WEBVTT 00:00:00.000 --> 00:00:01.988 align:middle line:90% [SQUEAKING] 00:00:01.988 --> 00:00:03.976 align:middle line:90% [RUSTLING] 00:00:03.976 --> 00:00:05.467 align:middle line:90% [CLICKING] 00:00:05.467 --> 00:00:11.207 align:middle line:90% 00:00:11.207 --> 00:00:12.540 align:middle line:90% GLENN ELLISON: Welcome to class. 00:00:12.540 --> 00:00:15.510 align:middle line:90% So this is 14.271. 00:00:15.510 --> 00:00:21.913 align:middle line:84% I should say that syllabus has a bunch of basic information. 00:00:21.913 --> 00:00:23.330 align:middle line:84% You'll notice the course is listed 00:00:23.330 --> 00:00:25.247 align:middle line:90% as co-taught by me and Tobias. 00:00:25.247 --> 00:00:27.080 align:middle line:84% COVID permitting, I'm going to be doing most 00:00:27.080 --> 00:00:28.372 align:middle line:90% of the lecturing in this class. 00:00:28.372 --> 00:00:31.820 align:middle line:84% I think the plan is I'll be doing 21 of the lectures, 00:00:31.820 --> 00:00:36.530 align:middle line:84% Tobias will be doing four, Jean Tirole will be doing two. 00:00:36.530 --> 00:00:42.140 align:middle line:84% And it's a basic course in industrial organization. 00:00:42.140 --> 00:00:45.950 align:middle line:84% I think these days, my primary responsibility in this class 00:00:45.950 --> 00:00:47.852 align:middle line:84% is to teach students who are going 00:00:47.852 --> 00:00:50.060 align:middle line:84% to write their thesis in industrial organization what 00:00:50.060 --> 00:00:52.350 align:middle line:84% they need to do to write an IO thesis. 00:00:52.350 --> 00:00:55.940 align:middle line:84% But the course has also traditionally been useful 00:00:55.940 --> 00:00:58.130 align:middle line:84% for many students in other fields. 00:00:58.130 --> 00:00:59.750 align:middle line:90% I think that's for two reasons. 00:00:59.750 --> 00:01:03.230 align:middle line:84% One is IO has always been at the forefront in economics 00:01:03.230 --> 00:01:06.410 align:middle line:84% and linking theory and empirics together. 00:01:06.410 --> 00:01:08.635 align:middle line:84% And then also in recent years, IO 00:01:08.635 --> 00:01:10.010 align:middle line:84% has developed a lot of techniques 00:01:10.010 --> 00:01:11.552 align:middle line:84% for what's referred to as structural, 00:01:11.552 --> 00:01:13.500 align:middle line:84% empirical industrial organization, which 00:01:13.500 --> 00:01:16.590 align:middle line:84% have become very, very popular in other fields. 00:01:16.590 --> 00:01:17.260 align:middle line:90% Anyway. 00:01:17.260 --> 00:01:21.990 align:middle line:84% So my goal here is really to teach the basics of IO theory 00:01:21.990 --> 00:01:24.298 align:middle line:84% and how IO people think about empirical work, which 00:01:24.298 --> 00:01:26.340 align:middle line:84% is sometimes different from how empirical work is 00:01:26.340 --> 00:01:28.110 align:middle line:90% done in different fields. 00:01:28.110 --> 00:01:32.010 align:middle line:84% The way I'm going to do that is you'll see on the syllabus 00:01:32.010 --> 00:01:35.730 align:middle line:84% we've got 12 topics or whatever we're going to cover. 00:01:35.730 --> 00:01:38.290 align:middle line:84% Almost all of them are going to get covered in two lectures. 00:01:38.290 --> 00:01:40.290 align:middle line:84% And typically in the first part of the semester, 00:01:40.290 --> 00:01:42.100 align:middle line:84% the pattern is going to be on Wednesday, 00:01:42.100 --> 00:01:43.900 align:middle line:84% I'm going to explain the theory of a topic. 00:01:43.900 --> 00:01:45.900 align:middle line:84% I'm going to go over the basic theory on a topic. 00:01:45.900 --> 00:01:47.733 align:middle line:84% And then on the Monday, I'm going to go over 00:01:47.733 --> 00:01:49.510 align:middle line:90% the empirical work on a topic. 00:01:49.510 --> 00:01:54.600 align:middle line:84% And my basic view of IO is IO is just an inherently theoretical 00:01:54.600 --> 00:01:56.100 align:middle line:90% field. 00:01:56.100 --> 00:01:59.970 align:middle line:84% There are many fields in which empirical work 00:01:59.970 --> 00:02:02.580 align:middle line:84% and natural experiments testing policies 00:02:02.580 --> 00:02:04.770 align:middle line:84% are what you want to do in studying that field. 00:02:04.770 --> 00:02:07.830 align:middle line:84% In IO, the questions we get asked are 00:02:07.830 --> 00:02:12.290 align:middle line:84% should Facebook be required to divest itself of Instagram? 00:02:12.290 --> 00:02:18.020 align:middle line:84% Or is it OK for Google to be paying Apple $12 billion a year 00:02:18.020 --> 00:02:21.600 align:middle line:84% to make Google the default search engine on iPhones? 00:02:21.600 --> 00:02:24.218 align:middle line:84% And these are not things that can be addressed 00:02:24.218 --> 00:02:25.260 align:middle line:90% with natural experiments. 00:02:25.260 --> 00:02:28.130 align:middle line:84% You can't create 100 different worlds 00:02:28.130 --> 00:02:31.760 align:middle line:84% and break up Facebook and Instagram in half of them 00:02:31.760 --> 00:02:34.610 align:middle line:84% and not break up Facebook and Instagram in the other half 00:02:34.610 --> 00:02:37.670 align:middle line:84% of them and then do a comparison of means 10 years later 00:02:37.670 --> 00:02:39.290 align:middle line:84% and see which was the right decision. 00:02:39.290 --> 00:02:42.510 align:middle line:84% It's a one time decision that will get made or not. 00:02:42.510 --> 00:02:45.300 align:middle line:84% And so you basically make decisions like that. 00:02:45.300 --> 00:02:47.975 align:middle line:84% You need to have a theoretical understanding 00:02:47.975 --> 00:02:50.600 align:middle line:84% of how the world works and how firms competing with other firms 00:02:50.600 --> 00:02:53.030 align:middle line:84% work and what do you think would happen on the two 00:02:53.030 --> 00:02:54.720 align:middle line:90% different market structures. 00:02:54.720 --> 00:02:59.000 align:middle line:84% And then you may never find out if you were right or wrong. 00:02:59.000 --> 00:03:02.600 align:middle line:84% While I think IO is inherently a theoretical field, almost 00:03:02.600 --> 00:03:06.590 align:middle line:84% all students getting good jobs in IO in the last 15, 00:03:06.590 --> 00:03:09.380 align:middle line:90% 20 years are doing empirical IO. 00:03:09.380 --> 00:03:13.115 align:middle line:84% And in part, this is because what happens is people in IO 00:03:13.115 --> 00:03:13.740 align:middle line:90% make up models. 00:03:13.740 --> 00:03:17.160 align:middle line:84% And the question is you can make up models that 00:03:17.160 --> 00:03:18.580 align:middle line:90% predict all kinds of things. 00:03:18.580 --> 00:03:21.000 align:middle line:84% Then you have to ask which model's assumptions are 00:03:21.000 --> 00:03:24.870 align:middle line:84% correct in terms of how people behave, how firms behave, 00:03:24.870 --> 00:03:27.930 align:middle line:84% and how large are those effects empirically. 00:03:27.930 --> 00:03:30.630 align:middle line:84% And so what typically happens, there's a lot of interplay in IO 00:03:30.630 --> 00:03:32.730 align:middle line:84% between you develop a theory, you then 00:03:32.730 --> 00:03:35.280 align:middle line:84% do empirical work that tries to examine the theory in places 00:03:35.280 --> 00:03:36.697 align:middle line:84% where it should or shouldn't work, 00:03:36.697 --> 00:03:38.850 align:middle line:84% and you find out how important are the effects 00:03:38.850 --> 00:03:40.840 align:middle line:90% the different models predict. 00:03:40.840 --> 00:03:42.840 align:middle line:84% Are the assumptions and models good or should we 00:03:42.840 --> 00:03:46.110 align:middle line:84% be looking at what-- in some sense, in any good IO empirical 00:03:46.110 --> 00:03:47.310 align:middle line:90% work, something doesn't fit. 00:03:47.310 --> 00:03:49.230 align:middle line:84% And then what doesn't fit gives you 00:03:49.230 --> 00:03:51.150 align:middle line:84% the idea of how do you refine the model, what 00:03:51.150 --> 00:03:52.525 align:middle line:84% else should be added to the model 00:03:52.525 --> 00:03:54.120 align:middle line:84% to get a better predictive model. 00:03:54.120 --> 00:03:56.430 align:middle line:84% So let me go ahead and get started. 00:03:56.430 --> 00:03:57.910 align:middle line:90% And I'm going to get started-- 00:03:57.910 --> 00:04:00.360 align:middle line:84% I should say normally when you design a class, 00:04:00.360 --> 00:04:02.550 align:middle line:84% you want to have the first class be really exciting 00:04:02.550 --> 00:04:05.850 align:middle line:84% and have something intriguing for students to get hooked 00:04:05.850 --> 00:04:07.110 align:middle line:90% on the class. 00:04:07.110 --> 00:04:08.600 align:middle line:90% That's not the case with IO. 00:04:08.600 --> 00:04:09.850 align:middle line:90% I have to start at square one. 00:04:09.850 --> 00:04:11.590 align:middle line:84% I have to start with monopoly pricing. 00:04:11.590 --> 00:04:13.170 align:middle line:84% I'm sure many of you have seen monopoly pricing 00:04:13.170 --> 00:04:15.300 align:middle line:84% in your undergraduate intermediate micro or principles 00:04:15.300 --> 00:04:15.820 align:middle line:90% classes. 00:04:15.820 --> 00:04:19.529 align:middle line:84% But I'm going to start with the beginning with monopolies 00:04:19.529 --> 00:04:22.230 align:middle line:84% pricing and then build from there. 00:04:22.230 --> 00:04:23.490 align:middle line:90% So anyway, first class. 00:04:23.490 --> 00:04:26.580 align:middle line:90% So this is 14.01. 00:04:26.580 --> 00:04:28.943 align:middle line:84% Monopoly seller of a single product. 00:04:28.943 --> 00:04:30.360 align:middle line:84% In this class, I'm generally going 00:04:30.360 --> 00:04:35.610 align:middle line:84% to be putting quantity on the x-axis and price on the y-axis. 00:04:35.610 --> 00:04:38.520 align:middle line:84% So a monopolist has this downward facing, 00:04:38.520 --> 00:04:44.880 align:middle line:84% this inverse demand curve p of x if it sells x units. 00:04:44.880 --> 00:04:48.100 align:middle line:84% Monopolist has a total cost function c of x. 00:04:48.100 --> 00:04:50.760 align:middle line:84% So if they produce x units, the total cost of production 00:04:50.760 --> 00:04:52.080 align:middle line:90% is c of x. 00:04:52.080 --> 00:04:54.060 align:middle line:90% What does the monopolist do? 00:04:54.060 --> 00:04:56.130 align:middle line:84% You've seen this many times, you maximize over 00:04:56.130 --> 00:04:59.700 align:middle line:90% x, x times p of x minus c of x. 00:04:59.700 --> 00:05:01.890 align:middle line:84% I take the first order condition and I 00:05:01.890 --> 00:05:03.220 align:middle line:90% get this familiar equation. 00:05:03.220 --> 00:05:06.850 align:middle line:84% Marginal revenue minus marginal cost equals 0. 00:05:06.850 --> 00:05:11.930 align:middle line:84% So the monopolist sets marginal revenue equal to marginal cost. 00:05:11.930 --> 00:05:13.400 align:middle line:90% Why am I starting here? 00:05:13.400 --> 00:05:15.840 align:middle line:84% Well, even in IO, we do things a little differently. 00:05:15.840 --> 00:05:18.830 align:middle line:84% The way people typically think about this first order condition 00:05:18.830 --> 00:05:23.060 align:middle line:84% is instead of putting marginal revenue on the left and marginal 00:05:23.060 --> 00:05:27.380 align:middle line:84% cost on the right, I'm going to take price minus marginal cost 00:05:27.380 --> 00:05:31.070 align:middle line:84% and move the minus x p prime at x to the other side. 00:05:31.070 --> 00:05:33.470 align:middle line:84% And then I divide both sides by p 00:05:33.470 --> 00:05:37.190 align:middle line:84% of x and I get price minus marginal cost over price 00:05:37.190 --> 00:05:40.466 align:middle line:90% equals this. 00:05:40.466 --> 00:05:43.460 align:middle line:84% And what's over here on the right side? 00:05:43.460 --> 00:05:47.330 align:middle line:90% p prime at x, x over p of x. 00:05:47.330 --> 00:05:54.530 align:middle line:84% So you can think of that as x over p of x, 1 over p prime 00:05:54.530 --> 00:05:57.380 align:middle line:90% at x. 00:05:57.380 --> 00:06:00.470 align:middle line:84% And then if you think about the inverse function theorem, 00:06:00.470 --> 00:06:07.340 align:middle line:84% 1 over p prime at x is kind of like dq / dp. 00:06:07.340 --> 00:06:09.500 align:middle line:90% Because p is the-- 00:06:09.500 --> 00:06:12.710 align:middle line:90% 00:06:12.710 --> 00:06:14.840 align:middle line:84% if you think about p as the inverse function of q, 00:06:14.840 --> 00:06:17.670 align:middle line:84% so this is like x over p prime at x. 00:06:17.670 --> 00:06:22.400 align:middle line:84% And so if I divide this, dq / dp times p over q, 00:06:22.400 --> 00:06:25.400 align:middle line:84% this is the elasticity of demand. 00:06:25.400 --> 00:06:29.880 align:middle line:84% So what a monopolist does is it sets its price 00:06:29.880 --> 00:06:32.780 align:middle line:84% so that price minus marginal cost over price 00:06:32.780 --> 00:06:36.720 align:middle line:84% is the negative of 1 over the elasticity of demand. 00:06:36.720 --> 00:06:40.040 align:middle line:84% And I start here because in macroeconomics, they 00:06:40.040 --> 00:06:42.200 align:middle line:84% seem to have this thing that when demand goes up, 00:06:42.200 --> 00:06:45.300 align:middle line:84% price goes up, and when demand goes down, price goes down. 00:06:45.300 --> 00:06:47.600 align:middle line:90% That doesn't happen in IO. 00:06:47.600 --> 00:06:49.940 align:middle line:84% When you think about firms with market power, 00:06:49.940 --> 00:06:54.920 align:middle line:84% pricing is based on demand elasticities, not demand levels. 00:06:54.920 --> 00:07:00.260 align:middle line:84% And on this left side, this is something that in IO we 00:07:00.260 --> 00:07:02.240 align:middle line:90% call it the Lerner index. 00:07:02.240 --> 00:07:04.123 align:middle line:84% Ignoring the marginal versus average, 00:07:04.123 --> 00:07:06.540 align:middle line:84% this is like what accountants would call the gross margin. 00:07:06.540 --> 00:07:08.750 align:middle line:84% This is the markup that you're charging 00:07:08.750 --> 00:07:10.800 align:middle line:90% as a fraction of the price. 00:07:10.800 --> 00:07:15.750 align:middle line:84% So if you charge double your marginal cost, that's 00:07:15.750 --> 00:07:18.780 align:middle line:84% a 50% markup, because 50% of your price is markup, 00:07:18.780 --> 00:07:20.320 align:middle line:90% 50% is cost. 00:07:20.320 --> 00:07:24.180 align:middle line:84% So this says the Lerner index is negative of 1 00:07:24.180 --> 00:07:26.560 align:middle line:90% over the elasticity of demand. 00:07:26.560 --> 00:07:29.530 align:middle line:84% So if, for instance, the elasticity of demand is-- 00:07:29.530 --> 00:07:33.940 align:middle line:84% if epsilon is negative 2, then this would say you do price 00:07:33.940 --> 00:07:38.770 align:middle line:84% minus marginal cost over price is 1/2. 00:07:38.770 --> 00:07:42.890 align:middle line:84% So if you're a monopolist, your elasticity of demand is 2. 00:07:42.890 --> 00:07:44.530 align:middle line:90% Your margin is 50%. 00:07:44.530 --> 00:07:46.360 align:middle line:84% You double your marginal cost when you 00:07:46.360 --> 00:07:47.840 align:middle line:90% think of your monopoly price. 00:07:47.840 --> 00:07:54.490 align:middle line:84% If your elasticity is 4, you do your 25% Lerner index. 00:07:54.490 --> 00:07:58.930 align:middle line:84% Lerner index is 1/4, which means you're raising your price by 1/3 00:07:58.930 --> 00:07:59.805 align:middle line:90% over cost. 00:07:59.805 --> 00:08:04.890 align:middle line:90% 00:08:04.890 --> 00:08:09.450 align:middle line:84% In 14.01, you certainly also did a diagram like this. 00:08:09.450 --> 00:08:11.460 align:middle line:90% So what happens? 00:08:11.460 --> 00:08:14.020 align:middle line:84% Firm sets marginal revenue equal to marginal cost. 00:08:14.020 --> 00:08:16.050 align:middle line:90% They have this monopoly price. 