental cost per unit. If I were to produce one more unit. how much do I need to? How much cost do I need? I need to pay. Okay? So the 1st level is one and second level of marginal cost is 1, that is simple price is below 1 to 1 right? If I ask my worker to work on weekday, but it’s below 1.5 and 1.5 is important. That is because 1.5 is the lowest point. Lowest point of my long run average cost right? So if the price market price you’re you’re giving me, you’re paying. My business is below my long run average cost. I’m not going to enter it. Okay? So that’s why the supply function or the quantity I want to supply. Given a price below 1.5 is 0, although it’s above one above my marginal cost. But it’s below my average cost. Okay, so I’m not going to enter. But something will happen when the market price you give me is higher than this lowest point of 1.5. Because that is an important point. Okay? Because we again in this decision problem, my entry decision, my entry decision depending on if your price is higher than or lower than my average cost. Now, if you increase your price just about 2. Right? And I can produce that 40 units. And that’s the best. Okay, that’s the best I can do, and the average cost I pay is 1.5 right? Which means that in the long run I’m going to cover my average cost. And we have the you know, the the math behind this. Why this decision is is optimal or is profitable. Right? Okay? So if the price is above 1.5 I decide to enter, but I will only ask my worker to work for 40 work to produce 40 units under the marginal cost of 2, which is higher than the market price you you give me right because the market price will give me here one to 2. So that is why the optimal supply is just to produce 40. Ask my worker to work on weekday produce 40, and I’m happy about that. I cannot do any better. Okay, now you can use this logic and proceed, and you will show that if the market price is 2 to 2. That is fine, because still lower, still lower than market price marginal cost of 3. So then, that’s well, the best scenario, and we, this market is so profitable that I will use all my capacity. I will ask my worker to work on the whole week. Right? Okay? Right? So I can ask my worker to to work the whole week, including Sunday. Okay? So that is the story behind our supply function. And I want to highlight. This is long rise supply function. It’s a decision made before I pay any fixed cost. So it’s I’m feeling facing this long run decision. So next I’m going to talk about slightly different timing of the decision, and the problem becomes whether or not I want to produce, you know. after I already invested my fixed cost, I already lease the equipment. So then the decision problem actually will have a different structure. And for some part the the your final decision will be the same. Optimization is same. But for some part it may differ. Okay. so suppose fixed cost has been paid for this particular week. Then after that the cost becomes signed. Cost. Same cost is the cost you already paid? Okay? And so now you cares about average variable cost instead of the so-called average cost. So every variable cost is the average cost, excluding the fixed cost because fixed costs already been paid. Even I don’t produce anything. I can now recover my fixed cost right fixed already been paid. You can think about us like in the Covid period. No, I already signed a lease of my restaurant for like 2 years. Right? So I already paid the money. Even if I’m I close my restaurant, I still need to pay the lease, pay the money, so that is like a short run that correspond to this short run supply function where I already paid a significant lump sum. Fixed cost. Right? So the show run supply function will be different. Let’s have a look. The logic is very similar. It’s just now I’m not using the average cost, which is this, the initially downward sloping and then increasing right. So every variable cost is flat in initially. Well, it’s because it’s because average variable cost here is constant right equals to the total viral cost, which is a labor cost in this particular example, right? Owning labor cost, which is 1 per unit. Okay? And the shape looks very similar. The only thing is there will be a different. There will be a market price that will generate different with short run response versus long run response. Okay, so the critical point, if you remember, is 1.5 previously, when we study the long run supply function. If the market price is between one and 1.5 we discussed, and we prove that it is not optimal for the firm to produce anything right. If you remember the right supply function which we’ve seen previously, let me try to go back right. This part. This is a long right supply function. This part, we argue that it’s not good for you to produce anything, because market prices below long run, average cost or average cost, which is 1 here. Okay, so then your supply function is. it will look the same as the long run supply function for higher price. But it’s just for this particular part it’s profitable to produce for the short run, but not profitable to produce for long run supply function. Okay? So the idea is very simple. So in the show, right? You can think about that. You are logging right? You’re lucky you are in a count, in a situation where you already pay, or your you know factory and large equipment. Right? You. You won’t take that into account in this decision horizon, right in this decision horizon. You will be motivated or guided by your average viable cost, because because some of the fixed costs at some cost. Right? But if you’re doing long term planning right, like the next financial year, right? Or the next 3 years. Right? So you will take into account all the fixed cost investments. So that is, that generates a different new supply function. Okay? So if we aggregate, you know, we get some very smooth line, and in a very general case. This is a visualization, so we still have every cost and every variable costs, and we have marginal cost right, and the idea is that the sh! The long rest of high function is starting from this right. This is the mean price for a long run. Average cost the mean, the mean market price that will convince me to supply anything is determined by the mean average cost. but in the short run this part I will still be happy to supply as long as it’s higher than the mean of my average barba cost. Okay. it’s useful to Apply what we learned to some more current situation of AI technology. Here there are multiple perspective. You can take and can apply our concept. I will take a particular one where we are thinking about those company decision. Okay, we’re thinking about whether or not you want to. Enter into this market. Okay? Via AI AI platform. AI company. Okay, in particular, be like A, you know. Ai Service provider like Open AI, okay, like, develop models like Chatgpt. But there are alternative perspective. You can take where you are. Customer of open AI. You are, you are, you know, thinking about investing or entering into smaller segment of the market. Right? You can you? Downstream? You you use chat, Gpt service right? And you’re a small service provider to some particular segment. Okay? Then the calculation can be different. For example, fixed costs will be very different, right? Because if you’re a small service provider developer that you use open AI service right? You pay well, you may pay by token. and you actually will not have a that high of a fixed cost, right? And the idea that those companies are so, you know, impactful is because they can translate all the problem they can. They can host so many small business, and they will bear all the fixed costs. So it’s very easy to attract developers and turn into this platform. But imagine, if you are this AI giant. So how is the investment problem different. Well, so AI technology will evolve, it will involve massive Iid for model training. Right? You have to purchase Gpus, right? Which can be very expensive. And also the model has, you know, return of scale right? So frequently you have to purchase many, many gpu, which is, which is going to be a very, very large investment comparing to other type of technology, right? And you also need engineering talent. Well, I’m not completely, you know, familiar or expert on the production function for these AI firms. But you can think of a scenario where a few, a handful of very talented people can make a huge difference. So the production function is very, very different from many of the traditional sector, where numbers matter matters much potential matters much more. But I think we are still learning about the production function. Right? So you you probably read about like recent case of some other AI platforms like Deepseek, and so on. Which really question the the exact, the production function of this? This platform, right? So, which inputs are generating a higher marginal return right? Exactly how costly it is. Actually, I think we don’t. We are not 100% sure about that. That can change a lot by technology. By our, you know, experimentations? Right? So it’s totally under some uncertainty. And that’s why you, you observe, like a huge response to the stock market. Once people learned about like someone claimed they can develop certain product right? So using our terminology, they can produce a certain level of productivity with a very low cost, right? So that that’s going to be tremendous, tremendous, and has a huge impact. But for this platform the send the cost. There are large, very large parties sank like Gpu, right? You have to invest a lot. And and you also hire people. But you know. But when you when you produce like a T-shirt factory. Right? When you produce. like output, right? It. The marginal cost becomes very, very low, right? Comparing to the initial investment model training and development and and equipment right? And for for traditional like, I call it like intelligence provider, like traditional human labor. the production function looks very different, right? So marginal cost. Imagine if I’m a content, I’m a like a advertisement company, right? I want to make videos. And I need to hire people right? So. And each additional people is like additional labor contract, right? And I need to pay them by hours and so on. Gonna be the marginal cost gonna be very expensive, but the fixed cost is not that high right? Especially if you hire people like on a temporary base, right by a particular contract on particular task right. But still the the marginal cost is going to be much higher for human right from having for a very large set of tasks compared to. Let’s see Gpt. And you can also make a comparison between like a search technology versus this AI technology and the AI technology can, you know can be applied to a larger set of task. Right? Maybe the marginal cost is at current state slightly higher than say, Google, search right? Because you can see that Google search is still free. But Chat Gpt and the other AI platform for the advanced models you need to pay. That is because marginal cost right? It’s not completely free. Or it’s not as costly as Google. Search right? Each Google search. I think my guess is probably a fraction of cents. Right? Very, very cheap. And well. later, we can talk about how? Why? Why? Is price at 0? Right? Because Google is a multi-sided platform. Okay? Because you can profit from advertisement, get paid by by a company who need attention. Right? So for me as a consumer when I search, I don’t need to pay anything. It’s also because the marginal cost is very low for search, right? But at current day many of us you pay for for for life model because of marginal cost. Okay? And okay, I think, but I have to say that is evolving. It’s still evolving stage. We don’t know. Because the technology is going to change very dramatically over the course of time. And we, we are yet to see what happens. And you can do the computations, and it’s very simple so every cost is total cost, divided by the quantity and given the scale of the large platforms like. Gpt, right, you can check the Internet, and how many tokens they use per per month, or how many queries they process the number. The queue will be super large, right? But still I I mean, the average cost as of today is still not as low as I mean. as I say, search technology or other technology. But it’s lower, I think. Especially I I mean, marginal cost is definitely lower than the human labor. For right? For example, when we do copywriting right? A copywriter and all we do like making slides right? Or do some like. Not that sophisticated task. or do some data cleaning right? So that is much cheaper. The marginal cost is much, much lower for Gpt or for this AI models comparing to human labor. And also, I think, one reason that is getting so much of hope or potential economic impact is because it’s it’s being we expect to apply this technology to a much larger set of tasks or problems than comparing to a search. Right searches only has only been applied to a given set of problem. But now we are thinking that this AI had will have a chance to be applied to a much larger set of of task. But at the end, what determine the price? Well will be influenced by the marginal cost right? So per unit, if it gets as cheap as like a search right? Then the price then you can expect it could happen. What could happen is the the model business model would be sort of similar to to Google. But I I but I I cannot see that, because there’s just so many uncertainty right? And firms when we decide whether or not to enter the the the industry, I will use marginal cost to decide the quantity right? How much I want to produce, whether or not. And I will talk about the the in our next set of slides. Compare the marginal cost with a marginal revenue. Okay? And this allows the AI platform to scale output very quickly at small. I wouldn’t. Yeah. So you’re still decreasing per unit cost right? And we also have this very high elasticity of supply because it’s digitize right? Because as long as you have the need. So I mean in a given time window. It’s not like you can scale it up, always can scale up in in 1 h or so. Maybe most of the time they can, but sometimes you, if you give them enough time, you can always get scale up the compute or the the service by a lot, so it makes it very hard for a second company or for other company to enter and compete right, because a new company. when you just start a business you need to take into what? Well, before you start the business you need to take into account average cost, right? Average cost include the fixed cost right, a lot of investment on the compute right or the initial investment. But the the incumbent we call it incumbent, or the the dominant current. Firm, right? The leader. Like open AI, they have scale. They are benefit more from the scale, right? Because more people using their platform right? So their investment get covered better than the followers. Right? Because open. Yeah, we have a lower average cost, right? Because of quantity effect. Because you have, you are early mover right? 