00:08:16.050 --> 00:08:18.600 align:middle line:84% They earn a profit on all the units 00:08:18.600 --> 00:08:20.250 align:middle line:84% they sell at a cost that's greater 00:08:20.250 --> 00:08:22.420 align:middle line:90% than the cost of producing them. 00:08:22.420 --> 00:08:24.228 align:middle line:90% And so this is the profit. 00:08:24.228 --> 00:08:26.520 align:middle line:84% This is the difference between the consumer willingness 00:08:26.520 --> 00:08:28.728 align:middle line:84% to pay for each unit and the price that they've paid. 00:08:28.728 --> 00:08:30.280 align:middle line:90% So this is consumer surplus. 00:08:30.280 --> 00:08:33.620 align:middle line:90% This is deadweight loss. 00:08:33.620 --> 00:08:38.130 align:middle line:84% Typically, a lot of IO is about market power is a bad thing. 00:08:38.130 --> 00:08:39.720 align:middle line:90% We want to prevent market power. 00:08:39.720 --> 00:08:42.770 align:middle line:84% Why do we want to prevent market power? 00:08:42.770 --> 00:08:45.500 align:middle line:84% Classic explanation for why you want to prevent market power 00:08:45.500 --> 00:08:46.740 align:middle line:90% is this triangle. 00:08:46.740 --> 00:08:50.717 align:middle line:84% You have this deadweight loss is a social inefficiency. 00:08:50.717 --> 00:08:52.550 align:middle line:84% People are willing to pay more than the cost 00:08:52.550 --> 00:08:54.260 align:middle line:84% of producing the goods, and yet those goods 00:08:54.260 --> 00:08:55.830 align:middle line:84% don't get produced and sold to them. 00:08:55.830 --> 00:08:57.420 align:middle line:84% That's socially wasteful that we have 00:08:57.420 --> 00:09:00.003 align:middle line:84% these opportunities for gains from trade that aren't realized. 00:09:00.003 --> 00:09:02.930 align:middle line:90% 00:09:02.930 --> 00:09:06.973 align:middle line:84% While people often focus on deadweight loss, 00:09:06.973 --> 00:09:09.140 align:middle line:84% there are a lot of other reasons why you might think 00:09:09.140 --> 00:09:10.980 align:middle line:90% monopolies are also a bad idea. 00:09:10.980 --> 00:09:16.497 align:middle line:84% So many of these are harder to actually explain economically, 00:09:16.497 --> 00:09:18.330 align:middle line:84% to realize I think they should be recurring. 00:09:18.330 --> 00:09:19.997 align:middle line:84% But one thing you hear people talk about 00:09:19.997 --> 00:09:23.210 align:middle line:90% is productive inefficiency. 00:09:23.210 --> 00:09:25.940 align:middle line:84% Certainly the question is, are monopolists 00:09:25.940 --> 00:09:30.140 align:middle line:84% going to produce things at the lowest cost possible? 00:09:30.140 --> 00:09:31.940 align:middle line:84% Your basic intuition should be yes. 00:09:31.940 --> 00:09:33.500 align:middle line:84% Monopolist gets all the extra money 00:09:33.500 --> 00:09:34.800 align:middle line:90% they make from cost savings. 00:09:34.800 --> 00:09:36.800 align:middle line:84% So monopolists ought to want to produce the most 00:09:36.800 --> 00:09:38.697 align:middle line:90% efficient technology possible. 00:09:38.697 --> 00:09:40.280 align:middle line:84% There are lots of discussions, though. 00:09:40.280 --> 00:09:42.830 align:middle line:84% If you think about the old Soviet Union, 00:09:42.830 --> 00:09:45.020 align:middle line:84% they had factories making things out of titanium 00:09:45.020 --> 00:09:47.187 align:middle line:84% that could have been made out of steel and that cost 00:09:47.187 --> 00:09:48.960 align:middle line:84% 5 times as much as they cost in the West. 00:09:48.960 --> 00:09:50.110 align:middle line:90% Why did that happen? 00:09:50.110 --> 00:09:52.772 align:middle line:84% Well, one thing that you hear is that if you 00:09:52.772 --> 00:09:54.980 align:middle line:84% have a firm that's a monopoly, they don't have anyone 00:09:54.980 --> 00:09:56.140 align:middle line:90% to compare themselves to. 00:09:56.140 --> 00:09:59.420 align:middle line:84% They don't realize there are better production technologies. 00:09:59.420 --> 00:10:01.580 align:middle line:84% Employees can be lazy in figuring out 00:10:01.580 --> 00:10:04.020 align:middle line:84% how to produce things and no one ever catches them. 00:10:04.020 --> 00:10:08.330 align:middle line:84% So productive inefficiency could be a cost of monopoly. 00:10:08.330 --> 00:10:09.900 align:middle line:90% Quality distortions. 00:10:09.900 --> 00:10:12.360 align:middle line:84% Again, this is more an informal complaint. 00:10:12.360 --> 00:10:14.720 align:middle line:84% But people look at Microsoft Windows 00:10:14.720 --> 00:10:18.050 align:middle line:84% and are often thinking of 30 different criticisms 00:10:18.050 --> 00:10:20.580 align:middle line:84% of Microsoft Windows and why does it suck in so many ways. 00:10:20.580 --> 00:10:21.840 align:middle line:90% Why can't Excel do this? 00:10:21.840 --> 00:10:23.280 align:middle line:90% Why can't Excel do this? 00:10:23.280 --> 00:10:25.250 align:middle line:84% Is it because Microsoft is a monopoly 00:10:25.250 --> 00:10:27.960 align:middle line:84% and has no incentive to improve the quality of its products 00:10:27.960 --> 00:10:30.050 align:middle line:90% because you've got to buy them? 00:10:30.050 --> 00:10:31.800 align:middle line:84% You've got to put Windows on your computer 00:10:31.800 --> 00:10:34.710 align:middle line:84% anyway, even if you don't like it. 00:10:34.710 --> 00:10:38.250 align:middle line:84% Rent seeking is an important thing. 00:10:38.250 --> 00:10:42.840 align:middle line:84% Sometimes we have monopolies for regulatory reasons 00:10:42.840 --> 00:10:45.070 align:middle line:90% or for capacity constraints. 00:10:45.070 --> 00:10:47.370 align:middle line:84% So, for instance, there's a limited number 00:10:47.370 --> 00:10:53.040 align:middle line:84% of licenses to provide cell phone service in every city. 00:10:53.040 --> 00:10:56.177 align:middle line:84% In the old days, when cell phone service first started, 00:10:56.177 --> 00:10:58.260 align:middle line:84% these were actually-- when the United States first 00:10:58.260 --> 00:11:00.210 align:middle line:84% adopted cell phones, there were going 00:11:00.210 --> 00:11:02.505 align:middle line:84% to be two licenses in every city. 00:11:02.505 --> 00:11:04.380 align:middle line:84% And they had what people refer to as a beauty 00:11:04.380 --> 00:11:08.028 align:middle line:84% contest to allocate those licenses. 00:11:08.028 --> 00:11:09.820 align:middle line:84% That is, firms didn't have to pay for them. 00:11:09.820 --> 00:11:13.988 align:middle line:84% You just had to apply to provide cell phone service 00:11:13.988 --> 00:11:15.780 align:middle line:84% and the government would pick the firm that 00:11:15.780 --> 00:11:19.047 align:middle line:84% seemed best, socially efficient to provide the cell phone 00:11:19.047 --> 00:11:20.880 align:middle line:84% service, and they would award those licenses 00:11:20.880 --> 00:11:23.310 align:middle line:90% to those winning parties. 00:11:23.310 --> 00:11:25.650 align:middle line:84% Often these winning parties were perhaps 00:11:25.650 --> 00:11:26.980 align:middle line:90% not the ones you would think. 00:11:26.980 --> 00:11:28.630 align:middle line:84% So, for instance, I don't know if the story is true. 00:11:28.630 --> 00:11:30.090 align:middle line:84% There's a story that in Washington, DC, 00:11:30.090 --> 00:11:32.298 align:middle line:84% one of the two licenses went to Katharine Graham, who 00:11:32.298 --> 00:11:33.883 align:middle line:84% was the editor of the Washington Post. 00:11:33.883 --> 00:11:35.550 align:middle line:84% Why would she know more than anyone else 00:11:35.550 --> 00:11:36.790 align:middle line:84% about running a cell phone company? 00:11:36.790 --> 00:11:39.030 align:middle line:84% Well, she ran a newspaper that would write stories 00:11:39.030 --> 00:11:40.030 align:middle line:90% about the politicians. 00:11:40.030 --> 00:11:41.670 align:middle line:84% So when they're thinking of who to give the license to, 00:11:41.670 --> 00:11:42.970 align:middle line:90% they give it to her. 00:11:42.970 --> 00:11:45.270 align:middle line:84% But in general, in many situations, 00:11:45.270 --> 00:11:46.980 align:middle line:84% you're going to find that if there 00:11:46.980 --> 00:11:48.930 align:middle line:84% are monopoly rents to be given out, 00:11:48.930 --> 00:11:52.122 align:middle line:84% someone has the power to give out the monopoly rents. 00:11:52.122 --> 00:11:53.580 align:middle line:84% And then when someone has the power 00:11:53.580 --> 00:11:57.150 align:middle line:84% to give out the monopoly rents, those people 00:11:57.150 --> 00:11:59.770 align:middle line:84% may take bribes from the people who are giving them. 00:11:59.770 --> 00:12:01.590 align:middle line:84% It may be that you have something 00:12:01.590 --> 00:12:04.470 align:middle line:84% akin to a political campaign where you're doing advertising 00:12:04.470 --> 00:12:07.440 align:middle line:84% or lobbying, and you have all these socially wasteful 00:12:07.440 --> 00:12:12.060 align:middle line:84% expenditures to try to be the one who earns these profits. 00:12:12.060 --> 00:12:13.920 align:middle line:84% If you imagine a bidding game where 00:12:13.920 --> 00:12:18.180 align:middle line:84% if everybody is going to give a bribe to the politician 00:12:18.180 --> 00:12:21.852 align:middle line:84% and the biggest bribe wins, what's your bribe going to be? 00:12:21.852 --> 00:12:24.060 align:middle line:84% You're willing to pay a bribe equal to the discounted 00:12:24.060 --> 00:12:27.990 align:middle line:84% value of all future profits that you're going to earn. 00:12:27.990 --> 00:12:31.530 align:middle line:84% In that game, you'd expect all of the profits 00:12:31.530 --> 00:12:34.110 align:middle line:84% would also then get dissipated socially. 00:12:34.110 --> 00:12:36.450 align:middle line:84% Well, actually, now if it's a bribe, 00:12:36.450 --> 00:12:39.370 align:middle line:84% you believe in efficient markets, it's just a transfer. 00:12:39.370 --> 00:12:42.690 align:middle line:84% And so it's just a transfer of wealth from consumers to the one 00:12:42.690 --> 00:12:44.250 align:middle line:90% person who's taking the bribe. 00:12:44.250 --> 00:12:48.480 align:middle line:84% But if the bribes are given in a socially wasteful way, 00:12:48.480 --> 00:12:53.250 align:middle line:84% all of this could also be a loss of monopoly power. 00:12:53.250 --> 00:12:54.900 align:middle line:84% And then my final comment on bribes 00:12:54.900 --> 00:12:57.010 align:middle line:84% brings up this sort of equity concern. 00:12:57.010 --> 00:13:00.180 align:middle line:84% Another problem with monopoly is that in a competitive market, 00:13:00.180 --> 00:13:01.650 align:middle line:84% the consumers will be paying this. 00:13:01.650 --> 00:13:03.510 align:middle line:84% In a monopoly, they're paying this. 00:13:03.510 --> 00:13:04.870 align:middle line:90% Consumers paying more. 00:13:04.870 --> 00:13:07.480 align:middle line:84% It's a transfer from consumers to firms. 00:13:07.480 --> 00:13:09.810 align:middle line:84% If you believe that the consumers have a higher 00:13:09.810 --> 00:13:12.090 align:middle line:84% marginal utility of income than the shareholders 00:13:12.090 --> 00:13:14.815 align:middle line:84% of the firm, that's also a concern 00:13:14.815 --> 00:13:15.940 align:middle line:90% we should be worried about. 00:13:15.940 --> 00:13:18.070 align:middle line:84% I should say that in IO, I mentioned that today, 00:13:18.070 --> 00:13:20.070 align:middle line:84% I'm never going to mention it again pretty much. 00:13:20.070 --> 00:13:24.010 align:middle line:84% Jean may talk about it when he gives his guest lectures. 00:13:24.010 --> 00:13:26.657 align:middle line:84% But in IO, all of these income effects 00:13:26.657 --> 00:13:28.740 align:middle line:84% you see in public finance, whatever, those are all 00:13:28.740 --> 00:13:29.590 align:middle line:90% going to disappear. 00:13:29.590 --> 00:13:33.090 align:middle line:84% So in IO, pretty much everybody has 00:13:33.090 --> 00:13:36.960 align:middle line:84% quasi linear preferences and everyone has preferences 00:13:36.960 --> 00:13:39.480 align:middle line:84% that look like v minus p, where everybody's 00:13:39.480 --> 00:13:42.128 align:middle line:84% got a marginal utility of income of 1 that's constant 00:13:42.128 --> 00:13:44.170 align:middle line:84% and everyone has some value from buying the good. 00:13:44.170 --> 00:13:46.620 align:middle line:84% And we just do additively separable utility 00:13:46.620 --> 00:13:49.320 align:middle line:84% and don't really worry about income effects and stuff. 00:13:49.320 --> 00:13:53.430 align:middle line:84% And obviously that's not to say that equity concerns 00:13:53.430 --> 00:13:55.440 align:middle line:84% and transfers across consumers aren't important, 00:13:55.440 --> 00:13:57.900 align:middle line:84% but it's just I think people in IO 00:13:57.900 --> 00:13:59.650 align:middle line:84% have made the decision that you don't want 00:13:59.650 --> 00:14:02.740 align:middle line:84% to be every single model talking about the incentive 00:14:02.740 --> 00:14:07.132 align:middle line:84% to distort what you have the firms doing in order to benefit 00:14:07.132 --> 00:14:09.340 align:middle line:84% the consumers with lower marginal utilities of income 00:14:09.340 --> 00:14:11.298 align:middle line:84% instead of higher marginal utilities of income. 00:14:11.298 --> 00:14:14.680 align:middle line:84% So we just do this, and you can think about equity concerns 00:14:14.680 --> 00:14:15.730 align:middle line:90% on top of it. 00:14:15.730 --> 00:14:17.410 align:middle line:84% But pretty much all semester I'm going 00:14:17.410 --> 00:14:20.750 align:middle line:90% to have that kind of utility. 00:14:20.750 --> 00:14:22.640 align:middle line:90% OK. 00:14:22.640 --> 00:14:24.840 align:middle line:84% That's your basic monopoly pricing model. 00:14:24.840 --> 00:14:27.500 align:middle line:84% What are some of the other things you need to know? 00:14:27.500 --> 00:14:29.990 align:middle line:84% First of all, most firms in the real world 00:14:29.990 --> 00:14:33.650 align:middle line:84% sell multiple goods instead of a single good. 00:14:33.650 --> 00:14:36.720 align:middle line:84% So the way I'm going to think about this is simplest example. 00:14:36.720 --> 00:14:40.640 align:middle line:84% Suppose the monopolist has two goods to sell. 00:14:40.640 --> 00:14:43.080 align:middle line:84% Now I'm putting price as the argument. 00:14:43.080 --> 00:14:46.950 align:middle line:84% So demand for good 1 is a function of p1 and p2. 00:14:46.950 --> 00:14:51.890 align:middle line:84% Demand for good 2 is also a function of p1 and p2. 00:14:51.890 --> 00:14:54.630 align:middle line:84% Let me stick with constant marginal costs here. 00:14:54.630 --> 00:14:57.980 align:middle line:84% So the total cost of producing x1 units of good 1 and x2 00:14:57.980 --> 00:15:00.996 align:middle line:84% units of good 2 is just c1 x1 plus c2 x2. 00:15:00.996 --> 00:15:03.740 align:middle line:90% 00:15:03.740 --> 00:15:06.060 align:middle line:84% So what's the monopolist's problem? 00:15:06.060 --> 00:15:10.820 align:middle line:84% It's now maximizing over the two prices. 00:15:10.820 --> 00:15:12.920 align:middle line:84% Price 1 minus marginal cost 1 times 00:15:12.920 --> 00:15:19.790 align:middle line:84% the number of units of good 1 plus p2 minus c2 dx times x2. 00:15:19.790 --> 00:15:22.080 align:middle line:84% What do the first order conditions look like? 00:15:22.080 --> 00:15:24.560 align:middle line:84% Well, if I differentiate, think about differentiating with 00:15:24.560 --> 00:15:28.670 align:middle line:84% respect to the price of good 1, if I differentiate with respect 00:15:28.670 --> 00:15:33.320 align:middle line:84% to p1, I'm going to get p1 minus c1 d(x1)/d(p1). 00:15:33.320 --> 00:15:38.990 align:middle line:84% Plus this times the derivative of that, which is x1 of P1 P2. 