1st mover, and you attract a lot of demand right? And so that will at the end of at the end of day affect the market structure. And it’s still very early to see now that you already can see. From, you know, some like stock price for Nvidia, for example. Right? Nvidia is like. If you buy their stock you are Well, if you bought early, then you you are pretty lucky, right? That is because, you know, Nvidia has its 1st or earlier mover advantage once the market start moving, you know, it’s hard for the follower to catch up because of. you know other firms. They don’t have the they do. They cannot wait right? They have to buy. And you know, competitors. They are not. there is no alternatives yet. Right so. And also, you know, if you are 1st mover you are, you can afford more investment right in R&D, and you’re hiring the best people invest more on technology. Right? So that is very beneficial. And that’s why you know, Nvidia, to some extent is being regarded by the market as a potential to have large market power. Right? Because of. we talk about the elasticity right? Because there’s not so many alternatives. So demand is not very last to the price, and you have a chance of to be a very you know to have not be power. Okay? So that eventually will translate into market concentration and market power. And I think you you are in, you know, school computing. So you know the frontier of innovation. So when you innovate, we’ll talk about in the second half. Your goal is to be like a monopoly be a monopoly right to be the 1st one, and you will enjoy a larger slice, much larger slice of pie, especially given the network effect. Okay. which will make your marginal cost every cost very low. And then, you know, it’s very hard, very hard, to to beat. Okay. okay, scale economy and potential monopoly. I think. Last lecture we talked about the case the interesting case where Google paid a lot paid like 20 billion to apple to be a default search engine. It’ll reveal a lot about the market. Okay, about market structure and profit. Right? So there are a few lesson lessons we can learn from this case first.st Well, Google is able to pay like 20 billion to apple right? That means that implies that being the default search engine is really profitable, right? Otherwise, why do I pay it? Right? I would. If I’m Google, CEO, I would just say that. Forget about it. I don’t care. You don’t need to put me as a default search engine, right? But the fact Google is able to pay so much meaning that being a default search engine is very profitable. right? It’s so profit that that it’s like definitely over 20 billions, right? Because remember, review preference if you’re able, if you’re willing to pay over 20 billion. That means you’re making much more than at least more than 20 billions right from being the default search engine. It also tells us the market power of apple right? Because Apple is apple is now like investing a lot for Google, right? Like working very high for Google to be a search engine, right? Because the margin cost of doing that is very low. So Google is just able to convince Google, sorry, apple is just able to convince Google to pay 20 billion, you know, without doing anything right? Just just without really incurring much of a cost. Right? So that says a lot about Apple’s market power, right? That is because you know, apple is a gatekeeper, right? So the concept of gatekeeper Apple is a gatekeeper of this ecosystem, which is very large in terms of economic value. Okay? So I I think in today, there’s lots lots of discussion on the antitrust domain as well. Where, this technology because of the huge economic impact huge, you know network effect a scale economy. It’s likely we’ll generate a lot, you know, a new like Microsoft or Apple. Right? Nvidia is looking like that. So I think our but our our understanding is really not that mature. And that makes sense, because the market, because no one knows what will actually happen, because the under our understanding about technology and the the market is still evolving. They’re constantly right. If you look at the price stock price of Nvidia, you you understand that constantly, people learn new information, form new belief about this market. But the fact that the value the market value is so fragile or so elastic to the new event tells you about the scale of the of the impact, right? So like a fraction of number like of the marginal cost of the can make a huge difference when you scale it up by by billions of people. Right? So, okay. that is a little bit of a broad discussion, but I think our. I always try to summarize my lecture with very few sentences. As long as you understand those sentences you are doing very well in the class. So our. our, our takeaway for the cost of production function is basically you want to use marginal cost to decide how much to produce. Okay, a marginal cost will affect how much you produce, and it will talk about pricing, because there is a fixed relation, one to one relation between price and quantity. Right? So equivalently, you can see that marginal will decide your price. That is, our discussion next. Okay, and every class will decide your entry decision. Okay, so and marginal cost for marginal decision. So that is the same story for a consumer decision. Okay. that that’s for a cost function, cost and supply function. And I will move on to our second part of today’s lecture, which is about pricing. So remember, in the T-shirt factory. our firm is price taking right? So they only control the quantity and price given by the markets. Now we’re talking about a more realistic setting where there is a demand function. Okay? And there is a relationship between price and quantity. That is what we developed in our 1st in our lecture 2 and lecture 3, right? So there’s a relationship between price and quantity. Okay, if I set a higher price, I will expect a lower sales or demand right. If I lower my price, I will expect my demand to increase. So that relation is given. Okay? And we are going to talk about this. The implication of this demand function in my pricing decision and pricing, you can think about it as the most important strategic decision firm can make. We will talk about. So in this lecture, the firm you can call it has some market power because it can influence the market price right? Because you can decide the quantity. And and you know, and or you can think about a setting the price that is like in the mo in most of the real worst world scenario. That’s the case, right even for our let’s see the regular store, right? You see, on the market, on the in the Mall. They all are based on a price posting mechanism where the seller post a price and expect a demand right? And at the end of today’s lecture, not the end of the slides I will have more sophisticated or complicated pricing mechanism, including, for example, auction. Right? So auction is very different, because auction is not me as a seller post any price right? Auction is, I ask my buyer to claim or to be how much she want to pay me. Right, that is option, that is, in the mechanism mechanism, design. Literature is very sort of different from pricing where the seller post price. But pricing is more regular. I say in also in e-commerce. Right? so I will post I will have some actual discussion in the end. But I will. First.st I will focus on the key methodology of pricing problem. And then this is built based on our understanding of demand. Okay, so firm face a trade off between price and quantity in, you know I no matter your e-commerce company, right? Or you are AI company like you are, check Gdp, or set price for your token right? And there will be a there will be a function between your price and the and the quantity right of sales. Okay, to sell more, they must charge less right, and the 2 have different impact on your profit. Sell more means, a larger quantity, make potentially larger profit, but charge less meaning lower margin. Right? I would define margin in today’s lecture. So what price should should the firm set right? So that involves a trade off? Okay. So should they simply apply a standard markup to cost. Actually, a lot of retailers, especially in the traditional business. They still do that right. So they have an understanding of the cost average cost marginal cost. Right? So what they do is they just multiply it by certain percentage, right? Like 20%. I want to earn 20%. So I’ve just used my cost information. And I’ve played and multiply by some percentage. which I think is which I think makes sense. But that’s not what especially digital firms does today. Because I see Amazon, or in some Chinese e-commerce company like Jd, what they can do is they can. They have a large database of sales, right? The prices, and they can do something called demand estimation, and they can estimate the demand. Elasticity. and the message we have today is that the optimal price? Okay, that maximize your profit will actually depend on demand elasticity. So what what does that mean? Well, it means that a constant multiplier to your cost right? This simple, like very classic old school way of selling price is not optimal, so optimum price should take into account how elastic your consumer to your price. And in today’s digital platform digital company, they have a large amount of data. What you can do is, you can estimate, and they can treat it as a very quantitative decision, and you can set a price that, to be very exact and maximize their price. You don’t need to multiply the same scalar to all the cost they have. Right. That is not going to be optimal. different price. Different product will have different elasticity, and that imply different price. Different optimal price. Okay? And at the end of the optimal price is a trade-off between margin and margin is per unit profit and the quantity sold. Okay, and we’ll prove that the price will be determined by margin cost divided by a function of demand elasticity. Okay, this is called the Elasticity a room. Okay? Okay? And and we’ll use a toy model to introduce this idea. Okay, I I imagine we’re talking about ice cream, a pricing of ice cream. Okay? So I have to run a chart. And I have to buy ice cream from upstream. Okay? And I pay a wholesale price so that the wholesale price actually my cost right as a retailer, as a retailer of ice cream. That is my marginal cost. Suppose I pay my upstream 10 is too high price, no one want to pay 5 of price, and I’ll get one unit. I expect to get one unit of demand. Okay, so there’ll be one buyer because there’s 1 student or who want to pay like who who is willing to pay 5 more. I will get one one more unit. Okay, so you can, you can, you can check that. The relationship is determined by this this global parameter. so that that sort of will determine the conversion rate of price and quantity. Okay. if I decrease my price by a lot to 15 for the rental right. Remember, the rental is 3 per unit, right? Because I have to pay the wholesaler for 9 right, and I make 8.5. So I make 24 of total cost. And I make 6.5, but if I increase my price a little bit further to 6.5. I decrease my price to a certain level of 3, right? Because my incremental cost here is my marginal cost. and which is constant because the the wholesaler doesn’t give me any discount. Okay? So I just always pay 6.5 right? Something important happened. And this level at this level. If I were to increase my price a little bit a decrease my price a little bit to 2.5 right, and my incremental cost is 3,000. So what do we learn from previous a discussion? Well, that means that’s a bad decision. Don’t want to do that right, because that decision leads to a loss. because marginal return is 2.5 and marginal cost of 7, what happens? Should I? Should I decrease my price a little bit? The answer is, yes, because I’m using. Let me use a different color. Okay, if I’m at 5 to 6.5. The reason is because the incremental revenue is 3. So this is a good deal. This is a good decision, because the marginal cost of that decision 6.5 generates the highest profit is higher than a lower price of 9 profit, and also higher than a higher price of 9 profit. Right? You can see that the optimal decision is defined by the marginal trade-off. Should I be so when you make the decision you think about locally. Divide right locally. If I cut my price a little bit more, what is the profit incremental profit? What is an incremental cost right incremental meaning a marginal. Okay, we use it interchangeably here. Okay. so since there is a 1 to one correspondence, I’m going to be more general and use some equation, and to to derive the exact formula of our elasticity rule. And but recall there’s a 1 to one correspondence between price and demand which is defined by the demand function or demand curve right, which in this case a linear function or Pnq. so there are 2 ways to model this problem. One is to decide the price, the other is to decide the quantity right? Because they have a 1 to one relation. So it doesn’t really matter. So I’m going to use the quantity as a control variable. Okay? As a policy or strength strategic, variable of going to decide the the output. So this face better with our previous discussion in the T-shirt factory. Also talking about deciding quantity. Okay. I did that. And I want to highlight. It’s not always the case. Right? If you are data scientist deciding optimal price for Amazon, or you are revenue manager for airline, you may decide you may use price as a as a control variable. But given that demand function is monotonic. Most of the time doesn’t really matter which way you use as a decision variable. Okay? So if slope is positive, then higher output leads to higher profits. Okay? And if slope is negative, then lower output leads to a higher profit. That is the review schedule we discussed right and and back here profit is a functional quantity, because we use quantity as a it’s agent, variable, right? So this is profit initially increase and then decrease. Right? So it’s a u shape as a function of quantity. Right? Quantity is increasing. Okay, when quantity increase profit initially increase, then decrease. So there is a optimal solution to the output level. And if we learn calculus, and if the profit functions will behave, let’s say it’s quadratic. Then we can use some, you know, calculus to to derive the optimal condition. Okay? And if it’s quadratic, actually, not only necessary, but also sufficient condition for optimal pricing. Okay, so this is the visualization profit profit a convention is to use the letter pop. Okay, as a function of quantity. Q. Right? So the axis Q. And why is the prophet? If you change the level of output queue right? You can see that initially, profit is increasing. the slope is greater than 0 right, and there will be a point where the slope equal to 0 right when you change output locally, the profit doesn’t change. So in calculus. We know that if this function is quadratic or convex, right? So this point is the optimal point of profit. So the level of quantity is optimal quantity, right? And it defines the optimal price. As a result, if you move beyond the optimal quantity pure like, let me call it pure star, which is defined by the 1st order condition equal to 0. Okay? And the derivative becomes negative. So the profit starts decreasing. Very simple visualization of the intuition right? For now we’re going to be a little bit. Use some algebra. Okay? Instead of this visualization in our earlier lecture, we talked about the elasticities right? And there we don’t take into account cost. Now we have everything. our table, everything ready. We already set up a demand function which defines the revenue a price, times, quantity. P. Times, Q. Right? And we also define total cost. So we can write the profit function. Okay, the profit function is pi, q pi is the notation we use for profit as a function of quantity equals to revenue right. Given the quantity I choose, minus the total cost conditional on the quantity I choose. Right. And a second equation is the marginal profit. We are taking a derivative. My total profit with respect to the quantity. So marginal quantity impact impact my marginal profit. Okay? And you can apply the chain rule. And you know, because it’s just 2 terms, and you can apply them separately taking derivative of quantity for the routing function and for the total cost function. Okay? And the and you, you can define this, and it’s very intuitive. I think you already learned this. The marginal revenue is the revenue with respect to the quantity. Right? So this is the 1st term here. This is my margin revenue. Okay, and this is my marginal cost, which is my total cost. With respect to quantity, I think we define this term explicitly in our cost function lecture, right? So the marginal profit in children is the marginal revenue minus marginal cost. So now we put everything together right? Because we already be has been using this note. Intuition. The decision of the optimal decision decision depends on the relationship between marginal revenue and marginal cost. Right? okay. So if you recall what it had in the ice cream example, right here, the marginal, we are talking about relation between an incremental which is marginal revenue and a marginal cost. Okay. so the claim is that in this visualization profit maximization defined by the defined by the marginal profit equal to 0, which means the marginal, which means this term equal to 0 right. this equal to 0 