00:15:38.990 --> 00:15:42.530 align:middle line:84% But now I'm going to get an extra term that I wasn't getting 00:15:42.530 --> 00:15:46.460 align:middle line:84% before, which is d(x2)/d(p1) times p2 minus c2. 00:15:46.460 --> 00:15:50.910 align:middle line:90% 00:15:50.910 --> 00:15:54.240 align:middle line:84% So one way to think about what we're getting is if I now do 00:15:54.240 --> 00:15:56.550 align:middle line:84% the same thing, if I divide by x, 00:15:56.550 --> 00:16:04.770 align:middle line:84% I get price minus cost plus x1 over d(x1)/d(p1) plus p2 minus 00:16:04.770 --> 00:16:10.140 align:middle line:84% c2 times d(x2)/d(p1) over d(x1)/d(p1) equals 0. 00:16:10.140 --> 00:16:12.540 align:middle line:84% So this is the same first order condition 00:16:12.540 --> 00:16:17.730 align:middle line:84% I had before, but with the addition of this extra term. 00:16:17.730 --> 00:16:21.190 align:middle line:84% People in IO refer to this as the diversion ratio. 00:16:21.190 --> 00:16:25.980 align:middle line:84% This is when I raise the price of good 1, 00:16:25.980 --> 00:16:30.240 align:middle line:84% I'm going to sell fewer units of good 2. 00:16:30.240 --> 00:16:32.440 align:middle line:84% When I sell fewer units of good 2, 00:16:32.440 --> 00:16:34.320 align:middle line:84% sorry, when I sell price of good 1, 00:16:34.320 --> 00:16:36.480 align:middle line:84% I'm going to sell fewer units of good 1. 00:16:36.480 --> 00:16:40.150 align:middle line:84% But when I sell fewer units of good 1, 00:16:40.150 --> 00:16:43.480 align:middle line:84% that's going to affect my sales of good 2. 00:16:43.480 --> 00:16:46.570 align:middle line:84% And so it can either raise my sales of good 2 00:16:46.570 --> 00:16:48.630 align:middle line:90% or lower my sales of good 2. 00:16:48.630 --> 00:16:52.960 align:middle line:84% But I have to take in that account for every unit of goods 00:16:52.960 --> 00:16:53.990 align:middle line:90% that I don't sell. 00:16:53.990 --> 00:16:57.650 align:middle line:84% It gets diverted possibly to a sale of another good. 00:16:57.650 --> 00:17:00.250 align:middle line:84% And I have to take the profits I would earn on those goods 00:17:00.250 --> 00:17:03.380 align:middle line:84% into account when I'm doing the first one. 00:17:03.380 --> 00:17:08.920 align:middle line:84% And so this gives me a different markup formula. 00:17:08.920 --> 00:17:13.849 align:middle line:84% Price minus marginal cost over price is minus 1 over epsilon. 00:17:13.849 --> 00:17:18.970 align:middle line:84% The elasticity of the good plus this other effect, which 00:17:18.970 --> 00:17:21.760 align:middle line:84% comes from thinking about the effect of my change 00:17:21.760 --> 00:17:24.330 align:middle line:84% in price of good 1 on sales of good 2. 00:17:24.330 --> 00:17:28.850 align:middle line:90% 00:17:28.850 --> 00:17:31.708 align:middle line:84% I should say before I go to the next page, one way you can also 00:17:31.708 --> 00:17:34.250 align:middle line:84% think about this is just you have to think about your pricing 00:17:34.250 --> 00:17:35.010 align:middle line:90% as a system. 00:17:35.010 --> 00:17:37.102 align:middle line:90% So, for instance, if you're-- 00:17:37.102 --> 00:17:38.810 align:middle line:84% the worst thing in the world you could do 00:17:38.810 --> 00:17:46.430 align:middle line:84% is imagine you're advising Apple on selling both blue iPhones 00:17:46.430 --> 00:17:48.090 align:middle line:90% and silver iPhones. 00:17:48.090 --> 00:17:51.260 align:middle line:84% And you do a study of the price elasticity of blue iPhones 00:17:51.260 --> 00:17:53.673 align:middle line:84% by raising the price of blue iPhones by $20 00:17:53.673 --> 00:17:56.090 align:middle line:84% and lowering it by $20 and seeing what that does to sales, 00:17:56.090 --> 00:17:57.930 align:middle line:90% holding all other prices fixed. 00:17:57.930 --> 00:18:00.530 align:middle line:84% You'll find that the blue iPhone is incredibly price 00:18:00.530 --> 00:18:04.595 align:middle line:84% sensitive price, because people switch and buy the silver one. 00:18:04.595 --> 00:18:06.970 align:middle line:84% That does not mean that you want to have a really, really 00:18:06.970 --> 00:18:08.392 align:middle line:90% low markup on the blue iPhone. 00:18:08.392 --> 00:18:09.850 align:middle line:84% You have to think of it as a system 00:18:09.850 --> 00:18:12.850 align:middle line:84% and what's the equilibrium solution to blue iPhone, 00:18:12.850 --> 00:18:15.670 align:middle line:84% silver iPhone, red iPhone, green iPhone, whatever. 00:18:15.670 --> 00:18:17.420 align:middle line:84% And think of all those things as a system. 00:18:17.420 --> 00:18:18.962 align:middle line:84% If you look at each one individually, 00:18:18.962 --> 00:18:20.200 align:middle line:90% you'll get it terribly wrong. 00:18:20.200 --> 00:18:22.000 align:middle line:84% And so in general, any time you're 00:18:22.000 --> 00:18:26.620 align:middle line:84% doing pricing of multiple goods, if there 00:18:26.620 --> 00:18:28.672 align:middle line:84% are very strong diversion ratios between them, 00:18:28.672 --> 00:18:30.130 align:middle line:84% you always want to make sure you're 00:18:30.130 --> 00:18:32.890 align:middle line:84% thinking of it as a system, not in terms of each one 00:18:32.890 --> 00:18:35.320 align:middle line:90% individually as its epsilon. 00:18:35.320 --> 00:18:37.990 align:middle line:84% But when these diversion ratios are smaller, 00:18:37.990 --> 00:18:40.300 align:middle line:84% you can think of it as it is 1 over epsilon 00:18:40.300 --> 00:18:42.310 align:middle line:84% pricing with some increase or decrease 00:18:42.310 --> 00:18:45.123 align:middle line:84% based on whether there is an effect of changing 00:18:45.123 --> 00:18:46.290 align:middle line:90% sales on the other products. 00:18:46.290 --> 00:18:51.050 align:middle line:90% 00:18:51.050 --> 00:18:51.880 align:middle line:90% OK, anyway. 00:18:51.880 --> 00:18:55.310 align:middle line:90% 00:18:55.310 --> 00:18:57.330 align:middle line:84% OK, so what does this formula look like? 00:18:57.330 --> 00:19:03.200 align:middle line:84% So my price is inverse elasticity minus price minus 00:19:03.200 --> 00:19:06.680 align:middle line:90% cost over price d(xi)/d(pi). 00:19:06.680 --> 00:19:09.200 align:middle line:84% If you think about the normal case, in the normal case, 00:19:09.200 --> 00:19:12.770 align:middle line:84% goods are sold at positive markups. 00:19:12.770 --> 00:19:15.450 align:middle line:84% So this would be a positive number. 00:19:15.450 --> 00:19:17.370 align:middle line:90% This would be a positive number. 00:19:17.370 --> 00:19:19.770 align:middle line:84% The d(xi)/d(pi) is a negative number. 00:19:19.770 --> 00:19:20.790 align:middle line:90% So this is negative. 00:19:20.790 --> 00:19:21.790 align:middle line:90% There's a negative sign. 00:19:21.790 --> 00:19:22.648 align:middle line:90% This is positive. 00:19:22.648 --> 00:19:24.440 align:middle line:84% So you can think of the whole thing in blue 00:19:24.440 --> 00:19:27.410 align:middle line:90% as a positive number normally. 00:19:27.410 --> 00:19:31.070 align:middle line:84% And so whether you price above or below 00:19:31.070 --> 00:19:33.470 align:middle line:84% the inverse elasticity is typically 00:19:33.470 --> 00:19:35.810 align:middle line:84% going to depend on whether this is a positive number 00:19:35.810 --> 00:19:36.815 align:middle line:90% or a negative number. 00:19:36.815 --> 00:19:40.010 align:middle line:90% 00:19:40.010 --> 00:19:41.970 align:middle line:84% Standard case, we're dealing with a monopoly. 00:19:41.970 --> 00:19:43.890 align:middle line:84% Pricing is often the substitutes case. 00:19:43.890 --> 00:19:46.670 align:middle line:84% The substitutes case is when I raise 00:19:46.670 --> 00:19:49.460 align:middle line:84% the price of the blue iPhone, sales of the silver iPhone 00:19:49.460 --> 00:19:50.690 align:middle line:90% go up. 00:19:50.690 --> 00:19:54.470 align:middle line:84% These goods are substitutes for consumers. 00:19:54.470 --> 00:20:01.130 align:middle line:84% So when that's the case, margins for both products are positive. 00:20:01.130 --> 00:20:05.360 align:middle line:84% And your margin exceeds the single product Lerner index 00:20:05.360 --> 00:20:06.230 align:middle line:90% formula. 00:20:06.230 --> 00:20:10.100 align:middle line:84% And the amount by which it exceeds that single index Lerner 00:20:10.100 --> 00:20:16.910 align:middle line:84% formula depends on what is dx minus i over pdpi. 00:20:16.910 --> 00:20:18.350 align:middle line:84% How large is the diversion ratio? 00:20:18.350 --> 00:20:20.730 align:middle line:84% This divided by this is the diversion ratio. 00:20:20.730 --> 00:20:23.060 align:middle line:84% So the larger is the diversion ratio, 00:20:23.060 --> 00:20:25.010 align:middle line:84% the larger is the markup you're earning 00:20:25.010 --> 00:20:27.650 align:middle line:84% on the other second product you're selling, 00:20:27.650 --> 00:20:30.025 align:middle line:84% the more you exceed the Lerner index formula. 00:20:30.025 --> 00:20:32.900 align:middle line:90% 00:20:32.900 --> 00:20:38.370 align:middle line:84% The other more interesting case is this complements case. 00:20:38.370 --> 00:20:43.010 align:middle line:84% So sometimes you have firms selling complements like 00:20:43.010 --> 00:20:46.730 align:middle line:84% hamburgers and French fries at McDonald's. 00:20:46.730 --> 00:20:48.890 align:middle line:84% If your price of hamburgers is lower, more 00:20:48.890 --> 00:20:51.720 align:middle line:84% people come to McDonald's, more people buy French fries. 00:20:51.720 --> 00:20:53.660 align:middle line:84% If your price for hamburgers is higher, 00:20:53.660 --> 00:20:55.490 align:middle line:84% there could be an income effect where people feel like I've 00:20:55.490 --> 00:20:56.430 align:middle line:90% spent too much already. 00:20:56.430 --> 00:20:57.750 align:middle line:84% I don't want to also buy the fries. 00:20:57.750 --> 00:20:58.790 align:middle line:84% But typically it's like the effect 00:20:58.790 --> 00:21:00.915 align:middle line:84% is going to be people don't come to your restaurant 00:21:00.915 --> 00:21:03.920 align:middle line:84% and then they don't buy fries at all. 00:21:03.920 --> 00:21:07.190 align:middle line:84% In the Apple case, if I don't sell them this iPhone, 00:21:07.190 --> 00:21:10.040 align:middle line:84% I'm not going to sell them my earbuds 00:21:10.040 --> 00:21:11.463 align:middle line:90% that I sell for a huge profit. 00:21:11.463 --> 00:21:13.880 align:middle line:84% If I don't sell them my iPhone, I'm not going to sell them 00:21:13.880 --> 00:21:17.360 align:middle line:84% the 17th charger that they've had to buy from me because I 00:21:17.360 --> 00:21:20.870 align:middle line:84% keep changing the way my charger and connector works. 00:21:20.870 --> 00:21:23.490 align:middle line:84% And so the complements case is the higher 00:21:23.490 --> 00:21:27.230 align:middle line:84% is the price of my iPhone, I sell fewer. 00:21:27.230 --> 00:21:31.180 align:middle line:84% If I raise the price of the iPhone, I sell fewer earbuds. 00:21:31.180 --> 00:21:39.790 align:middle line:84% And so in this case, now this becomes a negative number. 00:21:39.790 --> 00:21:44.940 align:middle line:84% You set price below the elasticity of demand. 00:21:44.940 --> 00:21:48.090 align:middle line:84% Very easily if this diversion ratio is large, 00:21:48.090 --> 00:21:49.620 align:middle line:84% you can often think of the diversion 00:21:49.620 --> 00:21:52.860 align:middle line:90% ratio is in some sense 1. 00:21:52.860 --> 00:21:57.160 align:middle line:84% For every iPhone I sell, I don't sell, 00:21:57.160 --> 00:21:59.460 align:middle line:84% that's exactly one pair of earbuds I don't sell 00:21:59.460 --> 00:22:01.650 align:middle line:84% or that's one charger I don't sell. 00:22:01.650 --> 00:22:06.870 align:middle line:84% So if this is a large number, if this diversion ratio is 1, 00:22:06.870 --> 00:22:09.870 align:middle line:84% the margin is high in the competing good, 00:22:09.870 --> 00:22:13.020 align:middle line:84% you can very easily end up with pricing one good below cost 00:22:13.020 --> 00:22:14.680 align:middle line:90% or even having a negative price. 00:22:14.680 --> 00:22:17.340 align:middle line:84% You may have thought you go into the store 00:22:17.340 --> 00:22:19.410 align:middle line:84% to buy an Inkjet printer and it's remarkable, 00:22:19.410 --> 00:22:24.420 align:middle line:84% but I can get this phenomenally complicated piece of electronics 00:22:24.420 --> 00:22:26.760 align:middle line:84% with all kinds of chips in it and hardware and precision 00:22:26.760 --> 00:22:28.783 align:middle line:90% stuff, and it's only $99. 00:22:28.783 --> 00:22:31.200 align:middle line:84% And then you get really pissed off when the thing runs out 00:22:31.200 --> 00:22:33.823 align:middle line:84% of ink seven days later and it's $59 for ink. 00:22:33.823 --> 00:22:35.490 align:middle line:84% And you're thinking yourself, how can it 00:22:35.490 --> 00:22:39.030 align:middle line:84% be that it only costs $99 to produce this printer 00:22:39.030 --> 00:22:44.520 align:middle line:84% and yet it costs $59 to get a bottle of ink this big? 00:22:44.520 --> 00:22:46.570 align:middle line:84% You would think the ink should only cost $0.50. 00:22:46.570 --> 00:22:48.400 align:middle line:84% And the answer is yes, obviously, yes. 00:22:48.400 --> 00:22:50.410 align:middle line:90% The ink only costs $0.50. 00:22:50.410 --> 00:22:52.890 align:middle line:84% But they're pricing the printer way 00:22:52.890 --> 00:22:54.730 align:middle line:84% below the cost of producing the printer 00:22:54.730 --> 00:22:58.750 align:middle line:90% to get you to buy the ink. 00:22:58.750 --> 00:23:01.913 align:middle line:84% I guess newspapers, these examples no longer resonate. 00:23:01.913 --> 00:23:04.580 align:middle line:84% There was a time when people had these things called newspapers. 00:23:04.580 --> 00:23:06.330 align:middle line:84% And instead of reading news on your phone, 00:23:06.330 --> 00:23:08.320 align:middle line:84% you got this big piece of paper delivered 00:23:08.320 --> 00:23:10.720 align:middle line:90% to your house every day. 00:23:10.720 --> 00:23:13.390 align:middle line:84% The way newspapers used to work is the newspaper would only cost 00:23:13.390 --> 00:23:14.660 align:middle line:90% $0.50. 00:23:14.660 --> 00:23:17.333 align:middle line:84% And if you think about how much does it cost to produce and get 00:23:17.333 --> 00:23:19.000 align:middle line:84% a newspaper to your house every morning, 00:23:19.000 --> 00:23:21.010 align:middle line:84% the answer was way more than $0.50. 00:23:21.010 --> 00:23:23.900 align:middle line:84% It would cost $1 or $1.50 or something like that. 00:23:23.900 --> 00:23:25.280 align:middle line:90% How did newspapers work? 00:23:25.280 --> 00:23:28.325 align:middle line:84% Newspapers worked because they sold them for $0.50, 00:23:28.325 --> 00:23:30.700 align:middle line:84% even though it cost $1.50 to produce it because they were 00:23:30.700 --> 00:23:31.960 align:middle line:90% full of ads. 00:23:31.960 --> 00:23:37.990 align:middle line:84% And they were full of classified ads that now go to wherever. 00:23:37.990 --> 00:23:40.460 align:middle line:84% They were full of real estate ads that are now on Zillow. 00:23:40.460 --> 00:23:43.430 align:middle line:84% They were full of ads for local car dealers. 00:23:43.430 --> 00:23:47.050 align:middle line:84% They were full of ads for all kinds of things. 00:23:47.050 --> 00:23:49.270 align:middle line:84% Whatever, eBay, everything that was sold on eBay 00:23:49.270 --> 00:23:52.180 align:middle line:90% was sold through the newspaper. 00:23:52.180 --> 00:23:54.520 align:middle line:84% The newspapers were selling the newspaper below cost 00:23:54.520 --> 00:23:58.970 align:middle line:84% because the cost of printing an ad was essentially 0. 00:23:58.970 --> 00:24:00.920 align:middle line:84% And so the cost of producing an ad was 0. 00:24:00.920 --> 00:24:04.150 align:middle line:84% They'd charge you $100 or $1,000 to list your house 00:24:04.150 --> 00:24:05.620 align:middle line:90% in the newspaper. 00:24:05.620 --> 00:24:09.400 align:middle line:84% Price minus cost was huge on the ads that made the margin 00:24:09.400 --> 00:24:12.010 align:middle line:90% on the newspaper negative. 00:24:12.010 --> 00:24:18.410 align:middle line:84% In some industries, it even becomes there's a 0 lower bound 00:24:18.410 --> 00:24:20.860 align:middle line:90% that becomes important. 00:24:20.860 --> 00:24:24.190 align:middle line:84% Google in some sense would like to pay you to use Google. 00:24:24.190 --> 00:24:25.360 align:middle line:90% Every time you do a-- 00:24:25.360 --> 00:24:29.350 align:middle line:84% Google's ad revenue is something like $0.06 or $0.08 a search. 00:24:29.350 --> 00:24:34.150 align:middle line:84% Every time you do a search on Google, Google is making money. 00:24:34.150 --> 00:24:37.870 align:middle line:84% The marginal cost of running a search for you on Google 00:24:37.870 --> 00:24:39.380 align:middle line:90% is very, very small. 00:24:39.380 --> 00:24:41.500 align:middle line:84% So when Google does the price search, 00:24:41.500 --> 00:24:44.900 align:middle line:84% it costs them a thousandth of a cent to run the search for you. 00:24:44.900 --> 00:24:47.690 align:middle line:84% They're making $0.08 profit on every search. 00:24:47.690 --> 00:24:50.410 align:middle line:84% They want to set the price of having Google on your phone 00:24:50.410 --> 00:24:50.922 align:middle line:90% negative. 00:24:50.922 --> 00:24:52.630 align:middle line:84% That's why Google is willing to pay Apple 00:24:52.630 --> 00:24:55.172 align:middle line:84% billions and billions of dollars to put Google there and make 00:24:55.172 --> 00:24:56.760 align:middle line:84% Google a default. In some sense, they 00:24:56.760 --> 00:24:59.100 align:middle line:84% want to have a negative price to get as many people 00:24:59.100 --> 00:25:00.050 align:middle line:90% to use it as possible. 00:25:00.050 --> 00:25:05.070 align:middle line:90% 00:25:05.070 --> 00:25:09.960 align:middle line:84% Any questions on multiproduct pricing? 00:25:09.960 --> 00:25:13.290 align:middle line:90% So product quality. 00:25:13.290 --> 00:25:16.470 align:middle line:84% So as I said, people often get frustrated 00:25:16.470 --> 00:25:18.570 align:middle line:84% with their Microsoft and other products 00:25:18.570 --> 00:25:21.925 align:middle line:84% and think this thing is so bad because they're a monopoly. 00:25:21.925 --> 00:25:22.633 align:middle line:90% I have no choice. 00:25:22.633 --> 00:25:27.610 align:middle line:90% 00:25:27.610 --> 00:25:30.190 align:middle line:84% Maybe they get frustrated with the Sloan cafeteria 00:25:30.190 --> 00:25:34.430 align:middle line:84% and are like, this place is closer than everywhere else 00:25:34.430 --> 00:25:36.640 align:middle line:84% and yet still people walk across to Brothers Market 00:25:36.640 --> 00:25:42.325 align:middle line:84% to get their lunch anyway, because it's not necessarily 00:25:42.325 --> 00:25:44.200 align:middle line:84% the price of the food at the Sloan cafeteria, 00:25:44.200 --> 00:25:47.530 align:middle line:84% although it's also high, but does Sloan cafeteria 00:25:47.530 --> 00:25:51.700 align:middle line:84% have the option to set the optimal product quality? 00:25:51.700 --> 00:25:55.190 align:middle line:84% Basic reasoning is going to be, in some sense, 00:25:55.190 --> 00:25:57.190 align:middle line:84% while we may think the monopolists often provide 00:25:57.190 --> 00:26:00.430 align:middle line:84% low quality products, at least the profit maximizing 00:26:00.430 --> 00:26:02.410 align:middle line:84% theory tells us, they ought to do pretty well 00:26:02.410 --> 00:26:03.710 align:middle line:90% on this dimension. 00:26:03.710 --> 00:26:06.280 align:middle line:84% So let's suppose you have a monopolist. 00:26:06.280 --> 00:26:08.770 align:middle line:84% It has a product that's a variable quality s. 00:26:08.770 --> 00:26:12.450 align:middle line:84% And s is a choice variable as well as the price. 00:26:12.450 --> 00:26:16.230 align:middle line:84% So suppose there's a constant marginal cost c of s. 00:26:16.230 --> 00:26:19.260 align:middle line:84% So you could think of this as either the quality of the food 00:26:19.260 --> 00:26:21.780 align:middle line:84% you're providing at a lunch or s could 00:26:21.780 --> 00:26:25.800 align:middle line:84% be the quality of the camera in an iPhone. 00:26:25.800 --> 00:26:28.715 align:middle line:84% The unit cost of producing this lunch or the unit 00:26:28.715 --> 00:26:30.840 align:middle line:84% cost of producing the iPhone is going to be c of s. 00:26:30.840 --> 00:26:33.090 align:middle line:84% And the higher quality you make it, the more expensive 00:26:33.090 --> 00:26:34.740 align:middle line:90% it's going to produce. 00:26:34.740 --> 00:26:37.300 align:middle line:84% And then I will do this as a constant marginal cost. 00:26:37.300 --> 00:26:38.790 align:middle line:84% So if you produce 100 iPhones, it's 00:26:38.790 --> 00:26:41.350 align:middle line:84% just 100 times the cost of producing one. 00:26:41.350 --> 00:26:46.020 align:middle line:84% If you produce 100 lunches, 100 times the cost of one lunch. 00:26:46.020 --> 00:26:48.450 align:middle line:84% Something we're going to do all the time in IO 00:26:48.450 --> 00:26:52.000 align:middle line:84% is take consumer heterogeneity very seriously. 00:26:52.000 --> 00:26:54.450 align:middle line:84% So I'm going to think about massive consumers who 00:26:54.450 --> 00:26:57.940 align:middle line:84% have types theta distributed uniform on 0, 1. 00:26:57.940 --> 00:27:00.790 align:middle line:84% The uniform 0, 1 here is without loss of generality. 00:27:00.790 --> 00:27:01.630 align:middle line:90% I do that a lot. 00:27:01.630 --> 00:27:05.070 align:middle line:84% It's just I index consumers by their percentile 00:27:05.070 --> 00:27:07.740 align:middle line:84% within the distribution of consumers. 00:27:07.740 --> 00:27:10.470 align:middle line:84% Here it's indexing within their distribution of consumers 00:27:10.470 --> 00:27:13.510 align:middle line:84% in terms of how high their valuation is for the good 00:27:13.510 --> 00:27:14.980 align:middle line:90% that people are buying. 00:27:14.980 --> 00:27:17.535 align:middle line:90% So consumers are going to get-- 00:27:17.535 --> 00:27:19.660 align:middle line:84% consumers are going to either buy one-- they either 00:27:19.660 --> 00:27:21.230 align:middle line:90% want one lunch or one phone. 00:27:21.230 --> 00:27:22.720 align:middle line:90% It's a unit demand case. 00:27:22.720 --> 00:27:26.020 align:middle line:84% And they're going to get utility v of s theta minus p 00:27:26.020 --> 00:27:28.360 align:middle line:84% if they buy one unit of a quality s good 00:27:28.360 --> 00:27:30.820 align:middle line:90% and their type is theta. 00:27:30.820 --> 00:27:32.590 align:middle line:84% And I'm going to assume that we've 00:27:32.590 --> 00:27:35.650 align:middle line:84% ordered the qualities and the types 00:27:35.650 --> 00:27:40.300 align:middle line:84% so that utility is increasing in quality for everyone, 00:27:40.300 --> 00:27:41.590 align:middle line:90% regardless of theta. 00:27:41.590 --> 00:27:44.980 align:middle line:84% And the high theta people have a higher willingness 00:27:44.980 --> 00:27:48.730 align:middle line:84% to pay than the low theta people, again, for any quality. 00:27:48.730 --> 00:27:50.530 align:middle line:84% That doesn't fit all products, but assume 00:27:50.530 --> 00:27:52.660 align:middle line:84% that there are these high types and low types 00:27:52.660 --> 00:27:54.820 align:middle line:84% and the high types are willing to pay more for everything, as 00:27:54.820 --> 00:27:56.487 align:middle line:84% if they're wealthier people with a lower 00:27:56.487 --> 00:27:59.830 align:middle line:90% marginal utility of income. 00:27:59.830 --> 00:28:02.290 align:middle line:84% Now imagine that in equilibrium the firm 00:28:02.290 --> 00:28:04.090 align:middle line:90% is going to sell q units. 00:28:04.090 --> 00:28:06.100 align:middle line:84% Given that the high theta people have 00:28:06.100 --> 00:28:10.780 align:middle line:84% the highest willingness to pay, if you're going to sell-- 00:28:10.780 --> 00:28:11.490 align:middle line:90% let me put my-- 00:28:11.490 --> 00:28:16.200 align:middle line:90% 00:28:16.200 --> 00:28:19.470 align:middle line:90% theta is uniform 0, 1 here. 00:28:19.470 --> 00:28:26.077 align:middle line:84% If I'm going to sell q units, what's 00:28:26.077 --> 00:28:27.660 align:middle line:84% going to have to happen in equilibrium 00:28:27.660 --> 00:28:30.810 align:middle line:84% is everyone whose type is above 1 minus q buys the good, 00:28:30.810 --> 00:28:33.790 align:middle line:84% and anyone whose type is below 1 minus q doesn't buy the good. 00:28:33.790 --> 00:28:36.948 align:middle line:84% Because if someone here is willing to buy the good, 00:28:36.948 --> 00:28:38.490 align:middle line:84% anyone with a higher type is strictly 00:28:38.490 --> 00:28:40.440 align:middle line:84% willing to buy the good at the same price. 00:28:40.440 --> 00:28:41.940 align:middle line:84% So if you're going to sell the good, 00:28:41.940 --> 00:28:46.320 align:middle line:84% this is how it's going to look in characteristic space. 00:28:46.320 --> 00:28:49.590 align:middle line:84% So therefore, what's your maximization problem? 00:28:49.590 --> 00:28:50.550 align:middle line:90% I'm going to maximize. 00:28:50.550 --> 00:28:51.967 align:middle line:84% I'm going to set it over, maximize 00:28:51.967 --> 00:28:55.560 align:middle line:84% over the quantity that I'm going to sell and the quality. 00:28:55.560 --> 00:29:00.600 align:middle line:84% If I'm going to sell to everyone in this interval, 00:29:00.600 --> 00:29:02.490 align:middle line:84% it's got to be that this person is 00:29:02.490 --> 00:29:05.350 align:middle line:84% just indifferent between buying and not buying. 00:29:05.350 --> 00:29:09.900 align:middle line:84% This person is willing to pay v of s 1 minus q, 00:29:09.900 --> 00:29:12.610 align:middle line:90% because 1 minus q is their type. 00:29:12.610 --> 00:29:15.310 align:middle line:84% So if I'm going to sell to these people and not to these people, 00:29:15.310 --> 00:29:18.010 align:middle line:84% it must be that this is the price I'm charging. 00:29:18.010 --> 00:29:21.010 align:middle line:84% So if I'm going to sell q units, my profit 00:29:21.010 --> 00:29:24.625 align:middle line:84% is q times v of s1 minus q minus c of s. 00:29:24.625 --> 00:29:27.305 align:middle line:90% 00:29:27.305 --> 00:29:32.400 align:middle line:84% I now want to differentiate that with respect to s to say, 00:29:32.400 --> 00:29:36.120 align:middle line:84% what is the optimal quality to choose? 00:29:36.120 --> 00:29:39.720 align:middle line:90% Notice the q. 00:29:39.720 --> 00:29:43.320 align:middle line:84% This q drops out of the first order condition. 00:29:43.320 --> 00:29:48.960 align:middle line:84% So I just get dv / ds evaluated at the monopoly level of s 00:29:48.960 --> 00:29:52.920 align:middle line:90% and 1 minus q equals dc / ds. 00:29:52.920 --> 00:29:57.750 align:middle line:84% So what this says is that the effect on the unit 00:29:57.750 --> 00:30:02.580 align:middle line:84% cost of increasing quality is exactly what 00:30:02.580 --> 00:30:04.920 align:middle line:84% the effect of that quality increase 00:30:04.920 --> 00:30:10.560 align:middle line:84% is on the evaluation of the marginal consumer. 00:30:10.560 --> 00:30:13.230 align:middle line:84% So the monopolist is choosing optimally-- 00:30:13.230 --> 00:30:16.170 align:middle line:84% choosing quality optimally from the perspective 00:30:16.170 --> 00:30:17.470 align:middle line:90% of the marginal consumer. 00:30:17.470 --> 00:30:20.010 align:middle line:84% And that's because the marginal consumer, if you raise 00:30:20.010 --> 00:30:22.218 align:middle line:84% the quality, that person will be willing to pay more, 00:30:22.218 --> 00:30:23.290 align:middle line:90% you can charge them more. 00:30:23.290 --> 00:30:25.440 align:middle line:84% And so you're improving quality if and only 00:30:25.440 --> 00:30:29.803 align:middle line:84% if the amount that the marginal consumers are 00:30:29.803 --> 00:30:31.720 align:middle line:84% willing to pay for that improvement in quality 00:30:31.720 --> 00:30:34.262 align:middle line:84% matches the cost of implementing that improvement in quality. 00:30:34.262 --> 00:30:36.870 align:middle line:90% 00:30:36.870 --> 00:30:41.650 align:middle line:84% So that in some sense, it's a partial optimality theorem. 00:30:41.650 --> 00:30:44.680 align:middle line:90% What would a social planner do? 00:30:44.680 --> 00:30:47.230 align:middle line:84% Well, a social planner would be choosing-- 00:30:47.230 --> 00:30:50.170 align:middle line:84% so let me say, so this would be what the monopolist ended up 00:30:50.170 --> 00:30:51.088 align:middle line:90% doing. 00:30:51.088 --> 00:30:52.880 align:middle line:84% What would the social planner end up doing? 00:30:52.880 --> 00:30:55.060 align:middle line:90% It's maximizing over q and s. 00:30:55.060 --> 00:30:58.360 align:middle line:84% Integral over all types from 1 minus q to 1, 00:30:58.360 --> 00:31:02.680 align:middle line:84% v of s theta minus c of sd theta. 00:31:02.680 --> 00:31:03.820 align:middle line:90% Maximize that. 00:31:03.820 --> 00:31:06.890 align:middle line:84% I differentiate with respect to s under the integral sign. 00:31:06.890 --> 00:31:09.550 align:middle line:84% So I get the integral from 1 minus q to 1 00:31:09.550 --> 00:31:14.740 align:middle line:84% of dv / ds minus dc / ds d theta. 00:31:14.740 --> 00:31:20.470 align:middle line:84% The integral from 1 minus q to 1 of dc / ds 00:31:20.470 --> 00:31:23.210 align:middle line:84% is just that's a constant function. 00:31:23.210 --> 00:31:25.700 align:middle line:90% So it's just q times dc / ds. 00:31:25.700 --> 00:31:29.590 align:middle line:84% So I've divided both sides by q and I get this expression. 00:31:29.590 --> 00:31:35.390 align:middle line:84% The effect on the marginal cost of increasing quality 00:31:35.390 --> 00:31:38.590 align:middle line:90% is equal to-- 00:31:38.590 --> 00:31:42.520 align:middle line:84% this is just an average across consumers of dv / ds. 00:31:42.520 --> 00:31:46.090 align:middle line:84% So what the monopolist is doing is instead of equating 00:31:46.090 --> 00:31:49.450 align:middle line:84% the cost of the quality improvement 00:31:49.450 --> 00:31:51.400 align:middle line:84% to the benefit for the marginal consumer, 00:31:51.400 --> 00:31:54.450 align:middle line:84% it's equating to the benefit for the average consumer. 00:31:54.450 --> 00:31:58.050 align:middle line:84% So there are two reasons in which the monopolist is not 00:31:58.050 --> 00:32:01.770 align:middle line:90% doing what we want socially. 00:32:01.770 --> 00:32:04.560 align:middle line:84% First is that monopolist is focusing 00:32:04.560 --> 00:32:07.950 align:middle line:84% on the marginal consumer rather than the average consumer. 00:32:07.950 --> 00:32:09.780 align:middle line:84% But the other thing the monopolist is doing 00:32:09.780 --> 00:32:13.950 align:middle line:84% is typically in this case, the monopolist 00:32:13.950 --> 00:32:17.828 align:middle line:90% is choosing this quantity. 00:32:17.828 --> 00:32:20.370 align:middle line:84% The competitive quantity that the social planner might choose 00:32:20.370 --> 00:32:21.310 align:middle line:90% might be down here. 00:32:21.310 --> 00:32:22.735 align:middle line:84% The social planner is going to serve-- 00:32:22.735 --> 00:32:24.090 align:middle line:84% for the social planner, there would 00:32:24.090 --> 00:32:26.215 align:middle line:84% be many more iPhones sold than there are with Apple 00:32:26.215 --> 00:32:29.610 align:middle line:90% as an iPhone monopolist. 