right? So that defines the optimal quantity, and that means the 2 term has to be equal right. The marginal revenue has to equal the marginal cost. This is the formal definition of this association problem. It’s very simple. But it’s, I think, if super important in all the decision problem. No matter. It’s a firm, firm decision. It’s an automatic decision, or it’s a consumer decision. Right? We call the pizza example. When you eat pizza, when you make your own decision right? You go into the go to a dining hall, decide how much you want to eat. Right, you will stop. When do you stop? Stop when the marginal return? It’s greater than it cost you, or approach a marginal cost. Okay, how many pizza you want to order from? That is also how you make it. Make your own decision. It’s very simple, but it’s a very powerful rule. Okay, it’s so powerful that I use this very crazy funds. Margin revenue equals to marginal cost. Okay? And what do you get from selling an actual unit. That is the definition of marginal revenue. Okay, so you got the price for which you sell it. But the additional marginal revenue is less than that. Okay? Why. let me show it precisely. Why? Well, suppose we are talking about, you know, like decreasing the price. Let me check if we are talking about decreasing price, or, okay, let’s just see how it goes. So marginal revenue we call it is the derivative of our quantity, right? For the revenue function, and the revenue function is price times. Q. Right? So when you change quantity, what happens to price? Times. Q. Well, there are 2 effects. Right? Apply the chain rule. There are 2 effects. The 1st is the price effect. Okay, the current level of price and the second is quantity effect. Okay. Q. Times, Dq, and Dp, okay, suppose, let’s see this term. Okay, this term is smaller than 0. Why is the case? This term is smaller than 0. Because if you want to sell more right. if you want to increase quantity, suppose quantity increase what happened to price. because this this term, this derivative. This slope is negative, right? The relation between price and the quantity is negative. If I want to suppose farm is making up quantity decision for whatever reason. Okay? So if I want to sell a little bit more, I have to decrease my price right? That is defined by my demand function. We are talking about a firm that is able to influence the market price that is decided or determined by the demand function. And that’s why Dp, Dq. Is more than 0. Right? So I can decompose my marginal revenue to 2 part by wise price. The other is quantity. Okay? And because this slope, parameter slope parameter is negative. So my margin revenue is always smaller than my current level price. Right? Okay? And the intuition is here because price must be lower in order for actual unit to be sold. Okay, this lowers the marginal on all units sold. Right? Okay? Because when you cut your price. you are, we’re talking about uniform pricing case different from, let’s say you can set price for all the consumers, for set different price for different consumers. Right? In that case you can you feel to send extra unit? You go to a consumer and talk to her and ask, or you negotiate right that. But now we’re talking about from that? Is it? That can only set one price to the whole market right? So if I do sell more, I have to revise my price to all the consumers. There is an additional impact, right? So that is the additional impact on the the marginal revenue when talking about changing price or set said, selling after unit. Okay. and marginal revenue. A little bit of algebra marginal revenue equals to this term. And I’m trying to link this term. We had in our previous slides, which is price plus this derivative times quantity. Right? Okay? And I try to link with elasticity. How do I do that? Well, so what is this term? This is Dpdq, right? Dpdq, and we can. divided, divided by by by P, and multiply by P. So it doesn’t change anything right. And look at this term. Well, it’s still not very intuitive. But if you move out to the denominator, what do you get? You get elasticity right? So this is the elasticity. the Dqdp, right? This is Dqdp. You just remove this term. You just remove this term to the denominator. This is what you get is, you can recognize this app flow. Right? Okay? And you do some rearrangement, and you will realize that the marginal revenue equals to price times one plus one plus inverse of absolute. Okay. It’s very simple algebra like high school, or you know, but with some, you know, with this derivative ratio, it’s a little bit unfamiliar, maybe, to some of you. But it’s right, it’s very simple. Okay? So therefore the margin revenue equal to marginal cost implies. Well, this is a marginal revenue. We just derived right equal to marginal cost, and you can define a function between price and marginal cost. So that is the elasticity formula in determine optimal price. Okay, so we introduce right? Or we promise in the beginning of today’s lecture. which is a relation between price and marginal cost right? Called the Elasticity group. Okay? Elasticity. Okay. so go go back to our introduction. And we said that you know the optimal price would depend on the elasticity. Right? It is not a constant multiplier to my marginal cost. Well, it depends on my margin cost. It is a multiplier cost, but it depends on elasticity. Right? Okay? So the optimal price equals to marginal cost, divided by one plus inverse of electricity, a little bit complicated looking. But idea is very simple, the optimal price in our in our toy model, very simple single product, optimal pricing problem, determined by true viable with a given functional form. So 2 vibrant marginal costs, and then the elasticity right? So you can immediately realize if a marginal cost increase, optimal price will increase right? So that is determined by the nominator here. and also depends on how elastic the demand curve the firm face. Right? Okay. And there are alternative arrangement. You can write it on other way. Price my minus marginal cost, because minus G times, one over epsilon or simply m equals to P. Minus mc. Divided by P. This M is margin. Okay, that I use this term a lot. It’s like a margin. Okay, I think I have this in the next slide margin next term. Okay. is P. Minus Mc. P. Is the price I charge right? Minus mc, the. that’s the actual money I charge. Right? So all the price I charge above my marginal cost. That’s all the money, the per unit profit I take right. divided by price, divided my price, the proportion of my price that is above my marginal cost, right proportion of my price. That is above the marginal cost. That is my margin. Okay? And you can see a relation between my margin and elasticity, because elasticity is negative, right? So minus epsilon is positive. And if you, if you, when we talk about, if absolute, is very big in absolute value. Okay? And what what does that mean? Well, that means, hmm. absolutely is very big in absolute value. Minus epsilon is positive, right? Because absolute is negative. So that means each one will be very small. Right? Imagine absolute minus 10. Right? So this will be one over 10, right? So that’s a small number so that means when demand is very elastic, like absolute equal to minus 10. The margin is small, right? But when right? So that is the relation here. You can visualize it here. This is a marginal cost, right? It’s a marginal cost. And this is the optimal price. Peacetime. Q, stop. Okay, and the margin is the proportion like the P. This P. Is well above the marginal cost. You can compare with the later case where P is closer to a marginal cost. Right? And you can check that elasticity here is more, is more or less elastic here. Well, because the demand curve the right curve. Here is price versus quantity. Right? So how do we interpret this? So you change your price. Decrease your price by a little bit. Quantity doesn’t change much here right along the demand curve. So it’s not very elastic here. not very elastic, right? But if you look at this at this level price, if I change my price a little bit. What happened demand is, very lastly, because quantity will change a lot right? We can draw some lines to help us visualize right? If I decrease my price, I’m drawing a lower line and correspond quantities. Here. Is he? Right? So a small change in price. We’ll change my demand by a lot. That’s what I mean by, you know. Demand is very lasting here. Right? Okay. But you go to the previous slide here. Demand is less elite, less elastic. Right? You can draw the line if I decrease my price by a little bit. Quantity doesn’t change much right? Comparing to a lower level. Right? Okay? So this is elasticity rule. It defines a real the margin I can charge above my marginal cost. Right? So the 1st M is the margin. The per unit price minus market cost the margin or the per unit. Profit. Divided price is the share of price that is above my marginal cost right? So that is something like I click as profit right? Like my profitability, something use. Very vague or very loosely speaking. And actually, what I’m more more useful concept called markup. You can find in the textbook. This markup is very frequently used especially by competition policy makers, which are interesting. Measure how competitive market is, so they will use this term second term. So this is more managerial. If you’re a manager, right? You’re looking within your company, you can. All you, we, you compare across different business. You can. What you can do is you can. sort of have a measure on the market power of like, how? What proportion of your price is about marginal cost? Or is is your problem right? But from competition, maker perspective, I’m going to use more objective measure. I’m going to normalize my marginal cost. Okay? Because marginal cost is more objective. It doesn’t depends on what firm does right? But I’m going to. Just okay. I’m going to compute the difference between price and marginal cost, right? And I I’m going to divide it by marginal cost, and that is, a concept we call markup. Okay, this is frequently used by policy makers. Okay, to measure how competitive your market is. If the markup is very high. What does that mean? That means the firm has market power. It’s able to charge a very high price comparing to marginal cost. Right? So recall the example. And I. We discussed the search engine, the default search engine. So the idea or the claim we made is that apple has a very large market power. Why? Because the because letting Google be a default search engine is not very costly. Right? So marginal cost is what it’s very low, right. It takes some of you like a few hours to set up things right, or even right. very small fraction of time, right? So set Google as a default engine in Ios, right? So the margin cost is very low. But what happens? Well, Apple is able to charge 20 billion. So that is very dramatic right? So that is a measure of markup or mighty power. So because, you know, if you are, if you own the whole ecosystem, you are the gatekeeper. Right? So you really dominate. And Google has to follow. You has to agree with what you ask, right? Otherwise. that’s the definition. Or that’s the concept of monopoly power. Right? So there are 2 aspects. The way is from the firm perspective. If you are entrepreneur, I would say, you want to be a monopoly right? Because that implies more profit, lot, larger market power. Right? You can charge very high price comparing to your marginal cost. So you are making profit. Right? So entrepreneur as a yeah. If you want to enter, enter this startup right? You want to start your own business. I wish you have a large market power. You have a can charge a high markup right or be a monopoly. But from policy maker perspective. it’s different. Okay? And we talked about the concept of antitrust. Right? So that is a term that used by the government. But it’s not. It’s very much hated by the company like Google, Apple Microsoft. They don’t like being called monopoly. Actually. in the internal like, there’s emails like among Google, like managers, leaders. They don’t talk. They don’t refer themselves as monopoly internally, and they hate that term because that term is not good. because policymaker don’t like monopoly right, because if you are monopoly. you will be able to charge very high price right and high price will imply. We’ll talk about that later, because right now we don’t have a aggregate measure of social welfare. Right? We had a measure of consumer welfare. And now we enter into the world of the firm decision problem, and later, when we put together everything, we have a concept called aggregate welfare, which is some of consumer welfare and firmware fan and a higher price. Okay, large market power on the firm side is a negative signal to the policymaker, because that implies firms are charging very high price that implies consumer surplus is lower right? So that is a market distortion, or that we will introduce a concept called deadly loss. Some some loss of welfare will result from firms market power. Okay? So it’s a mixed message depending on which side you are on. You are on the firm side that you lack monopoly power. That means profit. Right means rent economic rent, higher productivity because a little bit on marginal cost can generate much higher profit right? But from the policy maker perspective we also cares about consumer or from a consumer perspective. We don’t like market power. Right? Imagine, imagine open AI is the only platform, right? AI platform. There’s no other option. Then we are heading to a scenario of monopoly. Right? So so I think we can use Nvidia as example, right? So I think there’s recent criticism from the EU policy makers also from China. I guess some other Asian countries like. you know, Nvidia is on the on the list of a suspect of monopoly power, because they are the one of the only supplier of Gpu that is. circulating or available in the market. Right? What does that mean. That means Nvidia is able to charge a very high price. Have a high markup, not big market power, right? So whatever it charges, charges, right firms are willing to pay. That is because there’s no other option. Right? So that is market power. Okay. But I think in the end of this set of slides. I will talk about other considerations like Nvidia. It does not, you know, not maximize the short run profit function. And I want to put some context into our discussion. We are talking about short run profit. Right? P. Times. Q. That is the objective function we define for firm. That is simple, that is elegant, that is, capture the key intuition, but in reality firm, may maximize a long term, profit right? That is my current. P. Times, Q. Plus my future. P. Times, Q. And that will introduce more comp complexity of dynamic pricing because there’s network effect, right? There’s switching cost network effect scale economy. and also so consideration that may change from the incentive pricing. But I I put that at the end of the the slides. I think my time is up for today just to wrap up our discussion for today. we, we defined a very simple objective function for firm, which is to maximize the profit. Okay? And that function and that objective function can be solved. and it leads your decision that is determined by the marginal revenue and the marginal cost right? And you can actually solve for the decision problem. It at the end of day depends on the optimal price depending on depend on marginal cost and elasticity. Okay, depending on how elastic consumers are. The demand is and depends on my marginal cost. Okay? And we define content for margin, which, from firm perspective measures profitability. Right like the how much of my price is is actually above. From my marginal cost, like the rent I get, and also like a measure of market power, called markup. Okay. Firms. Ability to charge a high price comparing to its marginal cost. Okay, in the next lecture I will 1st revisit this part. And I will reapply this formula to our ice cream and solve for the optimal price, and show you that it’s the same as our simple excel sheet which you’ve seen before. Okay, I encourage you to work out this by yourself. Yeah, I mean, and I think, I will stop here. Okay, and we’ll meet in our next lecture next week. Okay.

5. Price and Competition

Okay, let’s get started today is our 5th lecture. We have one more lecture before we have the I think recess week or midterm break right. and our exam. Will be the 1st lecture after recess week so that will be Friday, March 7, th and the venue will be Mpsh 5. It’s a very large place to hold the more than 200 students. and so the venue is reserved for us from 3, 30 to 6 30. Our exam actually will take 90 to 100 min. It depends on, you know, the the length of question. So there’s still 10 min uncertainty, and then and I will start the exam at 4 20 to get some. I know some of you may need to travel from other venue. So I will give like, 20 min for. But just make sure you don’t deliberately be late. Okay? Because that will have some problems. Okay? The exam. Midterm takes a significant portion of this course assessment. So make sure you you are there on time. Okay? And and so the assessment mode will be hard copy, as you will have exam papers. All the question like we discussed will be Mcq, okay? And you will enter your answers. Record your final answers on Ocr form, which will be provided to you before the exam. At the exact venue, and you have to prepare your own pencil and eraser for the Ocr form. Because of the size of the class we would not be able to supply to all of you. and the exam. Content will cover our 1st half demand elasticity cost which we already cover pricing, which will be finishing today and competition. Okay, you can. Already you already can see all the slides in the folder progress and the primary resources will be lecture slides and the assignment questions. Okay? And and I will not provide any additional sample question or practice question, because I believe, is already a very large set of question at the end, you know. Exam, is we would talk about game theory, or we’ll cover the concept of equilibrium exact. Not single single agent. because I can give you many, many practice questions at the end, you know. You will make everyone’s welfare lower. Why? Because you personally will benefit from more practice. You feel like you are more confident. But everyone else feel the same way. So everyone practice 1,000 questions. Your average is still will be, I say, 80%. Right? So I just have to make the exam actually difficult, more difficult, because at the end I have to make the average at certain level. So. So that is a commitment. I’m not going to give you further assignment. Further practice questions. I think lecture slides, indeed, is a very useful reference. Because, I think most of the questions you can find related to our lecture content, or the somewhat similar to the assignment question. Okay. close book like we discussed. you are not allowed to have a either a cheat sheet or any other materials. It’s okay to have some very basic calculator, if you need. But I I will design the question in a way that really doesn’t demand. Large calculations will be very basic calculations. Okay. and I will collect back the exam papers after the exam. So this is for the midterm. so that will be on March 7.th be very careful and pay attention to the time and the venue and time. I think not that difficult, because it’s our lecture time, but the venue is different. We are not going to have a task here, because it’s to a small place cannot hold all of us. Okay, remind your friend as well. Okay. if we come here. You know, then no one will be here. Okay. that’s also a reason I want you to be early like, don’t wait until 4 20. If you come early, even you come to the wrong place, you can. Still, you know you, you will learn that. Realize you come to the wrong place sooner, and you still can still go to the right place. Okay, so this is for midterm. And our sign of why is due today. Actually, a lot of you already submitted your answers. And I think I mentioned in last lecture. There are 3 R. Temps. But the last of time counts. Okay, the 3 h time is only for the purpose. Some of you may mistakenly submit. Okay, then you have a chance to regret. Okay, and it doesn’t make sense to allow you to try 3 times and choose a maximum. It doesn’t make any sense right? So I can allow you to try 10 times and choose a maximum. What does that mean? Right? So, or you 100 times, then you just run program and simulate right? Of okay, that is for assignment. Okay, and for discussion. I’m I’m happy that there are lots of discussion going on, I think. Last lecture also I received some feedback. I will. I already encourage our ta to be more to put more attention on the discussion forum, and if there’s any special question you need particular attention, you feel is not fully assessed. Okay, because that is like a judgment call right. Our Ta may may not feel that way. But if you feel like you need more, you can just email our ta directly you can. CC, me, and you just say that I’m not sure if this question is fully addressed on the discussion forum. Or can you confirm this understanding is correct? Okay? But still, I think email is not a main channel for communication. Personally, I receive 100 emails per day. If I spend 3 min each and my day is done, okay? And okay? So we still want to want our attention on this discussion. Forum. Okay? And the particular question, I think there’s a quite a few clarification question for a particular and the the 25 I think I made some comments somewhere here. So remember what we did in our class. Well, in 1st lecture, we define consumer problem, and later on we define firms problem, and we want to use the minimum or the minimum amount of assumption that can, you know, deliver predictions. So the the. So if you think along this line. So what is a minimum assumption, you can assume a person well or a firm, right? So the minimum assumption that works very well, it’s just to assume they will do things that benefit them. Okay, so this, this is like the you can call it 1st principle, or, you know, just just minimum set of assumption. And that can help you make progress. Okay? And so then you just define for the problem, right for consumer. Define your choice, set and consider the constraint, and they will just do the best for them, and same for for for comp, for firms. There are exceptions to this. For example, the last couple of decades here. A lot of extension to the neoclassic economic models, such as behavior economics. Okay? Which account for all sorts of behavioral bias that people are bounding rational. For example, you’ve seen one example, the limited attention, right? You have present bias timing, consistency, self-control problem, all sorts of behavioral models. Right? But at the end of day. You can redefine or reformulate the objective function for the let’s say consumers, or 1st right? And you still assume they are maximized. Maybe they are maximized, Miss. Specified objective function. Okay? But it’s very simple and it’s very elegant. It just assume people are doing the best. Okay? You can apply that assumption, your daily life. It also will simplify your life a lot. Okay, okay, our class is being recorded. Right? Okay? so that is about assignment question. Okay, Balaji is our tutor and a graduate tutor. Okay, thanks for healthy. Okay. let me start our lecture formally. Last time, what? We defined a firm which has some market power in the sense that the firm face, financially elastic, right elastic, demand or face a regular demand curve right downwards. Open demand curve, and can choose either price or quantity is equivalent, because the demand function is well defined. Price and quantity is one to one. Okay. so. And we. we discuss a particular example of ice cream pricing of a firm can choose the optimal. We we use quantity as a decision variable. Choose the quantity to sell and keep in mind. That demand. Curve is downward sloping. If I want to sell more ice cream, I have to decrease my price right that makes sense. And we discussed this pricing profit maximization principle where I will find a decision that gives that sort of set my 1st order condition equal to 0, which means that the marginal return or marginal benefit equals to the marginal cost, and that defines my final decision in terms of quantity. Okay? And that that is basically the key principle for profit, maximization of a monopoly for, or a single firm that face a downward sloping demand curve. Okay? And we derive certain property of marginal revenue and marginal cost. And we define key concept, such as margin and markup, which measure firms ability to charge to raise the price above the marginal cost right for a firm that has large market power. I think you probably have firms come to mind, you could think about apple right, which is. has very premium. Brand is able to charge high price, comparing to their cost. So apple is the perhaps the largest company on earth as the largest market capitalization. Right over, I think, like 3 trillion Usd. And the others. The other top firms are Microsoft, which is also has. Loosely speaking, this is not a legal concept, because in legal term, when you claim they are, they are monopoly. You actually have to provide evidence. And those firm like we discussed, they don’t. They? Frequently. They don’t admit they are monopoly. Okay, because it’s not a good term because if you you know monopoly, then you will face accusations from the the government. Okay, so. But they have to say that they have market power, because, for example, Microsoft is the the main provider of operating system, right windows. and the apple is the main. Like the the gatekeeper. We use the term gatekeeper of Ios or the apple system right? That is like very powerful and the powerful. The corresponding technical term in our class is a market power. So they have the market power. They can raise the price pretty high. Okay, higher than the marginal cost, or they can make profit. Okay? I think last time we also derive the elasticity rule that defines the optimal price. okay, let me start from here of our ice cream pricing problem. So just recall that we have a fixed cost of 3 per ice cream, which we we buy it from some wholesaler who’s sell to us at 15 of fixed cost. So pi, which is a normal notation or the conventional notation for profit function. So pi as a function of Q, and you can actually write the other way, pis, function or price. And you can verify that. That gives you the same answer. Okay. So now we use pad, and with the improviable decision variable, we use Q. And we write out the profit function, which is a quadratic function of second order function of Q, right? Polynomial. Okay, so we know how to solve, for the maximum of this will behave. Function is basically a quadratic function, right? And you can take a 1st order condition. With respect to the decision, variable queue. Set it to 0 and you get the optimal quantity your optimal decision is to produce or to sell. Sorry to sell 7 units. Okay? And at the 7 units the optimal price is 20 of compute time, 1 right? 1, and there is a person value this product at 100 for this model, right. So government should give me 500, which I have no idea. Just random number. My valuation for Mac is 500 to me right? Because then there is 700, or meaning that. Suppose we are in a world of uniform pricing firm can only sell one price. What happens? Well. that means I have to lower price everywhere else right? There are so many people value my product. 2,000. Right? So I won’t do that. So you can see that market power implies certain distortion or welfare loss, because from a denying consumer has valuation higher than market cost. Okay? okay. And so let me let me start perfect computation. I introduced the concept. Homogeneous product. Products are similar, and there are a lot of competitors. None of them are large enough to affect the market price on their own. Okay. and perfect information about price and quality. So those are the assumptions for this particular benchmark model. Okay, a model that is almost perfect. Okay? And then, we already discussed a lot of deviations or real world case where those assumptions are violated. Right? First, st it’s very hard to find very like 100% homogeneous product, right? Even for rice, agricultural product. They can have differentiations, right? Okay? And firm. In reality, all have certain degree of market power. Okay, you have many stores on campus set selling as a coke. Right? They still have some degree of market power because their travel costs for us right? Visit another store, although it’s 0 point $5 cheaper. I may not go right because of travel costs, so we still have market power, certain degree of market power. So again, I want to highlight, our example is, it’s very extreme. A theoretical benchmark, okay. perfect information or price and quality, and in the second half we will relax this assumption. This is also standard assumption. In classic economics. Firms, consumer. Observe the price right, observe the price and the quality with no information symmetry. But this will. This is an assumption. Because I think we discussed for a lot of price, right? So even today, with Internet with very low search cost very low. Such cost, you can still say different prices for same product. right? Because search cost is never 0, because consumers have limited attention, have search costs and and right, so that that is called market friction. And you also have people like selling product. Well, you don’t know the quality right? That’s why we have online review system. Right? So they are selling an iphone. Right? How do I know it’s a true iphone? We have the reputation system and try to solve the information symmetry, because it’s always profitable for me to sell some very like old iphone right or some some bad product, or I don’t send you iphone at all. You just pay me right, and I will sell you send you nothing right? So that is called information symmetry right? Because I don’t know the seller there. There is no 100% guarantee he will set. He will send me what was promised, or the quality is good. Okay? And we also will assume free entry and free access to production method. This, again, is a very strong assumption. Right? Free entry is well, perhaps true for certain product. For example, I probably can sell Nike shoes online, right? So that is somewhat free, or the entry cost for me as a reseller of Nike. Property is low, right? But for certain products let’s see if you want to sell Lv. Or sell something you need. Well, there’s entry cost. It’s not free entry, right? Suppose like, if you want to sell things on campus, actually don’t know the policy. If you can sell, if it’s a market free entry or not, I assume it’s not right. Okay, but you. You can see that there’s always certain friction in a market. Right? But I will talk about market. There is no no such friction, and we’ll use that that perfect world as benchmark model to evaluate all the division. That’s our starting point. Okay. free access to production method. This is even more unrealistic because you, you, especially when we talk about innovation, right? Because we have patent protections right? And I don’t have. I don’t know how to develop. Let’s see a system that is the same as Ios, right or windows right? Because many of their stuff is patent, protected. Right? And it’s also complicated. I don’t have that talent. So all those are deviations to this set of assumption. Okay, so I just highlighted this. A set of assumption are very extreme. Okay, they are very classic assumption. So perfect competition is not a general statement of the world, and perfect competition is definitely not a goal for the firm. Right? We talk about because Firm wants to have profit. You are young, smart, ambitious. you know, graduate from a Us. Right? You don’t want to enter to a market to earn marginal cost. Right? You want to earn profit so perfect competition, not a goal for a firm. I would also agree. It’s not a goal for society. because we want to incentivize innovation. Okay, if there’s no profit to be different or to be better. Right? So we we were in a different world. That’s not how today’s economy and market works. We reward innovation. meaning that we reward certain degree of monopoly in the initially okay, you can have some monopoly power power initially so. But it’s very complicated issue. Let’s see. How long should I? How much should I protect your patent right, and how for how long should I protect it? Right? What can be at what? What can, which kind of innovation I can grant or patent? Those are all very serious and and difficult problems. Okay? And they involve freedom right? On one hand, you want to reward reward innovation. On the other hand, you want to make the market more competitive right? Because in a dream case. Right? If the government, let’s say EU or the Department of Justice in the United States. Okay, which is a computation policy agent of Ftc. okay, Federal Trade Commission or something so they care they will. They will go after you if they feel you have too much market power, or you are a monopoly. And reason is basically because you will raise the price too high above the marginal cost at certain at certain level, you know. You will lower the social welfare. Okay? So some industry behaves a more similar or closer to a perfect competition. I see agriculture a labor market. Unfortunately, this is not good news, because as labor market participant we want, we want to have some market power, right? We want to be able to bargain, get a higher salary. But unfortunately. right? So we differentiate to some extent. Right? But let’s say to undergrad from Soc, you actually compete with each other. And there’s some, you know, homogeneity, so the computing can be very, very strong, and I hope we will never compete with AI. I have some discussion at the end. Financial market, right? Like financial service and banks right for many people. They don’t really have a preference to it. Their service is very assume the bank are all safe or honest or reliable right, and they offer similar interest rates or similar return. Then I actually don’t have that much of residual preference, so they are close to a perfect competition world. Right? I would just look at the the promise, the return right? So that is a very simple world. So perfect competition is a convenient benchmark to study the effect of competition. So demand and their perfect competition is different. Okay, so your competitive market firms are very small relative to the market. and they will face infinitely elastic, demand curve. And importantly, they behave as price takers. So that is different from our T-shirt T-shirt example, T-shirt factory, right. Our T-shirt factory, even our ice cream factory ice cream seller is able to charge higher than marginal cost, and they are able to influence the price because they fear they face a financially elastic demand. Right and same for most of the companies you you experience, you you interact with in real world. They are they face a fine? They face they are price. They are not price taker, and they can influence the price. They can sell their price. Okay, but our example. Firms are price taker. The price is given by market right, and they are not able to in increase your price. Once they increase their price, people will switch. Think about bank right, or offer homogeneous contract to consumers to banks. Right? So consumer only cares about the return right the the return, and if I lower my return as a bank, I lower my return promise by a little bit. Consumers switch away. That is infinite, elastic. Okay. so I encourage you to go through the slides before our next lecture. I have to stop now, because I’m already taking your time. Too much time, but I still aim to finish the set of slides in our next lecture. Okay, so please take a look before our next lecture, and I will stop you. Okay. I think there’s a question. okay, I think there’s a question. Oh, class is over. Okay, I’m just reading some question from Zoom, okay, I’ll see you next week. okay, and please take a look at the this, the last set of slides before me term, which is I think we have still have quite a bit for competition. So I will talk about this in next lecture. I think the zoom question is about in previous slides on pricing. Let’s see this slide. Right, let me say. lowering the price. Let me stop recording first.st