00:32:29.610 --> 00:32:36.210 align:middle line:84% And so notice that these effects often go in opposite directions. 00:32:36.210 --> 00:32:38.430 align:middle line:84% One is that the monopolist is focusing 00:32:38.430 --> 00:32:41.250 align:middle line:84% on the marginal consumer rather than the average consumer. 00:32:41.250 --> 00:32:43.720 align:middle line:84% Because you're focusing on the marginal consumer, 00:32:43.720 --> 00:32:47.400 align:middle line:84% you may make the quality too low. 00:32:47.400 --> 00:32:51.060 align:middle line:84% It may be that the average consumer wishes 00:32:51.060 --> 00:32:52.980 align:middle line:84% that their iPhone had a camera that 00:32:52.980 --> 00:32:55.240 align:middle line:84% was as good as the camera on the Google Pixel, 00:32:55.240 --> 00:32:57.535 align:middle line:84% but it doesn't have a camera that's 00:32:57.535 --> 00:33:00.505 align:middle line:84% that good because the marginal consumer is not 00:33:00.505 --> 00:33:02.630 align:middle line:84% willing to pay as much for the camera improvements. 00:33:02.630 --> 00:33:05.920 align:middle line:84% So you make the camera lower than the average iPhone consumer 00:33:05.920 --> 00:33:08.170 align:middle line:84% wishes-- worse than the iPhone camera-- 00:33:08.170 --> 00:33:10.710 align:middle line:84% average iPhone consumer wants the camera to be. 00:33:10.710 --> 00:33:14.980 align:middle line:84% But the second effect is because only wealthy people and graduate 00:33:14.980 --> 00:33:20.230 align:middle line:84% students have iPhones, because relatively wealthy people have 00:33:20.230 --> 00:33:23.140 align:middle line:84% iPhones, all these people are not actually 00:33:23.140 --> 00:33:24.410 align:middle line:90% getting iPhones at all. 00:33:24.410 --> 00:33:26.680 align:middle line:84% And so it could be that the average-- 00:33:26.680 --> 00:33:29.530 align:middle line:84% with a social planner, you'd have this many iPhones sold. 00:33:29.530 --> 00:33:32.170 align:middle line:84% The average consumer would be somewhere down here. 00:33:32.170 --> 00:33:34.040 align:middle line:84% And serving the average consumer, 00:33:34.040 --> 00:33:37.480 align:middle line:84% the iPhone would be even worse than it is today. 00:33:37.480 --> 00:33:40.750 align:middle line:84% It's because Apple is only serving the high of the theta 00:33:40.750 --> 00:33:41.930 align:middle line:90% distribution. 00:33:41.930 --> 00:33:44.230 align:middle line:84% Their marginal consumer is higher 00:33:44.230 --> 00:33:47.020 align:middle line:84% than the marginal consumer would be if they were serving everyone 00:33:47.020 --> 00:33:48.615 align:middle line:90% with a $300 iPhone. 00:33:48.615 --> 00:33:53.010 align:middle line:90% 00:33:53.010 --> 00:33:55.710 align:middle line:84% But one corollary of this analysis 00:33:55.710 --> 00:33:59.940 align:middle line:84% is if you have a homogeneous population model, which people 00:33:59.940 --> 00:34:02.580 align:middle line:84% often do not thinking about it just to simplify things, 00:34:02.580 --> 00:34:04.650 align:middle line:84% you make all the consumers identical. 00:34:04.650 --> 00:34:08.300 align:middle line:84% If all consumers are identical, two things happen. 00:34:08.300 --> 00:34:11.219 align:middle line:84% One is the marginal consumer is the average consumer, 00:34:11.219 --> 00:34:13.530 align:middle line:84% because marginal and average are the same when 00:34:13.530 --> 00:34:15.449 align:middle line:90% all consumers are identical. 00:34:15.449 --> 00:34:18.330 align:middle line:84% The second is, in a homogeneous population model, 00:34:18.330 --> 00:34:20.590 align:middle line:84% you're always making an all or none sales decision. 00:34:20.590 --> 00:34:22.920 align:middle line:84% If everyone is identical, you either sell to everyone 00:34:22.920 --> 00:34:24.750 align:middle line:90% or you sell to no one. 00:34:24.750 --> 00:34:27.510 align:middle line:84% And typically you sell to everyone 00:34:27.510 --> 00:34:29.260 align:middle line:90% faced with those two choices. 00:34:29.260 --> 00:34:32.130 align:middle line:84% So in a homogeneous population model, 00:34:32.130 --> 00:34:34.270 align:middle line:84% there's a strong optimal quality theorem. 00:34:34.270 --> 00:34:35.760 align:middle line:84% That is, if you analyze something 00:34:35.760 --> 00:34:37.650 align:middle line:84% that's a homogeneous population model, 00:34:37.650 --> 00:34:40.360 align:middle line:84% you assume the monopolist is maximizing its profits. 00:34:40.360 --> 00:34:41.860 align:middle line:84% You're going to assume it's choosing 00:34:41.860 --> 00:34:43.800 align:middle line:84% the socially optimal quality, because there's 00:34:43.800 --> 00:34:45.820 align:middle line:84% no scope for those two to be different. 00:34:45.820 --> 00:34:50.960 align:middle line:90% 00:34:50.960 --> 00:34:53.129 align:middle line:90% Questions on that? 00:34:53.129 --> 00:34:53.629 align:middle line:90% OK. 00:34:53.629 --> 00:34:56.060 align:middle line:84% And this optimal quality reasoning, this 00:34:56.060 --> 00:34:59.380 align:middle line:84% will come in a few different times later in the class. 00:34:59.380 --> 00:35:03.730 align:middle line:90% 00:35:03.730 --> 00:35:06.400 align:middle line:90% Durable goods. 00:35:06.400 --> 00:35:10.570 align:middle line:84% So in many cases, people are selling 00:35:10.570 --> 00:35:14.350 align:middle line:84% durable goods that provide services over many periods. 00:35:14.350 --> 00:35:16.090 align:middle line:90% You're selling washing machines. 00:35:16.090 --> 00:35:17.320 align:middle line:90% You're selling cars. 00:35:17.320 --> 00:35:19.270 align:middle line:90% You're selling phones. 00:35:19.270 --> 00:35:21.250 align:middle line:84% Consumers are buying it today, and they're 00:35:21.250 --> 00:35:23.610 align:middle line:84% using that same product in multiple periods. 00:35:23.610 --> 00:35:27.390 align:middle line:90% 00:35:27.390 --> 00:35:29.520 align:middle line:84% One intuition you'll have sometimes 00:35:29.520 --> 00:35:32.200 align:middle line:84% is that this is a potential advantage. 00:35:32.200 --> 00:35:34.443 align:middle line:84% Firms can use this durable goods nature, 00:35:34.443 --> 00:35:36.360 align:middle line:84% the fact that people buy it at different times 00:35:36.360 --> 00:35:39.810 align:middle line:84% to move down the demand curve and sell the goods first 00:35:39.810 --> 00:35:41.760 align:middle line:84% to the people who have high values. 00:35:41.760 --> 00:35:45.260 align:middle line:84% You have whatever new Xbox is coming out. 00:35:45.260 --> 00:35:47.760 align:middle line:84% There's some number of people who line up outside the store, 00:35:47.760 --> 00:35:49.810 align:middle line:84% sleeping on the curb to buy the Xbox. 00:35:49.810 --> 00:35:52.320 align:middle line:84% They do not get it at the best possible price. 00:35:52.320 --> 00:35:54.010 align:middle line:90% They all pay full price. 00:35:54.010 --> 00:35:58.170 align:middle line:84% And then 14 months later at Christmas, that same Xbox 00:35:58.170 --> 00:35:59.620 align:middle line:90% is selling for much, much less. 00:35:59.620 --> 00:36:01.920 align:middle line:84% And maybe what Microsoft has done 00:36:01.920 --> 00:36:04.260 align:middle line:84% is they've walked down that demand curve, 00:36:04.260 --> 00:36:06.090 align:middle line:84% first selling it to the high value people, 00:36:06.090 --> 00:36:07.500 align:middle line:84% then waiting for the high value people to buy, 00:36:07.500 --> 00:36:09.375 align:middle line:84% then selling to the next highest value people 00:36:09.375 --> 00:36:12.240 align:middle line:84% and in some sense trying to discriminate 00:36:12.240 --> 00:36:14.640 align:middle line:84% between people based on when they want to buy the good. 00:36:14.640 --> 00:36:19.350 align:middle line:90% 00:36:19.350 --> 00:36:23.850 align:middle line:84% While somewhat appealing when you first hear the idea, 00:36:23.850 --> 00:36:27.450 align:middle line:84% if you think about rational, forward looking consumers, 00:36:27.450 --> 00:36:30.870 align:middle line:84% a durable goods monopolist is competing with itself. 00:36:30.870 --> 00:36:33.940 align:middle line:84% And you're competing with your future self. 00:36:33.940 --> 00:36:37.230 align:middle line:84% And if consumers think your future self is 00:36:37.230 --> 00:36:40.518 align:middle line:84% going to lower the prices of the Xbox, 00:36:40.518 --> 00:36:42.810 align:middle line:84% they're going to be hesitant to buy it on the first day 00:36:42.810 --> 00:36:43.950 align:middle line:84% that it comes out, because they're 00:36:43.950 --> 00:36:45.867 align:middle line:84% going to think, why don't I just wait a month 00:36:45.867 --> 00:36:47.200 align:middle line:90% and get it when the price drops? 00:36:47.200 --> 00:36:49.867 align:middle line:84% Or why don't I wait three months and get it when the price drops 00:36:49.867 --> 00:36:50.400 align:middle line:90% or whatever? 00:36:50.400 --> 00:36:51.997 align:middle line:84% And then the consumers who might have 00:36:51.997 --> 00:36:54.330 align:middle line:84% bought it six months out are going to think, why don't I 00:36:54.330 --> 00:36:57.145 align:middle line:84% wait till Christmas or whatever, where it's 00:36:57.145 --> 00:36:58.270 align:middle line:90% the day after Thanksgiving. 00:36:58.270 --> 00:37:00.450 align:middle line:84% I can get that thing at half price 7 months, 00:37:00.450 --> 00:37:02.375 align:middle line:84% 14 months after it comes out or whatever. 00:37:02.375 --> 00:37:05.470 align:middle line:90% 00:37:05.470 --> 00:37:08.740 align:middle line:84% So first thing I want to do is just think about 00:37:08.740 --> 00:37:12.610 align:middle line:84% let's suppose you do have this model with commitment power 00:37:12.610 --> 00:37:14.420 align:middle line:84% and with the forward looking consumers. 00:37:14.420 --> 00:37:16.420 align:middle line:90% What would you do? 00:37:16.420 --> 00:37:20.320 align:middle line:84% So simplest model of commitment power. 00:37:20.320 --> 00:37:22.850 align:middle line:90% I'm often going to do this-- 00:37:22.850 --> 00:37:25.660 align:middle line:84% I know in some subjects, especially macro these days, 00:37:25.660 --> 00:37:28.570 align:middle line:84% everything has to be an infinite horizon dynamic programming 00:37:28.570 --> 00:37:30.370 align:middle line:90% model. 00:37:30.370 --> 00:37:32.740 align:middle line:84% There's certainly plenty of infinite horizon dynamic 00:37:32.740 --> 00:37:34.120 align:middle line:90% programming models in IO. 00:37:34.120 --> 00:37:36.608 align:middle line:84% Often, though, in class, I'm just 00:37:36.608 --> 00:37:38.650 align:middle line:84% going to use two period versions, because I think 00:37:38.650 --> 00:37:41.358 align:middle line:84% the two period versions give you a lot of the intuition with much 00:37:41.358 --> 00:37:43.820 align:middle line:84% less complication than the Bellman equations would. 00:37:43.820 --> 00:37:45.730 align:middle line:84% So my simple model of commitment power. 00:37:45.730 --> 00:37:46.900 align:middle line:90% There are just two periods. 00:37:46.900 --> 00:37:49.740 align:middle line:84% Consumers can buy the good at either t equals 1 or t equals 2. 00:37:49.740 --> 00:37:50.690 align:middle line:90% So think about it. 00:37:50.690 --> 00:37:52.400 align:middle line:84% There's t equals 1, there's t equals 2, 00:37:52.400 --> 00:37:55.870 align:middle line:84% then the world ends and everything goes away. 00:37:55.870 --> 00:37:59.080 align:middle line:90% Good is perfectly durable. 00:37:59.080 --> 00:38:02.860 align:middle line:84% Consumers, again, are going to have types theta distributed 00:38:02.860 --> 00:38:04.010 align:middle line:90% on 0, 1. 00:38:04.010 --> 00:38:06.467 align:middle line:84% Here I'm going to make theta the actual value. 00:38:06.467 --> 00:38:09.050 align:middle line:84% So then I'm going to want theta to have a distribution capital 00:38:09.050 --> 00:38:10.940 align:middle line:90% f of theta. 00:38:10.940 --> 00:38:14.780 align:middle line:84% And so the consumer's utility is going to be 2 theta minus p 00:38:14.780 --> 00:38:17.360 align:middle line:84% if they buy it at t equals 1, theta 00:38:17.360 --> 00:38:19.190 align:middle line:84% minus p, if they buy it at t equals 2, 00:38:19.190 --> 00:38:20.790 align:middle line:90% and 0 if they don't buy. 00:38:20.790 --> 00:38:23.030 align:middle line:84% So the reasoning is like a type theta 00:38:23.030 --> 00:38:27.140 align:middle line:84% consumer gets utility theta every period they have the good. 00:38:27.140 --> 00:38:28.740 align:middle line:84% And so they buy it at t equals 1. 00:38:28.740 --> 00:38:31.065 align:middle line:84% They get to enjoy the Xbox in two periods. 00:38:31.065 --> 00:38:32.690 align:middle line:84% If they buy it at t equals 2, they only 00:38:32.690 --> 00:38:34.015 align:middle line:90% get to enjoy it for one period. 00:38:34.015 --> 00:38:36.698 align:middle line:90% 00:38:36.698 --> 00:38:38.990 align:middle line:84% I'm going to have a constant marginal cost of producing 00:38:38.990 --> 00:38:40.850 align:middle line:84% the good, and I'm going to assume 00:38:40.850 --> 00:38:42.710 align:middle line:84% the monopolist has commitment power 00:38:42.710 --> 00:38:44.530 align:middle line:84% and can set the price as p1 and p2 00:38:44.530 --> 00:38:46.405 align:middle line:84% that it's going to charge in the two periods. 00:38:46.405 --> 00:38:49.550 align:middle line:90% 00:38:49.550 --> 00:38:51.590 align:middle line:84% Basic observation in this model is 00:38:51.590 --> 00:38:54.980 align:middle line:84% that in this model, as I've written it, 00:38:54.980 --> 00:38:57.120 align:middle line:84% you don't want to walk down the demand curve. 00:38:57.120 --> 00:39:00.380 align:middle line:84% In this model, what you want to do is sell 00:39:00.380 --> 00:39:01.970 align:middle line:84% the good at the monopoly price at t 00:39:01.970 --> 00:39:05.500 align:middle line:84% equals 1 and commit to make no sales at all at t equals 2. 00:39:05.500 --> 00:39:09.470 align:middle line:90% 00:39:09.470 --> 00:39:12.020 align:middle line:84% So in this model, and obviously there 00:39:12.020 --> 00:39:13.370 align:middle line:90% are some firms that fit this. 00:39:13.370 --> 00:39:15.560 align:middle line:84% If you look at Apple's iPhone pricing, 00:39:15.560 --> 00:39:18.710 align:middle line:84% it seems like that iPhone comes out and it's $999 00:39:18.710 --> 00:39:21.260 align:middle line:84% and everyone knows that the iPhone 14 is coming out 00:39:21.260 --> 00:39:24.290 align:middle line:84% literally next week and the iPhone 13 price still 00:39:24.290 --> 00:39:25.820 align:middle line:90% has not dropped. 00:39:25.820 --> 00:39:28.610 align:middle line:84% They've set it at the monopoly price for iPhone 13s 00:39:28.610 --> 00:39:32.240 align:middle line:84% and they're going to keep it there and never sell another one 00:39:32.240 --> 00:39:33.550 align:middle line:90% until the 14 comes out. 00:39:33.550 --> 00:39:37.560 align:middle line:90% 00:39:37.560 --> 00:39:39.040 align:middle line:90% Why does this happen? 00:39:39.040 --> 00:39:42.970 align:middle line:84% Well, I'm going to think of my theta line. 00:39:42.970 --> 00:39:47.147 align:middle line:84% So suppose we've got our consumers. 00:39:47.147 --> 00:39:48.230 align:middle line:90% They're arranged by theta. 00:39:48.230 --> 00:39:53.860 align:middle line:90% 00:39:53.860 --> 00:39:56.350 align:middle line:84% What's going to happen in equilibrium? 00:39:56.350 --> 00:39:59.290 align:middle line:84% If I have some policy that involves 00:39:59.290 --> 00:40:02.980 align:middle line:84% selling the goods in both periods, 00:40:02.980 --> 00:40:05.950 align:middle line:84% I'm going to have some high types who buy at t 00:40:05.950 --> 00:40:14.140 align:middle line:84% equals 1, some intermediate types who buy at t equals 2, 00:40:14.140 --> 00:40:17.055 align:middle line:84% and some lower types who don't buy. 00:40:17.055 --> 00:40:19.710 align:middle line:90% 00:40:19.710 --> 00:40:21.510 align:middle line:84% Why is that what's going to happen? 