6. Demand

Okay, now, I’m going to start our last lecture for the 1st half. In the 1st half I recall, we studied the problem of consumer from the demand side. We, where we have a consumer who has a preference over over products or alternatives. And we define a concept of utility function in Difference Curve, where consumer, guided by these preference, I’m making some constraint optimization problem, make a choice of consumption bundles between apple banana between any 2 or multiple alternatives. And we define the concept of demand curve, demand elasticity to measure the property of demand. And most importantly, we have a concept called consumer surplus. which is used to to quantify how happy consumers are. And after setup consumers, decision problem, we introduce a supply side. just like, you know, regular market economy. You have the demand side and the supply side. The supply side is in charge of production, right? So buying resources or inputs like talented people labor right? Or computers lease a factory lease equipment investing, you know, buying patents, technology and firm is characterized by a function which is a mapping from those inputs to some output level or productivity level. Right? that function is called production function. So it’s a counterpart to the consumers. Utility function. Right? Once we have a supply of we have a production function. And we can define the firm’s decision, like pricing choice of prices or choice of quantity, right and price quantity basically is a 1 to one mapping defined by the demand curve. Okay, and we define many useful concepts, such as marginal cost, fixed cost, average cost, average variable cost. And as we highlight, those costs are very important in understanding production decisions, okay? And our last lecture will be on computation, where we are going to allow the the suppliers or the firms to interact. And we’ll have a concept called equilibrium. Okay? Equilibrium is a steady, a stable state where you know other parties making their optimal decision in this market. Okay? And it’s a stationary condition. Everyone is happy with themselves. No one will debate. and this concept come back again is a very important concept in economics. Also, you know, in physics you also have this concept called equilibrium. In the second half we’ll talk about, you know, a strategic interactions which we call games correspond to the famous game theory concept. So in game theory, you also have equilibrium, and you can have different type of equilibrium, for example, the most fundamental one which you will learn like in, I think. for example, graduate level, you will learn in the 1st half called general equilibrium. General equilibrium is similar to our the equilibrium concept. We’re going to set up today basically characterize the big picture of this market. how this market arrive at a resting point or steady point, the demand of supply sort of miss each other. And the market conclude, okay? And in the second half we will talk about other equilibrium concept, for example, Nash equilibrium. And if you learn more about economics game here you have some other type of equilibrium, Michael, perfect equilibrium. You can have even more equilibrium concepts, but those are the, I think, most classic and most important ones. But the fundamental idea is the same equilibrium gives you a steady stable state, where everyone happy with themselves, and this idea will be very clear in a second half. When we talk about equilibrium games, Nash equilibrium, where everyone’s everyone’s action is a best response to the other’s action. Think about, you know, in a class everyone cares about. Let’s say their their ranking right? So everyone make a make a choice on the level of effort. They want to make right. And your the equilibrium is a state where every student is making the optimal level of effort conditional, all the others choice. So no one want to make any change to the action. So that is like a steady state of equilibrium. And once we have, we clarify this steady state, we’re able to do something. okay, we can assess the efficiency of this market. Okay? And we are going to study a very ideal market called Perfect competition, where firms can enter, can enter into this market with with no cost. Okay, everyone have access to the technology. There’s no like monopoly of patent or technology. So you can. I would. I will talk about that later. There are many deviations from reality. The market you are familiar with, but it’s very useful benchmark. It’s something we have to understand in order to understand in order to appreciate. Why, you know, the free market or the market mechanism is preferred is so dominant today as a way of production? Or how do we, as a way of organizing our consumptions and production decisions? Is it is because it’s It’s very efficient and guided by the concept called computation and economic principle. Basically in one sentence, competition will eliminate above average profits or rents in the market. and such that the production will expand to the maximum level and is socially efficient. And this claim, it’s very strong. Okay? And there are many, many conditions that could fail and make this statement wrong, and I will highlight a few of them during the course of our lecture. But I will also make very strong assumptions, and I’ve talked about now, and to derive that very strong conclusion. Okay, so there will be a few parts. I will 1st characterize the setup of perfect computation and make predictions on what will happen in this market. and then we will analyze. Once we were able to solve one equilibrium, you are able to solve many by applying a similar logic idea. And there’s a type of analysis called comparative statistics. where we basically are dealing with problems where the environment change or the policy change or the technology change right? And how does that affect the equilibrium? And then you will resolve the equilibrium. Okay, derive the supply and demand conditions and conclude and settle the market and see how does the market move from one equilibrium to another? After you change, for example, the patent law? Or after you introduce a new technology, the market will change. And that type of analysis is called comparative statistics. Okay, so there’s some primitive. We call it primitive, something about the environment, about the policy, about people’s preference that changed. And how does this change affect the equilibrium of this market, or the equilibrium quantity, equilibrium price? And so on. We’ll see many examples later. And again we’ll return to our short run, long run analysis. But now we will endogenize from decision, and it will be many, many firms entering enter into the market. If you recall, we have done short run and long run analysis previously, but where we assume firms are priced, they are they are. They have market power. Right? Okay? Oh, we gave the price, and we have an endogenizer price. Or to really understand what is the equivalent level price in this perfect competitive market. And we’ll talk about applications as well as in this more current context of big data. And AI, okay, so we already covered this introduction part I would just sort of to recap and make sure we are on the same page. The perfect computation characterize a market of homogeneous product. and we will. We already mentioned this concept a few times previously. There is a corresponding counterpart called heterogeneous product. Where people’s the the product are different differentiated right? Using the jargon of, you know, economics. heterogeneous product means that product. They have some different attributes. They may have different flavor if using different materials. Different technology, different function, a different color different price. Okay, so no different price doesn’t really count that. Okay? So the homogeneous herogeneous really refer to the product characteristics. Okay. for example, brand, right? Brand allows the differentiation. you can have Apple versus Samsung. They have different brand, and they have many, many different details about product design that makes them heterogeneous product. Okay, so they are not homogeneous product. And now you realize that in reality most of product are somewhat heterogeneous. Okay, it’s very hard to find homogeneous product. We’ll give you a few examples. Suppose I’m hiring a programmer. Right? And the the task is so standard, and I don’t care about anything else. So there will be 2 market job market candidate that looks the same to me. Oh, then they are homogeneous. I don’t really care about one versus the other. Right? Or suppose I’m considering 2 contracts offering by offered by 2 banks. I only care about interest rate. Right. I trust the 2 bank equally. They are all very, very reliable. Okay, so then, that is a homogeneous contract. There’s nothing further details. But in reality firms try to differentiate. Try to avoid head-to-head competition. If you’re shopping on Amazon and you you can. You know Amazon has a feature called Buy box. I don’t know if you realize. So basically, the same product listing, there are 2 offer product offers under the same product listing. Let’s the iphone 15 right? Or pink color, 64 gb, RAM or something. Right? So they have exactly the same product listing, and 2 offers 2 different by 2 different sellers. And that product you could claim they are. They are homogeneous. Okay? And there are many, many competitors. So you’re you’re not monopolize the technology right? So in in real world context, firms have access to something, some technology that is include exclusive to them. Thinking about chat, Gdp has access to. Let’s see very advanced models, right? So that allows them to be differentiated or allow them to be vertically better than your competitors. So that doesn’t count as our perfect computation world. Okay, so imagine one day all those large language models. They are also good. They’re no different from one another. I don’t know if that they would ever happen. But you could argue that it already happened to certain degree. For example, if I’m my, if my need is only to do some polishing my writing right? You could argue that certain, you you already have multiple. like language model that can do very, very good, comparable to each other. Right? So that type of service that type of market becomes homogeneous right. But you could always come up with some need or some type of survey that is so advanced that allows one firm to differentiate it. Right? So this large language model is so good at python programming or handle certain project. Okay, firm always want to differentiate. Okay and perfect information of price and quality. we make this assumption. But I think we mentioned earlier that the real world market always have a lot of asymmetric information we’ll we will discuss in the second half will allow for information friction. Okay, for example, I’m not able to observe the product quality perfectly. So that happens in e-commerce where you have reputation system. Right? So, consumer. Look at a product review on Shopee. When you click on product, you cares about the product, really, because it could be the case. The seller will will send some, send you some bad product right? Or send you something else. If the seller has a low reputation, there always a chance. Be there. There’s some opportunistic seller, right? Okay. But we are not concerned about this complication. We leave to a second half when we talk about incomplete information, asymmetric information. Also, you can think about the case of Job market. Right? Let’s say, when you go to the market, apply for a job. Everyone will claim that the best right but firm has to really tell the quality. And that quality observed right. That is a case of estimated information. Then we will talk about screening, signaling. And how do? How do we communicate when there’s asymmetric information? One party knows more than the other. Okay, we will assume free entry and free access to production methods such that we can allow for many, many potential entrance. Okay. And perfect competition is not a general statement. We’re not claiming that the most of the market perfect competition. But you could claim that as a technology advance right? Man, more and more market will become perfect competitive market because the technology is going to be available to more and more people. Okay, because of technology advancement. Right? For example, you could argue that Google search in the last 2 decade allow people to search for information very easily, and some of the market becomes perfect competition like easily. If something you can Google search, search, use Google, search to find an answer. That type of service will become very competitive. Okay, I think in the older days, very, very. I don’t know. Today you still have this kind of service where you, if you want to book an airline ticket right? But there’s no like. People are not familiar with Google, or they don’t have Internet. They have to call some some intermediary which help them to book a flight. I don’t know, you guys, but do not if they still have this kind of service today, but now, I think if my my impression is more and more re being replaced by by Google search on on the Internet. Okay, I think we did also discuss in the last lecture that although we study perfect competition, we just use it as a benchmark. Okay, to understand market efficiency and competitive dynamics, and serve as a perfect ideal benchmark to evaluate market efficiency. Because when we talk about monopoly or oligopy where firm has market power right, and we will compare the market efficiency versus a competitive market, where I will show you in a moment that price will be pushed to marginal cost. Okay. and that is not a not good, especially if you. You are entrepreneur right? Because when you want to start your business, you want to make money, and you don’t want to want to price at marginal cost. So that’s when you you choose which major, which field you want to study in in university. Also, you want to study in a segment or in a field. That is not that competitive? Right? So why? Because in that case you will have some sort of market power. Right, you won’t. You will still have a market power, labor market power to earn a wage premium instead of earning something that is pushed to the lowest possible level. Okay? So I will come back to this point. I will when we discuss AI and the labor market. But now let’s get started. So in competitive market. There will be many, many suppliers. small or large, firms of different size, right? And they are so small comparing to the whole market, they don’t have any market power, and they are price takers. Okay? So this is an important concept. If you if you’re apple, you’re not price price taker, right? You can. Tim Cook can decide if if he wants to price that 2,000 right? Because they have market power. But if you are a small, tiny seller in a very large and competitive market. You don’t