00:40:21.510 --> 00:40:23.760 align:middle line:84% Again, it's a revealed preference argument. 00:40:23.760 --> 00:40:26.640 align:middle line:84% If this person is willing to buy the phone at t equals 2, 00:40:26.640 --> 00:40:28.390 align:middle line:90% this person gets a higher value. 00:40:28.390 --> 00:40:30.280 align:middle line:84% They strictly prefer buying it at t equals 2. 00:40:30.280 --> 00:40:34.090 align:middle line:84% So if anyone is buying it, everyone higher is buying it. 00:40:34.090 --> 00:40:36.660 align:middle line:84% So the people buying it have to be above everyone 00:40:36.660 --> 00:40:38.370 align:middle line:90% who never buys it. 00:40:38.370 --> 00:40:42.060 align:middle line:90% And then buying at t equals 1. 00:40:42.060 --> 00:40:44.190 align:middle line:84% If you buy at t equals 1, you may pay a high price 00:40:44.190 --> 00:40:46.140 align:middle line:84% for getting the Xbox on the first day 00:40:46.140 --> 00:40:49.180 align:middle line:84% or you can wait and get it later at a lower price. 00:40:49.180 --> 00:40:52.230 align:middle line:84% Again, if you prefer buying it at t equals 1, 00:40:52.230 --> 00:40:54.630 align:middle line:84% then anyone who's strictly higher value 00:40:54.630 --> 00:40:57.087 align:middle line:84% strictly prefers getting a t equals 1 relative to t 00:40:57.087 --> 00:40:59.670 align:middle line:84% equals 2 because they're getting the extra period of enjoyment 00:40:59.670 --> 00:41:00.520 align:middle line:90% of the good. 00:41:00.520 --> 00:41:04.991 align:middle line:84% That extra period of enjoyment is increasing in theta. 00:41:04.991 --> 00:41:08.240 align:middle line:90% So if I do want-- 00:41:08.240 --> 00:41:11.540 align:middle line:84% often in IO, it often makes sense to think about the choices 00:41:11.540 --> 00:41:14.900 align:middle line:84% not in price space, but in this data space. 00:41:14.900 --> 00:41:18.510 align:middle line:84% So if I want to implement this policy, how do I do it? 00:41:18.510 --> 00:41:22.340 align:middle line:84% Well, for this to be the cut off between people who buy at t 00:41:22.340 --> 00:41:25.310 align:middle line:84% equals 2 and don't buy, obviously you're 00:41:25.310 --> 00:41:29.535 align:middle line:84% going to have to have p2 equals theta 2 hat. 00:41:29.535 --> 00:41:31.160 align:middle line:84% In the second period, you're selling it 00:41:31.160 --> 00:41:32.750 align:middle line:90% at a price of theta 2 hat. 00:41:32.750 --> 00:41:35.900 align:middle line:84% So the people who have utility values above theta 2 hat 00:41:35.900 --> 00:41:37.010 align:middle line:90% buy at the second period. 00:41:37.010 --> 00:41:41.250 align:middle line:84% The people who don't-- utilities below that don't buy. 00:41:41.250 --> 00:41:43.140 align:middle line:90% What about in the first period? 00:41:43.140 --> 00:41:45.780 align:middle line:84% In the first period, you can't sell 00:41:45.780 --> 00:41:48.090 align:middle line:84% to this guy at a price of 2 theta 00:41:48.090 --> 00:41:51.270 align:middle line:84% 1 hat, because he'd be getting 0 utility buying 00:41:51.270 --> 00:41:53.080 align:middle line:90% at a price of 2 theta 1 hat. 00:41:53.080 --> 00:41:54.900 align:middle line:84% He would just wait until the second period 00:41:54.900 --> 00:41:57.180 align:middle line:84% and buy it at this low price and get positive utility 00:41:57.180 --> 00:41:59.250 align:middle line:90% by buying in the second period. 00:41:59.250 --> 00:42:01.440 align:middle line:84% So what you can do, the best you can do 00:42:01.440 --> 00:42:05.730 align:middle line:84% is to chart, set a price so that the utility he gets from buying 00:42:05.730 --> 00:42:08.195 align:middle line:84% in the first period exactly matches 00:42:08.195 --> 00:42:10.320 align:middle line:84% the utility he could get from waiting and buying it 00:42:10.320 --> 00:42:11.680 align:middle line:90% in the second period. 00:42:11.680 --> 00:42:13.650 align:middle line:84% And so what it tells you is that p1 00:42:13.650 --> 00:42:17.460 align:middle line:84% is going to be theta 1 hat bigger than p2. 00:42:17.460 --> 00:42:21.660 align:middle line:84% So p1 is going to be theta 1 hat plus theta 2 00:42:21.660 --> 00:42:23.400 align:middle line:90% hat, which would make sense. 00:42:23.400 --> 00:42:25.020 align:middle line:84% He's getting the one extra utility 00:42:25.020 --> 00:42:28.980 align:middle line:90% of a period of enjoyment. 00:42:28.980 --> 00:42:32.730 align:middle line:84% So you'd think about what's his incremental benefit from buying 00:42:32.730 --> 00:42:33.970 align:middle line:90% it early versus late? 00:42:33.970 --> 00:42:36.070 align:middle line:84% It's theta 1 hat is his incremental utility. 00:42:36.070 --> 00:42:39.535 align:middle line:90% That's what you're charging him. 00:42:39.535 --> 00:42:43.720 align:middle line:84% So monopolist problem looks like this. 00:42:43.720 --> 00:42:47.350 align:middle line:84% Maximize over theta 1 constrained by theta 1 hat 00:42:47.350 --> 00:42:49.180 align:middle line:90% at least theta 2 hat. 00:42:49.180 --> 00:42:54.940 align:middle line:84% The price minus cost at t equals 1 times the number 00:42:54.940 --> 00:42:56.770 align:middle line:84% of units you sell at t equals 1, which 00:42:56.770 --> 00:42:59.320 align:middle line:84% is 1 minus capital f of theta 1 hat, 00:42:59.320 --> 00:43:04.450 align:middle line:84% plus profits you get from second period sales 00:43:04.450 --> 00:43:08.608 align:middle line:84% times the mass of people who buy the good in the second period, 00:43:08.608 --> 00:43:10.150 align:middle line:84% which, again, is the number of people 00:43:10.150 --> 00:43:13.450 align:middle line:84% with types between theta 2 hat and theta 1 hat. 00:43:13.450 --> 00:43:16.480 align:middle line:84% So that's what my profit function looks like. 00:43:16.480 --> 00:43:19.360 align:middle line:84% Notice that I've got two different terms that 00:43:19.360 --> 00:43:21.220 align:middle line:90% involve theta 2 hat minus c. 00:43:21.220 --> 00:43:23.520 align:middle line:90% It's here and it's here. 00:43:23.520 --> 00:43:27.700 align:middle line:84% So if I group those together, it becomes my-- my profit 00:43:27.700 --> 00:43:31.030 align:middle line:84% function looks like theta 1 hat times 1 minus f of theta 1 00:43:31.030 --> 00:43:38.410 align:middle line:84% hat plus theta 2 hat minus c times this plus this. 00:43:38.410 --> 00:43:41.480 align:middle line:90% And then these things cancel. 00:43:41.480 --> 00:43:44.660 align:middle line:84% And so my profit function is theta 1 hat times 00:43:44.660 --> 00:43:49.520 align:middle line:84% 1 minus f of theta 1 hat plus theta 2 hat minus c times 1 00:43:49.520 --> 00:43:52.510 align:middle line:90% minus f of theta 2 hat. 00:43:52.510 --> 00:43:53.775 align:middle line:90% That's my profit function. 00:43:53.775 --> 00:44:00.660 align:middle line:90% 00:44:00.660 --> 00:44:04.890 align:middle line:84% Let's assume that the constraint was not binding. 00:44:04.890 --> 00:44:07.110 align:middle line:84% So if the constraint is not binding, 00:44:07.110 --> 00:44:09.022 align:middle line:84% I actually want to walk down my demand curve 00:44:09.022 --> 00:44:10.980 align:middle line:84% and set a high price and then set a lower price 00:44:10.980 --> 00:44:14.010 align:middle line:90% in the second period. 00:44:14.010 --> 00:44:17.820 align:middle line:84% This is just to separate maximization problems. 00:44:17.820 --> 00:44:21.730 align:middle line:84% I maximize this with respect to theta 1 hat. 00:44:21.730 --> 00:44:24.485 align:middle line:84% I maximize this with respect to theta 2 hat. 00:44:24.485 --> 00:44:28.440 align:middle line:90% 00:44:28.440 --> 00:44:31.380 align:middle line:84% Well, now let me go back to [14.]121 for those of you who 00:44:31.380 --> 00:44:32.580 align:middle line:90% have taken it. 00:44:32.580 --> 00:44:35.340 align:middle line:84% [14.]121 last year, [? Parad ?] did a lot of this functions with 00:44:35.340 --> 00:44:38.100 align:middle line:90% increasing differences stuff. 00:44:38.100 --> 00:44:41.160 align:middle line:84% If you look at this function theta minus c, 00:44:41.160 --> 00:44:44.500 align:middle line:84% 1 minus f of theta, this is like the monopoly profit function. 00:44:44.500 --> 00:44:49.080 align:middle line:84% This thing has increasing differences in c and theta, 00:44:49.080 --> 00:44:52.420 align:middle line:84% which means that this is a very general theorem. 00:44:52.420 --> 00:44:54.990 align:middle line:84% When a monopolist marginal cost is higher, 00:44:54.990 --> 00:44:56.135 align:middle line:90% it sets a higher price. 00:44:56.135 --> 00:44:58.660 align:middle line:90% 00:44:58.660 --> 00:45:01.540 align:middle line:84% This is basically the monopoly pricing problem 00:45:01.540 --> 00:45:03.340 align:middle line:90% with a cost of 0. 00:45:03.340 --> 00:45:07.220 align:middle line:84% This is a monopoly pricing problem with a cost of c. 00:45:07.220 --> 00:45:09.920 align:middle line:84% So the maximizer of this is larger 00:45:09.920 --> 00:45:12.640 align:middle line:90% than the maximizer of this. 00:45:12.640 --> 00:45:14.860 align:middle line:84% So if I ignore the constraint, this 00:45:14.860 --> 00:45:17.080 align:middle line:84% is the monopoly price when your cost is c. 00:45:17.080 --> 00:45:19.360 align:middle line:84% This is the monopoly price when your cost is 0. 00:45:19.360 --> 00:45:21.490 align:middle line:84% This tells you that theta 2 hat would be bigger 00:45:21.490 --> 00:45:23.020 align:middle line:90% than theta 1 hat. 00:45:23.020 --> 00:45:25.810 align:middle line:84% So if I assume this constraint is non-binding, 00:45:25.810 --> 00:45:29.740 align:middle line:84% I get the contradiction that if that constraint is not binding, 00:45:29.740 --> 00:45:32.950 align:middle line:84% I find that theta 2 hat is strictly greater than theta 00:45:32.950 --> 00:45:34.930 align:middle line:84% 1 hat, and that contradicts the assumption 00:45:34.930 --> 00:45:37.570 align:middle line:84% that the constraint is not binding. 00:45:37.570 --> 00:45:38.330 align:middle line:90% Why is this? 00:45:38.330 --> 00:45:42.460 align:middle line:84% In some sense, what's going on here is-- 00:45:42.460 --> 00:45:44.590 align:middle line:84% almost imagine with this durable good, 00:45:44.590 --> 00:45:47.110 align:middle line:84% instead of selling those Xboxes in the first period, 00:45:47.110 --> 00:45:48.045 align:middle line:90% I rent them to people. 00:45:48.045 --> 00:45:49.670 align:middle line:84% I get all those people waiting in line. 00:45:49.670 --> 00:45:52.030 align:middle line:84% I give them a one period rental on the Xbox, 00:45:52.030 --> 00:45:53.845 align:middle line:84% and then I take them all back and then 00:45:53.845 --> 00:45:55.470 align:middle line:84% I rent them again in the second period. 00:45:55.470 --> 00:45:59.134 align:middle line:90% 00:45:59.134 --> 00:46:01.830 align:middle line:84% Anyone that I'm selling it to in the first-- 00:46:01.830 --> 00:46:03.930 align:middle line:84% all these people who get it in the first period, 00:46:03.930 --> 00:46:05.850 align:middle line:84% they're going to rent it back from me 00:46:05.850 --> 00:46:08.530 align:middle line:84% in the second period at a price of theta 2 hat. 00:46:08.530 --> 00:46:11.280 align:middle line:90% 00:46:11.280 --> 00:46:14.340 align:middle line:84% When I'm selling extra units in the first period, 00:46:14.340 --> 00:46:16.770 align:middle line:84% any extra units I sell in the first period 00:46:16.770 --> 00:46:19.770 align:middle line:84% were units that I was going to sell in the second period 00:46:19.770 --> 00:46:20.890 align:middle line:90% anyway. 00:46:20.890 --> 00:46:24.995 align:middle line:84% So the cost of selling them one period early is 0. 00:46:24.995 --> 00:46:27.920 align:middle line:84% I'm already producing them to sell in the second period. 00:46:27.920 --> 00:46:31.530 align:middle line:84% So given that I'm producing them to sell in the second period, 00:46:31.530 --> 00:46:34.410 align:middle line:84% there's 0 cost to me of selling them in the first period 00:46:34.410 --> 00:46:36.150 align:middle line:90% instead. 00:46:36.150 --> 00:46:39.060 align:middle line:84% So it's as if those units I'm going to sell anyway 00:46:39.060 --> 00:46:42.660 align:middle line:84% have a marginal cost of 0 instead of a marginal cost of c. 00:46:42.660 --> 00:46:45.420 align:middle line:84% And so because I'm going to produce them and sell them 00:46:45.420 --> 00:46:47.230 align:middle line:90% anyway, they have no cost. 00:46:47.230 --> 00:46:50.610 align:middle line:84% So that makes me even more want to sell them rather than less 00:46:50.610 --> 00:46:51.370 align:middle line:90% want to sell them. 00:46:51.370 --> 00:46:54.920 align:middle line:90% So this doesn't make any sense. 00:46:54.920 --> 00:46:57.860 align:middle line:84% So this would say is that monopolist selling 00:46:57.860 --> 00:47:02.213 align:middle line:84% durable goods, this whole idea, it's a strictly bad idea 00:47:02.213 --> 00:47:04.130 align:middle line:84% to walk down your demand curve, because you're 00:47:04.130 --> 00:47:07.110 align:middle line:84% taking these goods that you're going to produce anyway. 00:47:07.110 --> 00:47:09.440 align:middle line:84% They really have no cost to selling to people early, 00:47:09.440 --> 00:47:12.510 align:middle line:84% and you're losing those people's enjoyment. 00:47:12.510 --> 00:47:14.390 align:middle line:90% It doesn't make any sense. 00:47:14.390 --> 00:47:16.388 align:middle line:84% These ones, you have to be hesitant to produce 00:47:16.388 --> 00:47:17.430 align:middle line:90% because they have a cost. 00:47:17.430 --> 00:47:20.480 align:middle line:84% These ones have no cost of production. 00:47:20.480 --> 00:47:25.940 align:middle line:84% So in some sense, someone at Apple saw this lecture, 00:47:25.940 --> 00:47:27.990 align:middle line:84% they realized this is what you do. 00:47:27.990 --> 00:47:30.050 align:middle line:84% You do not bring out the new iPhone 00:47:30.050 --> 00:47:31.790 align:middle line:84% and then make its price go down and let 00:47:31.790 --> 00:47:33.660 align:middle line:90% people wait for the price cut. 00:47:33.660 --> 00:47:34.860 align:middle line:90% You just bring it out. 00:47:34.860 --> 00:47:36.420 align:middle line:90% You fix the price forever. 00:47:36.420 --> 00:47:38.360 align:middle line:84% And that's the optimal monopoly pricing 00:47:38.360 --> 00:47:39.520 align:middle line:90% policy with durable goods. 00:47:39.520 --> 00:47:49.758 align:middle line:90% 00:47:49.758 --> 00:47:51.050 align:middle line:90% Simple numerical example there. 00:47:51.050 --> 00:47:52.175 align:middle line:90% I don't need to go over it. 00:47:52.175 --> 00:47:56.580 align:middle line:84% So where would price drops come from in real life? 00:47:56.580 --> 00:47:59.088 align:middle line:90% 00:47:59.088 --> 00:48:00.880 align:middle line:84% Obviously we need to do something different 00:48:00.880 --> 00:48:05.170 align:middle line:84% from this model to get prices of new goods to come out. 00:48:05.170 --> 00:48:06.610 align:middle line:84% I think we do often have the view 00:48:06.610 --> 00:48:08.207 align:middle line:90% that there are many new things. 00:48:08.207 --> 00:48:09.290 align:middle line:90% The new product comes out. 00:48:09.290 --> 00:48:10.240 align:middle line:90% It's got a really high price. 00:48:10.240 --> 00:48:13.070 align:middle line:84% And then if you wait for it, the price will decline over time. 00:48:13.070 --> 00:48:14.790 align:middle line:90% Why does that happen? 00:48:14.790 --> 00:48:17.410 align:middle line:90% 00:48:17.410 --> 00:48:20.800 align:middle line:84% One thing you can get is different preferences. 