have market power because market price is given. and we will discuss how this market price is derived, and it’s derived by equilibrium condition. When you take into account our competition. Okay? And you’re not able to deviate from the market price. If you increase your price a little bit, right, people will switch to all the other suppliers, because there’s infinite many supplier in the market. Okay. so this is like theoretical ideal of of of markets. And on this graph you see, the blue one is a market. It’s an aggregate, you see, a large queue here, right? You see a large queue which is the aggregate demand. On the on right, you see a 1. Firms demand curve any price. You see, a small queue is a quantity. Okay? So the firm cannot cannot make any decision. Basically, it’s a price taker price is given by the market. Okay? And you just decide giving you capacity, how how many units you can produce in this market. And this large queue will show you data that will be an aggregation across all the small queue. There’ll be many, many sellers right? And you just aggregate. And this is a source of like innovation, because firms want to innovate. They. They want to avoid this perfect competition. So they want to be different, right? And they want to have market power and set price. And that is the really the source of innovation. But as the technology progress, right? More and more old technology outdated and they will become perfect competition market. Imagine now that you still produce like a very low tech cell phone, right? So that market is gonna be very competitive. But if you’re producing like very high end, with perfect with with the most advanced camera and and and operation system and hardware like apple right keep advancing. And and the basic motivation is is to be is to differentiate, and is to have the market power right to charge a higher price. If the firm, unfortunately are, you know, situated in a perfect competitive context become a price taker. You will face a infinite elastic demand curve. So as long as you increase your price a tiny bit, you will lost all the demand. Okay, so the demand elasticity apply. Our formula will be infinite. So we can apply our formula for marginal revenue for this firm and marginal revenue will equal to using our formula, p. Times one plus one over epsilon. And that gives you P, because absolute is one over absolute goes to 0. Okay, so the margin revenue is always always equal to the price and profit. Maximization price, marginal revenue equal to marginal cost that implies price will equal to marginal cost. Okay, so that concludes this proof in this perfect comparative market, the firm will charge the marginal cost. So what does that mean? That means firm is not able to earn any profit that is a false or competition, because there are just so many homogeneous competitor who are as good as you, and they are always willing to supply or to provide a further unit at the marginal cost. Right? So that sort of competition force will push down the price to the marginal cost. So, under perfect computation, each firm supplier will be given by the marginal costs. Okay, as long as price is greater than the minimum of average cost. This is the insight we derived the earlier lecture we talked about cost right short run, long run cost and long run. The entry decision is determined by average cost. As long as average cost is fine, meaning that price is greater than average cost firm, you know it’s not not. It’s still. you know, not losing money, and they will stay in this market, or they will choose to enter this market, and their price will equal to the marginal cost. So use our old example. T-shirt factory. remember, the tissue factor will pay 1 per hour, a weekday, 5 days a week, 8 day, 8 h per day. So it’s that. That is 40 days. I’m sorry. 40 h, okay, and 2 h for Saturday up to 8 h and 1. So this is some conclusion we have derived previously. So together, this this optimal condition lead to the short run, the long run supply curve. I think we discussed this part in our previous lecture. So basically. this is the long run, right, long run. And this part, this is a critical point. Let me use a different color and be this color. It’s a critical point, because this part, this part still profitable for show run production, right? Because it’s a, the average viable cost is basically marginal cost. If you produce 40 units, if you ask your worker to only work on the weekday, and the wage is 1 is profitable to to produce or to sell as a short run decision. So the short run supply curve allows for lower price and a lower supply curve. Okay, this is something we’ve learned before. And again should recall our conclusion. So the long run, a supply decision determined by the average cost, but the short run supply decision determined by the average viable cost. and the difference is that app that in the show right some of your costs are sent cost right, like your equipment or your facility. Maybe you know already been paid so that it will not enter your strategic decision because marginal decision determined by marginal cost and and micro wrapping. Okay, and the market supply curve will be an aggregation of many sellers. Right? Their individual supply curve the same as the demand function demand. Curve is aggregation across all the consumers. Valuation? Right? You have many, many consumers. When you change the price. You have many, many marginal consumer. They make the decision. Some will change their mind will drop out the market. Some will switch in and start buying the product. When you move up or down your price. but same thing on a supply side. You have many, many firms which has their own cost structure. Okay? And you will make the decision to to produce to supply, and you can aggregate across many of these firms, and you will get something called market supply. In this example you can see that the market supply capital Q. At a given price given level price for P. Prime is equal to the sum of the 2 firms supply. And you can. You have many, many of these small firms, and you can aggregate across all of them. So at this given price and the aggregate quantity this market is willing to supply is q. 1 plus q. 2. The idea is very simple. And I have this example of California electricity production. And you can have many, many firms. Okay? And they are producing electricity. Okay, they have their generator, their their infrastructure. And they have different variable cost and fixed cost, different capacity, different production or cost structure plus and production schedule. Right? And you can aggregate across all of them. Right? Imagine you are changing the market price, and you can compute how many capacity, how many supply, exists in this market? This is the incomplete list. Okay, so just give you an idea. I don’t think you can aggregate to get the exactly supply schedule in the next slide. So this is basically an example. You can see a lot of heterogeneities just like for consumer, right? Consumer have different tastes. Some cares more about brand, some cares more about price, right? Some cares. Some use this AI more for for writing, some use it for coding, so they value different things. There’s lots of consumer heterogeneity in tests, but you can still aggregate, you get at market demand same thing. Some firm here may have higher fixed costs, some may have higher variable cost, some may have larger capacity. Some may be smaller. Okay, but you don’t really care from market perspective. If you are, you know, market designer, you can just take the all their production function right? And you can just aggregate condition any price, and you can compute how many each of them, how much quantity they are willing to supply, and you will get the curve which you can see here right? So if the price is 0, of course no one want to supply. But as soon as you start increasing your price, more and more people will will switch in. Mobile firm. Will are willing to to supply the the electricity right? So you can get some some very realistic curve for given market. Okay. there’s a comment here. Okay, the market supply curve is basically the supply curve. Okay? Because previously. okay, I think supply curve will come later. Actually, so supply curve correspond to a demand curve. Right? Demand curve is basically characterize the relation between price and quantity from the consumer side. Right? So give a price. How many quantity aggregate in aggregate this market consumer? All the consumer in this market are willing to consume. Okay, so that is our demand. Curve right? And I will show you the market will arrive at equilibrium or a steady point of P. Star. Q. Star. Okay? And the remaining element, in addition to the demand curve, which is downward, sloping the supply curve. Okay, the supply curve is upward, sloping, because when you increase the price more and more consumer and more and more firms are willing to to produce for you. Okay, so this is the this one looks, you know, like a step function, because you don’t have infinite many suppliers. Right? So if you have infinite many supplier, and this market is infinitely large. You have smoother sort of lines. Okay? But you still have, you know, different density of firms, different type of firms there. For example. and it’s 5 5,000 MW, right? 5,000 MW, you have to increase your price a lot to encourage more supply. Okay, so that is the reason is because there are just not that many of firms in this. given your technology, given your production function, it’s just not profitable for for many firm to to jump in to start increase their production. Okay, so so this you you have very complete curvature. But realistically, they reflect the the complex production function in this market. When you when you do the aggregation okay, I will. I will move on. Let me know at the end we can still discuss. If you have, you have further questions. Okay. so market equilibrium. Okay, we have the demand function and we have supply function when it’s upward sloping when downward sloping, and they will cross a certain point right? The point they cross. The the 2 line 2 curve cross is basically our market equilibrium, where the the quantity demanded is the same as the quantity supplied and the price it’s also agreeable between the supply side and the demand side. Right? So they come with point. Right this market. They come up. They come with a point that both sides are happy. Right? The demand side are happy with this level of price, and they’re willing to consume 1,000 units. and the supply side are happy with this price and willing to supply 1,000 unit. So the market conclude. So that is the equilibrium where the supply meets the demand. And you can imagine if price is higher than P. Star in the next slide. You can see this idea in in a graph where you have the downward sloping blue line, the demand curve and upward sloping supply curve. Okay. and the the crossing point. Here is P. Star and Q. Star, that is the equilibrium. If price is higher. then the equilibrium price PP. Ph. Is higher than P. Star. So what happened? Well, at Ph. you can see that at this level price this is demand right? A point I can find on demand curve. So this is the quantity demanded right? And this is the quantity that firm are willing to supply. You have overproduction right or excessive supply. Okay, so what will happen? Well. then, supplies greater than demand have to sort of sell out all the product, all the units, all the products firm have to decrease their price right? So pricing will fall so that it will not be an equilibrium point. Right? Imagine sometimes it could happen, because firm mispredict, the market. Right? For example, covid calm, right? So so. Then, a lot of the supply of supply side of firms mispredict the market so there could be a issue of oversupply. But if they are just over production over capacity issue, right? That, for example, airlines schedule too many flights, right or or Google, by Amazon by too many server. Right? So what they can do is they can decrease your price right to make the supply and demand meet, because because at the at the that P. Star, you know, the market is not able to absorb or utilize so many units of products, right? So they have to decrease the price. But in the other case, what if price is lower than the P. Stock? Well, so that is pl in this graph? Right. let’s take a look right? So PL. Correspond to this quantity and this quantity right? So this queue and this queue is the point on a supply curve. So that is the quantity firm is willing to supply. Because now price is lower, right? So more and more firm will say, I’m not interested right owning a smaller number of firm will stay. We’ll say that, you know. Okay, I’m willing to supply, or if for individual firm, like a T-shirt factory, they would. The T-shirt factory will say, Okay, at this price. I only I only work on weekdays right? I don’t work on weekend at this price to have a excess demand right or or a shortage of of supply right? Because now price is low, actually more consumer. Get interested, right, more consumer, get interest, but not. You don’t have enough firm that are interested in, because they are not able to make any money right because they have. They still make money based on their cost, the cost function right, the cost structure. So that is the idea. So the equilibrium is the P. Star and Q. Star, where price and quantity exactly equal for the demand curve and supply curve. So this is the most famous law of supply and demand. Okay, right? You always in any economic book. You’ll see a supply curve and demand curve right up downward, sloping upward, sloping across that point. And that point is the market equilibrium point. Okay? And you can argue why all the other points are not equilibrium, because either they are too little consumer interest, or there are too many or too little supply from the firms. Okay, the market price is not high enough. Okay. Now, we derived equilibrium of price and quantity for for a single for for a given market. Okay? Or given equilibrium. What you can do is you can make predictions on what will happen if there is some change in the market condition. And that type of analysis is called comparative statistics. comparative statistics. Basically, it’s an analytical analysis on to make predictions. For example, what if we have government make some environmental friendly policy right? Or they have some some control on Co 2, right? Or the increase interest rate. How this will impact the industry in the affect the market. Okay? So this type of analysis is known as comparative statistics. And what is the important change in other factors is sometimes called primitive of the market. Okay, it could be changing policy or technology. And how does it affect the equilibrium prices and quantities? The answer is to we can compute a new equilibrium. Okay? And we can, you know, analyze how demand curve will shift our supply curve will shift right and how that? Where will be the new equilibrium, price and quantity. How does that depend on, let’s say, the elasticity of some supply or elasticity of demand when the curve is flatter or the curve is steeper. How does that affect the impact of the policy? Right? Because I will, as I will show you. If the curve is steeper or flatter. Then, you know, the new equilibrium will will be different. Okay, I see the price. the change, magnitude, change in price can be different. And this point we have been made in in the 1st half. So it’s important to to distinguish, really to translate the change in policy, to the change in our model. Right, so is it a change, a shift in the a shift along the demand or along the supply curve? A movement along the demand supply curve? Or is the shift of the curve right to the left or to the right. So I’ll give you a few examples. Real world examples so Taiwan as many of you may know, is a major production hub for computing device that chips right? So what happens if there’s an earthquake in in Taiwan, which happens many years ago. You were very young, I think that was like beginning of the century. What if there’s an earthquake? What does that do to the market price of chips. Right? Okay. and you can also think about an introduction of new technology. How does it affect grammarly, which is like a A platform that helps you to polish your your writing. I don’t know your I know you’re familiar with this. So there was one called Grammarly. Apparently they are, you are, you know, impacted by the life model a lot so that can can. You can conduct some analysis to sort of quantify the impact to to them. I actually don’t know what happened to them. okay. And also you can, what happened to housing price if there’s a change if the stock market is going down right? So what is the concept that is important here to make this prediction? Well, it’s the income elasticity, right. So previously, we in early on we introduced a concept called Income Elasticity, where, for example, because Palo Alto is in Silicon Valley, where you have a lot of tech entrepreneurs, right? Or employees at Google, for example, right? Their wealth, their income is affected a lot by stock price. and if stock price decrease, so that is an income shock to many of them. Right? And then you have. If there’s an income shock. Well, we know that production decision is elastic, too right, or will depend on income, right? So that is called income elasticity of demand. Right? So if their income change, the demand curve will change right, there will be shift of the demand curve. and that will affect housing price. Right? Okay. Let me use this earthquake example to visualize. How