00:48:20.800 --> 00:48:22.660 align:middle line:84% Suppose the people get utility not 00:48:22.660 --> 00:48:24.400 align:middle line:84% from enjoying the good, for instance, 00:48:24.400 --> 00:48:28.120 align:middle line:84% but from having the Xbox before any of their friends do. 00:48:28.120 --> 00:48:30.227 align:middle line:84% I'm the one guy who has the brand new Xbox. 00:48:30.227 --> 00:48:32.060 align:middle line:84% Everyone has to come to my house to play it. 00:48:32.060 --> 00:48:34.870 align:middle line:84% I somehow get status in my friend group 00:48:34.870 --> 00:48:38.860 align:middle line:84% out of being the guy with the new Xbox. 00:48:38.860 --> 00:48:41.290 align:middle line:90% Fashion clothing. 00:48:41.290 --> 00:48:42.070 align:middle line:90% Clothes come out. 00:48:42.070 --> 00:48:43.810 align:middle line:90% Fashion designers introduce it. 00:48:43.810 --> 00:48:45.890 align:middle line:84% Someone has a copy some number of months later. 00:48:45.890 --> 00:48:51.200 align:middle line:84% But the people who wear the $3,000 dresses get utility from 00:48:51.200 --> 00:48:53.960 align:middle line:84% knowing that they have the genuine $3,000 dress that no one 00:48:53.960 --> 00:48:57.180 align:middle line:84% else has because it's just the new good that comes out. 00:48:57.180 --> 00:49:00.410 align:middle line:84% So if you give people preferences for owning goods 00:49:00.410 --> 00:49:03.050 align:middle line:84% that other people don't have, then it 00:49:03.050 --> 00:49:05.630 align:middle line:84% can make total sense to have some set of people-- 00:49:05.630 --> 00:49:08.360 align:middle line:84% in some sense, almost that's going back 00:49:08.360 --> 00:49:10.250 align:middle line:90% to an optimal quality theorem. 00:49:10.250 --> 00:49:11.720 align:middle line:84% If there are people out there who have a high willingness 00:49:11.720 --> 00:49:13.760 align:middle line:84% to pay for things that other people don't have, 00:49:13.760 --> 00:49:15.380 align:middle line:84% you create a high quality product, 00:49:15.380 --> 00:49:17.880 align:middle line:84% which is a product that other people don't have, and sell it 00:49:17.880 --> 00:49:18.740 align:middle line:90% to them. 00:49:18.740 --> 00:49:20.450 align:middle line:84% And then maybe once you've sold it 00:49:20.450 --> 00:49:22.520 align:middle line:84% to people who have those preferences, 00:49:22.520 --> 00:49:24.980 align:middle line:84% then you can sell it to the other people later. 00:49:24.980 --> 00:49:29.900 align:middle line:84% So that would be one way you can get early adopters. 00:49:29.900 --> 00:49:32.780 align:middle line:84% I think another thing that often comes up in the tech sector 00:49:32.780 --> 00:49:38.120 align:middle line:84% is just costs that are changing over time or costs that 00:49:38.120 --> 00:49:40.940 align:middle line:84% are changing over time because of technological improvements, 00:49:40.940 --> 00:49:43.280 align:middle line:84% because of learning by doing, because of capacity 00:49:43.280 --> 00:49:45.080 align:middle line:90% constraints at the factories. 00:49:45.080 --> 00:49:48.420 align:middle line:84% Early on, it's very costly to produce goods. 00:49:48.420 --> 00:49:49.230 align:middle line:90% The costs are high. 00:49:49.230 --> 00:49:50.930 align:middle line:90% Therefore, you set a high price. 00:49:50.930 --> 00:49:54.598 align:middle line:84% As the technology level changes, it 00:49:54.598 --> 00:49:56.640 align:middle line:84% becomes cheaper and cheaper to produce that good. 00:49:56.640 --> 00:49:59.060 align:middle line:90% You get better at producing it. 00:49:59.060 --> 00:50:00.530 align:middle line:90% Your costs go down. 00:50:00.530 --> 00:50:05.710 align:middle line:90% You drop your price as a result. 00:50:05.710 --> 00:50:12.100 align:middle line:84% Third explanation is firms may have a commitment problem. 00:50:12.100 --> 00:50:18.250 align:middle line:84% That is, if you think about how is this pricing supposed to go. 00:50:18.250 --> 00:50:21.110 align:middle line:90% I bring out my new phone. 00:50:21.110 --> 00:50:24.478 align:middle line:90% I price it at $1,000. 00:50:24.478 --> 00:50:26.770 align:middle line:84% Everyone who wants that new phone comes out and buys it 00:50:26.770 --> 00:50:27.910 align:middle line:90% at $1,000. 00:50:27.910 --> 00:50:30.100 align:middle line:84% And then I'm supposed to sell, in this example, 00:50:30.100 --> 00:50:34.660 align:middle line:84% without phones that break, I sell none for the next two years 00:50:34.660 --> 00:50:36.800 align:middle line:84% or whatever until my new product comes out. 00:50:36.800 --> 00:50:38.630 align:middle line:84% So I sell a whole bunch of phones. 00:50:38.630 --> 00:50:40.160 align:middle line:90% I keep the price high. 00:50:40.160 --> 00:50:41.540 align:middle line:84% Everyone bought it at t equals 1. 00:50:41.540 --> 00:50:43.323 align:middle line:90% No one buys it at t equals 2. 00:50:43.323 --> 00:50:45.490 align:middle line:84% If you imagine the firm sitting there at t equals 2, 00:50:45.490 --> 00:50:47.740 align:middle line:84% they must be thinking once everyone has already 00:50:47.740 --> 00:50:52.060 align:middle line:84% bought the product, if this was not like any repeated game, 00:50:52.060 --> 00:50:55.400 align:middle line:84% imagine this was the last iPhone that was ever going to exist. 00:50:55.400 --> 00:50:57.080 align:middle line:90% Why keep selling 0? 00:50:57.080 --> 00:51:00.895 align:middle line:84% Why not take all those people who have low values? 00:51:00.895 --> 00:51:02.770 align:middle line:84% The other people are going to be disappointed 00:51:02.770 --> 00:51:04.540 align:middle line:84% when you cut the price and wish they had bought it later, 00:51:04.540 --> 00:51:06.860 align:middle line:84% but it's too late for them to actually buy it later. 00:51:06.860 --> 00:51:09.988 align:middle line:84% Once you've already sold the phones, it is optimal for you 00:51:09.988 --> 00:51:12.280 align:middle line:84% then in the second period to go ahead and cut the price 00:51:12.280 --> 00:51:15.130 align:middle line:90% and sell to more people. 00:51:15.130 --> 00:51:18.760 align:middle line:84% And so if you don't have some way 00:51:18.760 --> 00:51:22.480 align:middle line:84% to commit yourself not to lower the price after all 00:51:22.480 --> 00:51:24.730 align:middle line:84% the early adopters have bought, you're 00:51:24.730 --> 00:51:26.212 align:middle line:84% going to want to lower the price. 00:51:26.212 --> 00:51:28.420 align:middle line:84% And then once you're going to want to lower the price 00:51:28.420 --> 00:51:29.830 align:middle line:84% and equilibrium, you will lower the price 00:51:29.830 --> 00:51:31.390 align:middle line:84% and then you'll be in the position of everyone 00:51:31.390 --> 00:51:33.740 align:middle line:84% in the first period thinking you're going to lower the price 00:51:33.740 --> 00:51:36.323 align:middle line:84% and not buy it because the price is too high given what you're 00:51:36.323 --> 00:51:38.570 align:middle line:90% going to lower it to and so on. 00:51:38.570 --> 00:51:41.840 align:middle line:84% So firms could be lowering prices 00:51:41.840 --> 00:51:49.100 align:middle line:84% because they lack the ability to commit to a pricing structure. 00:51:49.100 --> 00:51:51.710 align:middle line:84% So for instance, if you think about Apple 00:51:51.710 --> 00:51:54.110 align:middle line:84% versus some company that's a new startup that may never 00:51:54.110 --> 00:51:56.420 align:middle line:84% have a second phone, you could imagine 00:51:56.420 --> 00:51:58.940 align:middle line:84% that Apple is able to commit because they 00:51:58.940 --> 00:52:00.380 align:middle line:90% have some reputation. 00:52:00.380 --> 00:52:02.120 align:middle line:84% A new entrant is not able to commit, 00:52:02.120 --> 00:52:04.473 align:middle line:84% because no one knows if they're going to be around. 00:52:04.473 --> 00:52:06.140 align:middle line:84% We don't know that they're a firm that's 00:52:06.140 --> 00:52:07.470 align:middle line:90% not going to cut the price. 00:52:07.470 --> 00:52:09.442 align:middle line:84% And so it may be that that's a disadvantage 00:52:09.442 --> 00:52:11.900 align:middle line:84% that you get if you're not someone who is known to be there 00:52:11.900 --> 00:52:13.160 align:middle line:84% for a long time and known to have 00:52:13.160 --> 00:52:14.620 align:middle line:84% this reputation for high pricing. 00:52:14.620 --> 00:52:18.610 align:middle line:90% 00:52:18.610 --> 00:52:22.030 align:middle line:84% I put on the slides, I don't want to take the time to do it. 00:52:22.030 --> 00:52:26.770 align:middle line:84% In some sense, I think that over time people in IO 00:52:26.770 --> 00:52:29.110 align:middle line:84% believe less and less in commitment problems 00:52:29.110 --> 00:52:31.600 align:middle line:84% and in firms being unable to exploit their monopoly 00:52:31.600 --> 00:52:34.280 align:middle line:84% power because they have these severe commitment problems. 00:52:34.280 --> 00:52:38.890 align:middle line:84% But anyway, I put on the slide a simple numerical example 00:52:38.890 --> 00:52:42.550 align:middle line:84% of imagine theta was uniform 0, 1. 00:52:42.550 --> 00:52:43.940 align:middle line:90% Cost was 0. 00:52:43.940 --> 00:52:46.090 align:middle line:90% d of p is 1 minus p. 00:52:46.090 --> 00:52:48.635 align:middle line:90% What is it that you do? 00:52:48.635 --> 00:52:50.260 align:middle line:84% And without worrying about the algebra, 00:52:50.260 --> 00:52:51.510 align:middle line:90% you can look at it afterwards. 00:52:51.510 --> 00:52:52.518 align:middle line:90% What you do is-- 00:52:52.518 --> 00:52:53.810 align:middle line:90% do I have it on the next slide? 00:52:53.810 --> 00:52:55.270 align:middle line:90% Yeah. 00:52:55.270 --> 00:53:00.400 align:middle line:84% Whereas in this model, if you were a monopolist with d of p 00:53:00.400 --> 00:53:05.340 align:middle line:84% as 1 minus p and two periods, the monopoly price is 1/2 per 00:53:05.340 --> 00:53:05.840 align:middle line:90% period. 00:53:05.840 --> 00:53:07.630 align:middle line:90% So you'd set a price of 1. 00:53:07.630 --> 00:53:10.510 align:middle line:84% And so what you get, instead of setting a price of one and then 00:53:10.510 --> 00:53:13.210 align:middle line:84% sticking to it and selling to half the consumers, 00:53:13.210 --> 00:53:16.340 align:middle line:84% the solution turns out to be a set price of 9/10 in the first 00:53:16.340 --> 00:53:19.950 align:middle line:84% period and then drop it to 3/10 in the first period. 00:53:19.950 --> 00:53:22.220 align:middle line:84% You sell to some people, but less than half 00:53:22.220 --> 00:53:24.620 align:middle line:84% in the first period, even though your price is lower. 00:53:24.620 --> 00:53:28.070 align:middle line:84% And then you sell to some people in the second period. 00:53:28.070 --> 00:53:32.750 align:middle line:84% So two things that come out in this simple algebraic example 00:53:32.750 --> 00:53:36.120 align:middle line:84% is you get a decreasing price sequence. 00:53:36.120 --> 00:53:37.700 align:middle line:90% This is the commitment problem. 00:53:37.700 --> 00:53:39.920 align:middle line:84% The price sequence has to go down 00:53:39.920 --> 00:53:41.873 align:middle line:84% because if your price didn't go down, 00:53:41.873 --> 00:53:43.790 align:middle line:84% no one would buy in the second period and then 00:53:43.790 --> 00:53:46.165 align:middle line:84% the second period, you would be getting 0 profits instead 00:53:46.165 --> 00:53:47.280 align:middle line:90% of positive profits. 00:53:47.280 --> 00:53:50.690 align:middle line:84% So in a problem with commitment, you always get p2 less than 1/2 00:53:50.690 --> 00:53:51.830 align:middle line:90% p1. 00:53:51.830 --> 00:53:56.060 align:middle line:84% And you always get profits are lower without commitment than 00:53:56.060 --> 00:54:00.200 align:middle line:84% with commitment, because if the profits were better with this 00:54:00.200 --> 00:54:03.530 align:middle line:84% 9/10, 3/10 thing, that's what you would have chosen when you 00:54:03.530 --> 00:54:04.970 align:middle line:90% had commitment power. 00:54:04.970 --> 00:54:06.680 align:middle line:84% So commitment power is always the best 00:54:06.680 --> 00:54:08.690 align:middle line:84% you can do when you have commitment power. 00:54:08.690 --> 00:54:12.500 align:middle line:84% So what you do in these sort of problems with a self control 00:54:12.500 --> 00:54:14.570 align:middle line:84% problem is always you earn lower profits 00:54:14.570 --> 00:54:16.195 align:middle line:84% and you do walk down your demand curve. 00:54:16.195 --> 00:54:18.960 align:middle line:90% 00:54:18.960 --> 00:54:24.820 align:middle line:84% So firms lacking this ability to commit is one explanation. 00:54:24.820 --> 00:54:26.838 align:middle line:84% And firms are worse off because they 00:54:26.838 --> 00:54:28.880 align:middle line:84% can't commit to doing what they would like to do. 00:54:28.880 --> 00:54:32.790 align:middle line:90% 00:54:32.790 --> 00:54:33.930 align:middle line:90% OK. 00:54:33.930 --> 00:54:38.790 align:middle line:84% The Coase conjecture is even more extreme commitment problem. 00:54:38.790 --> 00:54:42.300 align:middle line:84% I think in the 1970s or whatever, 00:54:42.300 --> 00:54:46.470 align:middle line:84% this was something the Chicago School of Economics took 00:54:46.470 --> 00:54:49.890 align:middle line:84% seriously that if a monopolist has no market power and has 00:54:49.890 --> 00:54:52.650 align:middle line:84% to price at marginal cost because commitment power is 00:54:52.650 --> 00:54:53.770 align:middle line:90% so severe. 00:54:53.770 --> 00:54:56.460 align:middle line:84% So the Coase conjecture is let's take 00:54:56.460 --> 00:55:00.450 align:middle line:84% the extreme version of this where I set a price at t 00:55:00.450 --> 00:55:01.240 align:middle line:90% equals 0. 00:55:01.240 --> 00:55:04.620 align:middle line:84% I set a price at delta t, which you think of as one day later, 2 00:55:04.620 --> 00:55:07.470 align:middle line:90% delta t, 3 delta t, and so on. 00:55:07.470 --> 00:55:12.000 align:middle line:84% And assume that consumers have utility theta e to the minus rt 00:55:12.000 --> 00:55:14.880 align:middle line:90% if they purchase at time t. 00:55:14.880 --> 00:55:17.790 align:middle line:90% So consumers are discounting. 00:55:17.790 --> 00:55:21.480 align:middle line:84% Firms are setting their prices over and over and over again, 00:55:21.480 --> 00:55:24.150 align:middle line:84% and the consumers have this kind of discounted expected utility 00:55:24.150 --> 00:55:26.070 align:middle line:90% and are forward looking. 00:55:26.070 --> 00:55:29.610 align:middle line:84% The Coase conjecture says that under some conditions, 00:55:29.610 --> 00:55:35.050 align:middle line:84% the monopolist profits go to 0 as delta t goes to 0. 00:55:35.050 --> 00:55:37.930 align:middle line:84% So if you make this severe commitment problem 00:55:37.930 --> 00:55:39.940 align:middle line:84% that every day I'm going to choose a new price, 00:55:39.940 --> 00:55:42.315 align:middle line:84% every day I'm going to choose a new price with no ability 00:55:42.315 --> 00:55:45.040 align:middle line:84% to commit to that price, the theorem 00:55:45.040 --> 00:55:50.530 align:middle line:84% is that actually the more often you can change your price, 00:55:50.530 --> 00:55:53.830 align:middle line:84% the more severe this commitment problem gets. 00:55:53.830 --> 00:55:56.860 align:middle line:84% And in the limit as delta t goes to 0, what you do is basically 00:55:56.860 --> 00:55:57.900 align:middle line:90% you do almost like that. 00:55:57.900 --> 00:56:00.490 align:middle line:90% 00:56:00.490 --> 00:56:02.560 align:middle line:84% You start your price out closer to 1, 00:56:02.560 --> 00:56:06.490 align:middle line:84% but then it just drops to 0 and it goes to 0 almost instantly. 