do we do this type of analysis. you can see the demand curve. Stay the same. Okay idea is that? Well, if there’s a a shock in Taiwan, so we have to understand what it, what is shock mean actually in our model. So the mid in this model, the shock basically is a supply shock, right? Meaning that, you know, because consumer of chips are all over the world right? They are not impacted by this earthquake. But Taiwan is the production hub for chips. Okay? So if there’s an earthquake, many of the factory are impacted right? They are not able to supply. So what does that mean? That means at a given value or price right at a given value of price you will have fewer or lower supply, so that means that the supply curve will sort of rotate right or shift upwards. Right? So when you increase the price you have a at a given level of price, you’ll have a lower production right? You have a lower production at any given level of price. Okay? And so the supply curve will shift from s 1, or you can call it rotate. It really depends on the exact production function, and how this shock really affect the production. Add the market equilibrium will ship from q. 1 to Q. 2, and move from this intersect to this right to this point. So that is what we predict will happen if there’s a supply shock, negative supply shop and price increase right price increase and and quantity decrease. Quantity. Move from q. 1 to q. 2, and price increase from q. 1 to from q, 1 to p. 2. So that’s what we predict to happen. I think the earthquake quake is around September in this is which year in 1990, 1999, is it okay? Or, Oh, yeah, before 2,000, actually. And you can see the price increase dramatically. Right? So that is same as we predicted. But in a longer run it goes back to normal because the supply size able to recover from this shock. So this is negative environmental shock which you can recover. But you can think about there are some shock that are more long term right like technology shock. Right? So like the introduction of Chat Gdp will be a shock to a company called Grammarly, but that shock is not recoverable right? So the the firm has to make some managerial decision and how to navigate the new technology landscape. Okay, this is why example. And we can be more precise even with just with the curves. We can be more precise because we we’re able to sort of predict the size of the response depending on the the slope of the curves. Okay? Or how elastic is supply and demand. If there is a supply shift because of some supply shock, either technology or some, you know disaster, then? Well, the prediction is that equilibrium effect on prices and quantity depends on the curvature or the elasticity of the demand curve. I will show you next slides. On the other hand, if there is a demand, shift, consumer taste, change, or market, expand right well. and the change in equilibrium, price, and quantity will depend on the slope of the supply curve. So this is how you visualize this impact. So look at the top 2 graph. you can see a shift of the demand curve right? So the demand curve shift to the right right. Move to the right. So what does that mean? Well, market span right from d 1 to d. 2 market span, because at the same level of price I draw a line. I draw horizontal line. You can see the implication to the 2 demand curve. Is that market expanded right at same level of price? I have more demand. Right? Q is bigger at D 2 than at q, 1, okay. D. 2, and then at d 1, right? So market expanded. Okay, you can think about if this is the demand curve for for a hotel. Okay? Or for airline. It may be some holiday or some events right? So they are more consumer coming. The demand curve shift. Or if if I say, if a company is doing some promotion right running advertisement on the Internet on Google, right? So then demand curve will shift to right right because you build some awareness more consumer learn about your product. You reach a larger consumer base. Okay. so suppose you have this shift or demand curve. Well, the change in price and quantity, and it’s very intuitive will depend on the slope of the supply curve right? So on the left graph, the supply curve is flat. What does that mean? Well. the supply curve is flat, meaning that supply is very elastic, right? Because you increase your price a little bit and supply respond by a lot. Right? That is the the to the left, right, the left graph. This part supply is very elastic. You change your price by a little bit, right, and demand, respond a lot. And this one. you change your price by a lot, but sorry supply. You change your price. You increase your price by a lot, but quantity doesn’t change much. So supply is in classic. The curve is steep. Okay, you can see that a relative change when there is a shift of demand depending on the it assisted you on the slope of the supply curve. Okay. if the supply curve is very steep, meaning that the supply curve is elastic, then the effect will be a large price response. A large price increase and a relatively small quantity. Expand expansion right? But if the supply curve is flat. well, then, you will see a large increase in quantity. a relatively small increase in price. right? And same exercise which you can do. Suppose you have a change in supply curve, like the Taiwan earthquake example, where supply curve shift up. What does it mean? That means the supply shrink right? So the sum of capacity are gone. Okay, some of the production capacity are gone, and market supply side experience, some negative shock. Well, the effect on price and quantity depends on the slope of the demand curve. If the demand carries were steep. if the demand curve is very steep, then we expect a very large impact on price right? But if the demand curve is very flat like this one, right? So we expect mostly effect will happen to the quantity. Okay, so this is the comparative statistics depending on the exact curvature of the supply and demand curve. You can do this kind of market analysis competitive analysis, and to derive the new equilibrium prices and quantity, and to make predictions on the impact of impact of some, you know, environmental change. And we can have more examples for different events. And how does it affect the market equilibrium in different market? So if Opec, which is organization of oil supply that controls the oil supply. Well, if if Opec increased oil output, what happened to a world price? And actually, those 4 examples, okay, correspond to each case we consider here, okay. so you can class classify all those events into a supply shock or demand shock and try to make a prediction on price impact of price or quantity. And can you can make some, you know, assessment? Whether you know if it’s a supply shock, is the supply is the demand. It curve, steep or not steep. Okay? Or if it’s a demand shock, right? Demand expansion. That is a game, a popular game. And so market span for the hotel rooms. Well, is the supply curves elastic or not? Right? Right? So I can give you some hint here, for example, hotel rooms. Is it the supply hotel room, elastic or not. Well. one could argue that you can come with different perspective. But here is one you could argue that a hotel supplies elastic because it’s capacity, constraint, right? It’s just there, right. And if you are exhausting the capacity right? It’s very hard to find actual room right? Because it’s fixed in the show run cannot adjust number of rooms for hotel. Right? That’s why, when there’s like, I think, fy in Singapore the hotel price is crazy, right? Because supply. Is not that elastic when you have huge event? Right? And it helps. If you have Airbnb, so you can, you can. You can actually think about the sharing economy, what they do to this market. Well, in a lot of case they make the they increase the capacity. They actually make the supply a lot more elastic. Right? So that’s the idea for many of the the the It technology. It makes the market a lot more elastic. Think, think about if in the old days, when you only have taxi right, and it’s full time for everyone, right? And the license is very expensive, and you have to have all sorts of certifications, and so that supply is highly regulated. and that supply is very inelastic, especially when you meet the capacity constraint. So that’s the old model, like, 20 years ago. Yeah, I don’t know like the old day when I travel to a new city, right? It’s always very hard to find a find a taxi, especially during rush hour. because the capacity constraint. But now this is better. Even you have Taylor Swift. Concert right, Taylor Swift. I think I was. I was still able to find a grab that night. Well, why? Because because we have a grab right? Very elastic supply. When the firm, when the platform can increase the price for the drivers, right? Increase in which and you have a lot of more people who are a casual driver. They’re not full time. They’re part-time. They just, you know, just have time, and they find the price reasonable. They switching right? And they start. And that supply that makes the supply a lot more elastic. right? And that is a very nice feature of the sharing economy. Same for Airbnb. Right? So because you have much larger potential level of supply. If you take into account Airbnb, right? Because you can switch between like a hotel mode. And and that hotel mode very easily that basically increase the pool of hotel supply. and same for aws, server. Right? So you let’s say in China they have the single day for e-commerce, right? So that day the server load is super high, right super super high, like you have. a few years ago they still celebrate sort of the G. The Gmv. For that that day is 1111th or single day like a shopping event. And you need to have very have a lot of compute right? A lot of server to manage that amount of traffic. but because of, you know, the the cloud computing that actually allows for the the supply to be a lot more elastic. So that is something something good that we we didn’t have a Pre, the, you know the technology revolution. Okay, so this is the basic idea of comparative statistics where your decision, your prediction for the market equilibrium will change, depending on the market condition. Okay. so now I’m going back to the short run and long run analysis and try to make some discussion and predictions in our context of perfect competition, and well recall what what is the most important difference between show, run decision. The difference is, the cost is different, right? The cost structure is different. It depends on the timeframe you are. You? You use to evaluate the decision. and you will take into account different costs. Some costs are sunk costs in the long run in a short run, right, but they are marginal costs in the long run. Okay? And class is very important in determining the supply. Okay. from firms. Entry decision to there the quantity they want they are willing to supply. Okay, so that is very important. And in our context, let’s say, in the last year so there was this news Sam Elman was proposing. I think he denied. But there was a lot of news saying he proposed, or he was discussing some 7 trillion dollar semiconductor vision. So what is this doing to the market? Well. you can see that if there is a large investment in the semiconductor industry, suppose you find a replacement or find some competitor, or you, you just build a competitor for a media, or you build a competitor for I think. Tsmc, right? So the there is a Semiconductor Manufacturing Company in Taiwan, which is which has, like huge market share for this manufacturing of a semiconductor. Well, in the second half we are going to talk about you. You will solve games or solve market competition and market equilibrium. When compare the as in the monopoly equilibrium with the equilibrium that has a huge implication on the quantity of market supply and the welfare and the price. Well, if you’re able to sort of invest a lot in a semiconductor market. Well, so that will have. That will sort of expand the supply by a lot. Right? Because one reason now there is a shortage or the market price so high is because in a lot of the the key Lexi. the key sort of input right, or the the equipment like Tsmc or Nvidia. They are actually very concentrated, right? Very concentrated, meaning that they are almost they occupy a huge sort of market share for that market segment. Right? So then, you will have a problem where firms have market power, and they have a tendency to maximize their profit, which means that they will price high, and they also have capacity, constraint, right? Meaning that they are not able to to expand to the, to the extent that they maximize the social welfare. But if you’re able to coordinate a large investment in hardware in the infrastructure or building, let’s see, this semiconductor factory right that is able to capable of producing a lot of the hardware right? That will have a lot of implications to the to the market. Right? You will solve the the key The the reason there’s a shortage is because some investment is not profitable. Right? So if you are able to coordinate with, let’s say the government right, and some, you know, all coordinate across many of the the firms that are willing to build or to opt in, and to provide, to commit on this fixed cost right, and then the market supply will be much higher, because once you already commit or invest it on those hardware or the factories right, and then the remaining decision will be will depend on the variable cost. Right? So that will sort of help reduce the reduce the the cost right, or improve or expand the production. Okay, and that will have impact on a downstream market. Right? But frequently because there’s externality, I will talk about externality later. There’s analogy to this. And individual firms is not able to internalize this in this externality. Okay. for example, Nvidia is a monopoly firm. Okay, based just based on their market share in certain market right? Producing more benefit, the society more than benefit Nvidia. Why? Because because Gpu has a downward sloping demand curve right? So if Nvidia is to expand the production, what happens? Because that was looking demand curve, it has to decrease the price right? So that goes back to our discussion on the monopoly or the the pricing decision of a firm that has market power. So it’s cheating off the marginal return versus the marginal cost, right marginal return marginal cost right? But it’s different from a perfect computation case, right? Because in a perfect competition case the marginal return always equals to the marginal cost. Right. But, Nvidia, if you are firm with market power, this epsilon is not minus infinity. You feel elastic demand. Okay? And marginal revenue is actually well, you will set your price such a marginal revenue equal to marginal cost. But sorry you’ll you’ll set your your your price and margin revenue because you mentioned a cost. Right? Okay? So there’s externalities of production that a monopoly firm will not internalize. Where are we? Okay? Because if if the monopoly from is to produce more. It’s competing with itself. Right? It has to be has to lower the price. It has to lower the price. So there, there’s a cost of doing that. Okay, but to the society lowering the price doesn’t really hurt anyone right? Lowering, because the price is a transfer from the firm to the consumer right. but is but to Nvidia lowering the price is costly. Right? So you have this this issue of when firm has market power. The supply may not be socially optimal. and we can look at the supply function and the perfect competition. Compare the short run versus the long run you could have in the short run. For example, marginal cost is above average cost. The margin cost is the right line. It’s it’s going high as you you are expanding too much because a given firm has some capacity constraint. Okay. on some. This economy of scale and every cost is downward, sloping, and in the show right? We have marginal cost is above the sorry. The market price is above the average cost right at this level. So this firm, actually earning actual profit, right? Because price is higher than the average cost. Right? Because price is here and every class at the optimal quantity is here. Okay, so price is above the average cost. Right? So what happened in the long run? Well, in the long run you have more company joining right? Because this market has read right because this market price suppose the company another firm entrepreneur right, has access to the same technology which is our assumption and the perfect competition, and that from we enter into this market because this market has rent right. The price is not low enough. So you have more firm coming in. Okay, you have more from coming until the point where the price is low enough. The price goes to here. Right? Okay. When the price is pushed down to the long run average cost. Bye, if you have. if you your price still higher than the long run, average cost will happen well, in the long run, you will help from joining to this industry because this industry has rent right? So it’s like in in us, student keep applying for Soc or a computer or Cs major. Why? Because or or BA, right? So because those major, you have a you have a red right? You have some market power in labor market and charge high. which, right? But in the long run, what will happen? The long run? Very long. Run well, in a very long run. So well, there’s just so many people coming to the market to push the price down until the price equals to the long run average cost right to the lowest point of long run. At that point no one want to join anymore, because because Csmb becomes a average major at us. Right? So now it’s it’s still very, very attractive. But again, this this example is not perfect. Because it’s not a perfect competitive market, right? Because, we don’t have infinite many of students right? So even in the long run, and we actually control the entry. Right? The entry to sales is not. It’s to to associate is not free entry, right? You have to have very good grades or something. I don’t know the details. You guys know much better than I do. Okay? So that’s the difference. Right? Because in a perfect competition market firm they don’t need to have high Gpa to enter into the market. Right? So they have infinite money or potential entry, and in the long run you’ll push the price down until it meets a longer average. Every cost. Okay? And competition is in the long run. So all firm, our assumption, all firm, have access to the same technology. and there’s no entry cost at all right. You have, even in manual firms, that thinking about entering into this market and that competition will push the economic profit to 0. Okay, or to the opportunity cost. You can think about that way. So there are some sort of average level of income for all, the for for capital or for labor in this society, right. And the idea of economic rent or economic profit goes to 0 basically means that this particular industry is no different from average or the average outside option opportunity cost. Okay. so that is the implication for competition along right? But in the tech industry a significant difference is that not all firms have access to the same technology. Right? So that is why tech industry is attracting so many capital from the stock market. Right? There’s just so many people willing to invest in this industry. Why? Because because firm have talented people right? And those are not. You know, those are not like everyone can have right. Those are very scarce. Scarce. Resource. Okay, you don’t have infinite infinite supply of the top talented people, and you don’t have you. You have patents right. And like some technology that is ahead of that is not accessible to others. And the background is a discussion of open source and closed source. How does that affect innovation? And that is particularly interesting. Given the you know the the interesting case, open AI, where, the name is called open right? So early days. They are evasioned themselves to have very open like at least more open than average company, right? But they are actually being perceived as it goes on being perceived as slightly less open than the average like very frontier tech company like Google, for example, Google shares so many of the research in the conference papers and publications, right? And really help drive a lot of the innovation. Early days. that sort of becomes the foundation of the current wave of progress we have witnessed. Okay. and there’s another event happened quite recently, which is the deep seek which I think, according to some news they claim to have, very low cost in producing some quite capable models. and that sort of trigger. Some discussions on on the open source versus close source model in driving innovation, or even, you know, driving the success of a company. This is a very, very complicated discussion. I will talk about some of the, you know, innovations and how to reward innovation, how to incentivize innovation in the second half. But I think at the same time, you also see, like a Nvidia stock, respond a lot right to this news. Well, so at least some part of it is, because because it’s it’s because competition we will talk about in the second half that a monopoly is able to attract much higher higher profit. So that’s why, if I frontier innovative, firm, right? And the capital market is willing to invest a lot of money on you, and and that is based on the expectation that you have the monopoly power. But if the market is very different, as someone open, sourced everything. So what happens? Well, that that that means you are getting closer to the model we are talking about in the class, which is the perfect computation model. What if everyone have access to a technology right? So you lose the monopoly power? Right? Because so that is, that makes a huge difference, because for monopoly you’re a very high price, and in the second half you will solve some of the. You know the the games where, just by introducing one more firm right, you will lose a lot of monopoly power, and you will lose a lot of profit. But if you are, you’re moving from monopoly to a perfectly competitive market. That’s that is a huge loss of the the profit potential right? Because once everyone has access, then the market becomes very competitive, and you can in extreme case, when this, the setup is exactly what we discuss in our comp perfect competition model that the firm lose all the ability to make any profit. Okay? Right? So in that case, you are just. You’re not like a frontier AI company, right? It becomes like a a small company in a perfect competitive world. But again, it’s not in no way, no way close to that. But it do make a huge difference. This access to technology. Okay. in terms of firms, profitability. And also you have to stay very active, you know, respond to all the possible opportunities. And when when we how it works in a tech company, as we discussed earlier, is because of networking effect, because success has this path dependency, right? Success has this path dependency. If you are early mover, you have a huge advantage. Right? If you are, let’s say, open. AI have lots of users. Right? The user, you can accumulate large amount of data right? So it’s a game. It’s a competition among the firms. so they they constantly want to release their, their their new model progress, and to convince the the investors they are ahead of the market, or they have access to superior technology that is not accessible to others. Right? So that is a signal they want to. They want to send, and they want to. They keep doing that because tech industry is very special in a sense that they have huge network effect. If you’re a leader, you enjoy a lot of network effect and benefit. And recently, there’s a news in China, also that Wechat is sort of collaborating with deep seek. you know, like opening to sort of integrating your model to the to the to the wechat ecosystem, so that also may have, if they are serious about that collaboration that may also have a huge impact. And reason is basically because of scale effect, right? Because wechat is the largest ecosystem in China. They have, like over 1 billion very active, intensive user. And if you are, if the companies are able to successfully integrate, that means AI company, let’s say deep will have access to a very large user base. Right? So that implies a huge amount of network effect in terms of data. Right? But I I’m not not I. I don’t know how how exactly that will work, how serious that that would be. But still I I think at this stage key advantage that Openai has is is very large user base, right? It’s still like number one, in terms of active users. That means a lot of network effect, right? And data is another source of of scale that return that have high return to scale. So that’s why the same thing for Nvidia, right. You always want to keep ahead. Right? Keep introducing new models and try to scare away potential competitor. Okay, and try to build confidential investors. Okay, to present themselves. Have Mono have some monopoly access to to new technology? okay, I will. I will. I will conclude with an example of tax. And it’s very simple. So if the government imposed a tax, who will bear the cost of the tax well, this is another example. I think it’s a classic analysis that we we can do. Because in a lot of the industry that we have, we do have some sort of government intervention. And the idea between a lot of government interventions. Because government believe there is actuality, then market is not internalizing, like smoking, right? So we have tax on smoking. In Singapore. We have tax, or we have some, you know, fees for car, right of a car license or so that is also because potentially because of externality, because capacity constraint, right? Because traffic there’s externality of traffic. If everyone drives a car potentially, the waiting time will be high, the traffic will be overly crowded. So government is trying to impose some tax on on that to help the user to internalize externality right? Because when I pay the money when I pay, I’m a smoker. When I pay, buy a cigarette and start smoke. I don’t take into account the harm I I cause to my neighbors right because I smoke, and some my neighbors suffer right, but I don’t take into account that in my decision of buying a cigarette. so in that case the Government can help to correct the market by imposing a tax to charge me 2 more dollars. Right? Why, 2 more dollars! Well, it’s more complicated, and I will show you in the graph, but just say that the Government charged me 2 more dollars, that is, to help me take into account the harm I’m causing to others right? So next time, when I make a decision to buy the cigarette and I have to spend 2 more dollars. So that helped me correct the actuality in my decision. Okay, but it’s more complicated as showing this graph, because when government imposed a tax. it will be shared between the producers and the consumer right? Exactly. How do they share this tax will depends on the pervature of the demand curve you can. You can think about the tax of tea being that a shift of the supply curve by T, and you can see 2 sort of equilibrium point Pre and post tax. Okay, create the price. Level is p. 1, and post the price that will be p. 2. And there’s a gap between this price, which is the price goes to consumer. That’s the price on demand curve between the price goes to consumer versus a price received by the supplier. Right? Okay? So that is the idea. So that is the effect of tax. So this gap is t the tax, the money taken away by the government? Okay? And you can see that there are 2. You can divide this segment by 2 parts. So the 1st part is between p. 1, p. 2. That is a change in the out of pocket price by the consumer. So that is the part of the tax bear by the consumer. Okay? And you have another second part. which is this part right? That is thereby the producer. Okay, that is bad bear by producer and overall. T. Amount of money is taken away from this market by government. Okay? And idea most most of the times because of externality. For example, there are some other type of actionable correction which is not imposing tax by giving reward right? You also have that having a key, right? So I think, in a lot of the advanced economy, like, I think, in Singapore, and perhaps I guess in also in some other country like Korea. I guess where the birth rate is too low. A government will give. Subsidize right. Give subsidy to a young couple to have kid right? Because having kid has actuality. right? So that helps the the society. So that’s why government is giving away money or government want to help build a manufacture in Singapore right? Or in some country like in some country. They say in the Us. They want to subsidize the green industry right like Ev right? And they will. They will subsidize for consumer if you buy, and I will. Actually, government will help you cover some of the money. So that is because people believe it has positive functionality. So by by a word, you give you some extra money. The government is trying to correct the I’ll help you in internalize. Ex the positive externality, right? Help you internalize. So when you make a decision, you will take into account 2 basically based on government calculation. They believe you know your action, your consumption for this product has social benefit. Okay, landing bomb and try to wrap up. Okay? So there’s a short discussion for the car in the Europe. The example is based on the I think, the second second, most expensive country, which is Denmark. Did I remove that? I think Denmark is second, most expensive in the world. Okay, and I’ll leave you to figure out who is the most expensive country. Okay, okay, I will. I will. Skip that discussion. So tax will be bare will will be covered more by the consumers for the buyer. If the demand is elastic and it will be, it will be covered more or bear more by the seller on the supply side. If demand is elastic. it’s very simple. Imagine like well, what do we mean by elastic? Well, elastic means buyer have outstanding right elastic meaning demand elastic meaning. Consumer can choose something else, and they’re happy to choose something else. Right? That’s what we mean by demand is very lasting. Right consumer will switch away immediately if you increase the price. If that’s the case, meaning consumers have good option, right and consumer will switch away a customer being very elastic. They are essentially the same thing. And if that’s the case, means consumer have bargaining power. So in that case, in this market, okay, so the cost of tech will be bare by the salad. Okay, that is basically demand, okay, you can think about you and your partners allocating some some work right? So whoever has a better outside option have a higher backup power. Okay. The other, who don’t have a good outside option. You have to bear more of the work. Okay? I think I will have some discussion which I already discussed in the class, and and I think I will leave you to the next lecture, I have 3 slides on on the labor market post. And Pre AI, how does that interact with labor market power? And the idea, basically, I mentioned earlier of this lecture that labor market to some extent you can think about us relatively homogeneous, especially for, like, if you’re not extremely like talented people like our associate. Undergrad. Right? You have access to the best education. And you know, like you, you are really good in terms of programming and all the knowledge of Ai. okay. And you are top tier. So. But from a lot of the work is actually quite homogeneous. Right? If you think about it being a copy editor, right? Some are being a translator, right? Like doing translation for people. And you know a lot of the work. They are the labor market like the services or the the output. The product is quite homogeneous. That is the most I would say dangerous, or that will be shocked or have large displacement potentially by by AI, and the reasons because AI is sort of offering very close substitute. Right? So because for that type for the market of that type of skills right? By which we mean very, basic, very homogeneous. You are competing with what you are competing with. some like a infinite supply right? Like in our perfect computation model, you have continuous if you are doing copy copy editing, right? So you’re competing with largely model or AI, which has infinite supply and increasingly low marginal cost. Right? So so more and more skills or tasks will become the type of market we motivated or we analyze in this class in today’s lecture, that is perfect competition. because the nature of the task is. it’s being homogenized by AI, because AI can do as good as anyone can. And there’s also infinite supply of these skills right? And and that supplies at very, very low marginal cost, right? And if you apply our analytical framework, and you will realize that that implies this market will be very, very competitive, and that will push because they are infinite. Entry right? And I will push the price down to the marginal cost. and the marginal cost for AI is crazily low, right comparing to the margin cost for for for a human being. So I will leave a little bit more discussion to the next lecture. But don’t worry. The last 3 slides will not be in the exam. So we conclude our 1st half. I just have a little bit of leftover discussion. I will leave to the second half. And again let me repeat our midterm information. Our midterm will be in a different venue. Okay. take a look and remind your friend who tend to forget things. Okay, like me. Okay, just remind your friend. Send them a Whatsapp or email or a chat. Our meeting will be a different venue. And it’s the same time as our lecture. And you will have 29. Mcq, okay, 29, Mcq, and you will have 100 min. Okay, so you can do the calculations on how how much time you have? Per question. You have 29 questions, Mcq, Abcd, and one correct answer. You will have to finish in 100 minutes, and you will have to. Take a pencil with you a reason most likely, because you need to correct your answers. and we will have close book exam. Okay? And the best reference for the midterm is our lecture notes, which cut which cover the 1st 5 topic demand elasticity cost, pricing and competition. Okay. and assignment assignment y, you can have the answers here. Where is he? Escape? We have some sort of suggested answer. Take a look. Okay, especially the the one that you didn’t do. Well, take a look on the solution. And okay, we will not have lecture next week, because next week is recess week. I hope you have a a very good week. I know many of you will be traveling. But I will see you back. No. Why are you laughing? Okay. okay. I may misunderstand something you can tell me later. Okay, so I will meet you the week after after recess week, and that is for our midterm. Okay, try to come early. If there’s anything you can. Still, if you come to the wrong venue, you can. Still, there’s still a chance to correct. Okay, so okay, that’s basically all. I will see you next. I will see you on March 7.th Okay, have a good recess week. Do we have questions? Okay, let me 1st stop recording.