00:56:06.490 --> 00:56:08.950 align:middle line:84% And in the limit as delta t goes to 0, and in fact, 00:56:08.950 --> 00:56:11.740 align:middle line:84% your price goes to 0 instantly and the monopolist 00:56:11.740 --> 00:56:13.110 align:middle line:90% earns 0 profits at all. 00:56:13.110 --> 00:56:19.200 align:middle line:90% 00:56:19.200 --> 00:56:20.610 align:middle line:90% Why does this happen? 00:56:20.610 --> 00:56:23.178 align:middle line:90% 00:56:23.178 --> 00:56:24.720 align:middle line:84% Again, it's the example, the argument 00:56:24.720 --> 00:56:27.670 align:middle line:84% is really there's nothing else that could happen in this model. 00:56:27.670 --> 00:56:33.090 align:middle line:84% So suppose I did this example where 00:56:33.090 --> 00:56:36.180 align:middle line:84% if I have commitment power, I just set the price at 1 00:56:36.180 --> 00:56:37.180 align:middle line:90% and never change it. 00:56:37.180 --> 00:56:40.560 align:middle line:84% So if I put time on this axis and p of t on this axis, 00:56:40.560 --> 00:56:43.030 align:middle line:84% this is what the monopolist wants to do. 00:56:43.030 --> 00:56:45.920 align:middle line:90% And when you have no-- 00:56:45.920 --> 00:56:47.990 align:middle line:84% when you had two periods, I think I showed you 00:56:47.990 --> 00:56:50.250 align:middle line:90% that this is some sense. 00:56:50.250 --> 00:56:51.990 align:middle line:90% That's what the monopolist does. 00:56:51.990 --> 00:56:54.690 align:middle line:84% It sets the price here and then drops the price down. 00:56:54.690 --> 00:56:56.990 align:middle line:84% If you have three periods, the monopolist 00:56:56.990 --> 00:56:58.550 align:middle line:84% is going to do something like this, 00:56:58.550 --> 00:57:03.940 align:middle line:84% set the price here, set the price here, set the price here. 00:57:03.940 --> 00:57:06.370 align:middle line:84% Imagine the prices didn't go to 0. 00:57:06.370 --> 00:57:09.400 align:middle line:84% If prices didn't go to 0, it must be when you do this thing 00:57:09.400 --> 00:57:13.390 align:middle line:84% and very rapidly, you get some smooth curve. 00:57:13.390 --> 00:57:19.000 align:middle line:84% Like this is what the monopolist does when it's changing prices 00:57:19.000 --> 00:57:20.540 align:middle line:90% every second. 00:57:20.540 --> 00:57:25.000 align:middle line:84% It just drops its price smoothly over time. 00:57:25.000 --> 00:57:31.660 align:middle line:84% But dropping your price smoothly over time at a finite slope just 00:57:31.660 --> 00:57:33.610 align:middle line:90% isn't going to ever be optimal. 00:57:33.610 --> 00:57:35.120 align:middle line:90% And why is that? 00:57:35.120 --> 00:57:39.880 align:middle line:84% The reason is that imagine this was what you were going to do 00:57:39.880 --> 00:57:43.750 align:middle line:84% and you decided this time to instead of charging this price, 00:57:43.750 --> 00:57:46.240 align:middle line:84% I'm just going to jump ahead one period 00:57:46.240 --> 00:57:50.780 align:middle line:84% and charge this price instead and then follow the same path. 00:57:50.780 --> 00:57:53.080 align:middle line:84% So I just say, OK, this one period, I'm 00:57:53.080 --> 00:57:55.510 align:middle line:84% just going to jump ahead and charge that price instead 00:57:55.510 --> 00:57:57.230 align:middle line:90% of this price. 00:57:57.230 --> 00:58:00.860 align:middle line:84% If I jump ahead one period, I move all future sales back 00:58:00.860 --> 00:58:03.400 align:middle line:90% by delta t. 00:58:03.400 --> 00:58:12.080 align:middle line:84% The gain for moving those things ahead is first order in delta t. 00:58:12.080 --> 00:58:14.610 align:middle line:84% Because it's got an e to the minus delta t in it. 00:58:14.610 --> 00:58:17.150 align:middle line:84% I just move sales forward one period. 00:58:17.150 --> 00:58:22.720 align:middle line:84% What have I lost by dropping the price from this to this? 00:58:22.720 --> 00:58:26.570 align:middle line:84% Well, I could have earned this price on this tiny, tiny number 00:58:26.570 --> 00:58:28.580 align:middle line:90% of units I sell in that period. 00:58:28.580 --> 00:58:31.380 align:middle line:84% I now get this price on those units. 00:58:31.380 --> 00:58:36.740 align:middle line:84% And so my loss from having jumped forward 00:58:36.740 --> 00:58:39.980 align:middle line:90% is this thing times this thing. 00:58:39.980 --> 00:58:43.490 align:middle line:84% That's second order in delta t, because the number of products 00:58:43.490 --> 00:58:46.670 align:middle line:84% I'm selling in each delta t unit interval is like a delta t, 00:58:46.670 --> 00:58:49.160 align:middle line:84% proportional to delta t, and the price drops 00:58:49.160 --> 00:58:51.200 align:middle line:90% are proportional to delta t. 00:58:51.200 --> 00:58:55.375 align:middle line:84% So the gain for moving things forward would be first order. 00:58:55.375 --> 00:58:57.750 align:middle line:84% The loss for moving things forward would be second order. 00:58:57.750 --> 00:58:59.880 align:middle line:84% So you would always want to move things forward. 00:58:59.880 --> 00:59:01.820 align:middle line:84% And so what must be the case is it's not 00:59:01.820 --> 00:59:04.220 align:middle line:84% the case that these things are both order delta t. 00:59:04.220 --> 00:59:07.310 align:middle line:84% It's got to be that this thing is falling 00:59:07.310 --> 00:59:10.670 align:middle line:84% at a nearly infinite rate so that the loss and gains 00:59:10.670 --> 00:59:13.420 align:middle line:90% both become first order. 00:59:13.420 --> 00:59:15.580 align:middle line:84% So anyway, so the Coase conjecture 00:59:15.580 --> 00:59:19.090 align:middle line:84% says there are some conditions under which this happens. 00:59:19.090 --> 00:59:22.032 align:middle line:84% There's a big literature on the Coase conjecture. 00:59:22.032 --> 00:59:24.490 align:middle line:84% If you read the Fudenberg, Tirole Game Theory textbook, not 00:59:24.490 --> 00:59:27.790 align:middle line:84% the IO textbook, it's got pages and pages and pages 00:59:27.790 --> 00:59:29.440 align:middle line:84% proving versions of Coase theorem. 00:59:29.440 --> 00:59:32.620 align:middle line:90% I think an IO right now-- 00:59:32.620 --> 00:59:34.210 align:middle line:84% I should say it's Coase conjecture 00:59:34.210 --> 00:59:35.530 align:middle line:84% even though it's a theorem, whereas the Coase 00:59:35.530 --> 00:59:36.890 align:middle line:90% theorem is really a conjecture. 00:59:36.890 --> 00:59:39.115 align:middle line:84% But the Coase conjecture is true. 00:59:39.115 --> 00:59:39.740 align:middle line:90% It's a theorem. 00:59:39.740 --> 00:59:42.670 align:middle line:84% People have proved it under various assumptions. 00:59:42.670 --> 00:59:44.570 align:middle line:90% Why isn't it true? 00:59:44.570 --> 00:59:48.195 align:middle line:84% There's more of a literature in IO on why it doesn't happen, 00:59:48.195 --> 00:59:49.570 align:middle line:84% and I don't think anyone believes 00:59:49.570 --> 00:59:50.940 align:middle line:90% it should happen these days. 00:59:50.940 --> 00:59:52.110 align:middle line:90% Why wouldn't it happen. 00:59:52.110 --> 00:59:57.710 align:middle line:84% First of all, firms don't change their price every day. 00:59:57.710 --> 00:59:59.597 align:middle line:84% It takes time to change your price. 00:59:59.597 --> 01:00:01.930 align:middle line:84% In particular, it takes time after you change your price 01:00:01.930 --> 01:00:04.300 align:middle line:84% for all consumers who are supposed to buy at that price 01:00:04.300 --> 01:00:05.580 align:middle line:90% to buy at that price. 01:00:05.580 --> 01:00:08.410 align:middle line:84% So if you decide I'm going to sell my phone for-- 01:00:08.410 --> 01:00:12.100 align:middle line:84% my Xbox at $399 instead of $459, you're 01:00:12.100 --> 01:00:14.690 align:middle line:84% going to have to wait a month for all the $399 consumers 01:00:14.690 --> 01:00:16.410 align:middle line:84% to get around to coming in and buying it. 01:00:16.410 --> 01:00:19.570 align:middle line:84% So it's not like the consumer's buying instantly. 01:00:19.570 --> 01:00:20.990 align:middle line:90% It's also costly. 01:00:20.990 --> 01:00:23.990 align:middle line:84% You're not going to change your price every day. 01:00:23.990 --> 01:00:26.540 align:middle line:84% Firms like Apple can have reputations 01:00:26.540 --> 01:00:29.630 align:middle line:90% for not buying their goods. 01:00:29.630 --> 01:00:32.180 align:middle line:84% You can also have inflows of new high willingness 01:00:32.180 --> 01:00:33.005 align:middle line:90% to pay customers. 01:00:33.005 --> 01:00:34.880 align:middle line:84% So even if Apple didn't have this reputation, 01:00:34.880 --> 01:00:36.920 align:middle line:84% there are people who are very high value iPhone 01:00:36.920 --> 01:00:38.630 align:middle line:90% users whose phones break. 01:00:38.630 --> 01:00:40.130 align:middle line:84% And then when their phones break, 01:00:40.130 --> 01:00:43.670 align:middle line:84% those new high theta people show up wanting to buy a new iPhone. 01:00:43.670 --> 01:00:47.270 align:middle line:84% So in some sense, there's always high theta people showing up 01:00:47.270 --> 01:00:47.930 align:middle line:90% in the market. 01:00:47.930 --> 01:00:50.097 align:middle line:84% And with high theta people showing up in the market, 01:00:50.097 --> 01:00:53.065 align:middle line:84% that gives you another reason to keep your prices high. 01:00:53.065 --> 01:00:59.090 align:middle line:90% 01:00:59.090 --> 01:01:01.800 align:middle line:84% Consumers not being all rational and forward looking. 01:01:01.800 --> 01:01:05.390 align:middle line:84% Again, that argument that I gave you was all the consumers 01:01:05.390 --> 01:01:08.360 align:middle line:84% are anticipating the future price drops. 01:01:08.360 --> 01:01:12.080 align:middle line:84% If you have some consumers who don't realize the price is going 01:01:12.080 --> 01:01:14.120 align:middle line:84% to drop, then you're going to want 01:01:14.120 --> 01:01:15.995 align:middle line:84% to exploit the consumers who don't understand 01:01:15.995 --> 01:01:17.412 align:middle line:84% that the prices are going to drop. 01:01:17.412 --> 01:01:19.220 align:middle line:84% That's going to change that first order 01:01:19.220 --> 01:01:22.070 align:middle line:90% condition I gave you. 01:01:22.070 --> 01:01:24.750 align:middle line:84% There are also many strategic reasons. 01:01:24.750 --> 01:01:28.790 align:middle line:84% So, for instance, you can rent the good 01:01:28.790 --> 01:01:31.040 align:middle line:90% rather than selling the good. 01:01:31.040 --> 01:01:32.973 align:middle line:84% Like in the old days, it used to be 01:01:32.973 --> 01:01:35.390 align:middle line:84% that AT&T, when they were a phone monopolist in the United 01:01:35.390 --> 01:01:37.160 align:middle line:90% States, didn't sell telephones. 01:01:37.160 --> 01:01:38.582 align:middle line:90% They only rented telephones. 01:01:38.582 --> 01:01:40.790 align:middle line:84% And you had to rent your phone from AT&T. They didn't 01:01:40.790 --> 01:01:42.660 align:middle line:90% offer sales as an option. 01:01:42.660 --> 01:01:45.800 align:middle line:84% When Xerox first started selling copy machines, 01:01:45.800 --> 01:01:47.760 align:middle line:90% they didn't sell copy machines. 01:01:47.760 --> 01:01:48.960 align:middle line:90% They rented copy machines. 01:01:48.960 --> 01:01:50.780 align:middle line:84% If you rent them instead of selling them, 01:01:50.780 --> 01:01:52.550 align:middle line:84% you don't face this commitment problem, 01:01:52.550 --> 01:01:54.740 align:middle line:84% because just every month you're renting to your current set 01:01:54.740 --> 01:01:55.400 align:middle line:90% of consumers. 01:01:55.400 --> 01:01:57.690 align:middle line:84% You're setting your optimal price to keep renting it. 01:01:57.690 --> 01:01:59.240 align:middle line:90% That is the monopoly price. 01:01:59.240 --> 01:02:01.830 align:middle line:90% 01:02:01.830 --> 01:02:04.170 align:middle line:84% The other thing you can do is a tricky thing 01:02:04.170 --> 01:02:06.340 align:middle line:84% that seems like it's favorable to your customers. 01:02:06.340 --> 01:02:08.830 align:middle line:84% You can say, I have this very customer friendly policy. 01:02:08.830 --> 01:02:11.970 align:middle line:84% If I ever sell this iPhone to any customer at a lower price 01:02:11.970 --> 01:02:15.270 align:middle line:84% than I sold it to you, I'm going to rebate you the difference 01:02:15.270 --> 01:02:16.060 align:middle line:90% automatically. 01:02:16.060 --> 01:02:17.893 align:middle line:84% It's going to come back to your credit card. 01:02:17.893 --> 01:02:20.677 align:middle line:84% It's a best price guarantee that you never 01:02:20.677 --> 01:02:22.260 align:middle line:84% have to worry about I bought my iPhone 01:02:22.260 --> 01:02:23.635 align:middle line:84% and got ripped and got ripped off 01:02:23.635 --> 01:02:25.260 align:middle line:84% and someone else paid less than me. 01:02:25.260 --> 01:02:28.480 align:middle line:84% Again, I adopt that most favored customer policy. 01:02:28.480 --> 01:02:29.800 align:middle line:90% I commit to it. 01:02:29.800 --> 01:02:32.490 align:middle line:84% Then once I've sold my phone to all the high value customers, 01:02:32.490 --> 01:02:34.620 align:middle line:84% I don't want to cut the price, because if I cut the price, 01:02:34.620 --> 01:02:36.010 align:middle line:84% I'm cutting the price for everyone. 01:02:36.010 --> 01:02:38.177 align:middle line:84% That's back to the same thing as the rental problem. 01:02:38.177 --> 01:02:43.000 align:middle line:90% 01:02:43.000 --> 01:02:45.430 align:middle line:84% And firms can also strategically, 01:02:45.430 --> 01:02:47.560 align:middle line:84% I can just build a factory that only produces 01:02:47.560 --> 01:02:48.740 align:middle line:90% so many phones per month. 01:02:48.740 --> 01:02:51.157 align:middle line:84% I build a factory that only produces enough phones to sell 01:02:51.157 --> 01:02:52.550 align:middle line:90% to everyone at one size. 01:02:52.550 --> 01:02:55.190 align:middle line:84% I'm not tempted to raise my price. 01:02:55.190 --> 01:02:57.160 align:middle line:84% So I think the general view in IO 01:02:57.160 --> 01:03:01.450 align:middle line:84% is that there are enough reasons why the consumer is not 01:03:01.450 --> 01:03:02.270 align:middle line:90% being rational. 01:03:02.270 --> 01:03:04.960 align:middle line:84% The inflows of high willingness to pay customers 01:03:04.960 --> 01:03:08.390 align:middle line:84% are such that the extreme Coase thing shouldn't happen. 01:03:08.390 --> 01:03:11.170 align:middle line:84% And if you were in a situation where the extreme Coase 01:03:11.170 --> 01:03:14.677 align:middle line:84% problem were there, firms would come up with a way around it. 01:03:14.677 --> 01:03:16.510 align:middle line:84% Obviously, there are some antitrust concerns 01:03:16.510 --> 01:03:19.720 align:middle line:84% with some of these, but there are ways 01:03:19.720 --> 01:03:21.080 align:middle line:90% to avoid the Coase conjecture. 01:03:21.080 --> 01:03:26.380 align:middle line:84% I think we don't necessarily think that this is so important. 01:03:26.380 --> 01:03:29.100 align:middle line:84% And I should say that this is something 01:03:29.100 --> 01:03:30.350 align:middle line:90% you'll get later in the class. 01:03:30.350 --> 01:03:32.530 align:middle line:84% We start doing-- a lot of people in IO 01:03:32.530 --> 01:03:36.070 align:middle line:84% are doing structural models of dynamic situations 01:03:36.070 --> 01:03:38.320 align:middle line:84% where you're always assuming firms follow these Markov 01:03:38.320 --> 01:03:39.920 align:middle line:90% perfect strategies. 01:03:39.920 --> 01:03:42.920 align:middle line:84% This Markov maximization is, in some sense, always 01:03:42.920 --> 01:03:45.375 align:middle line:84% a firm with commitment problems where 01:03:45.375 --> 01:03:47.750 align:middle line:84% it's every period the firm is choosing the optimal action 01:03:47.750 --> 01:03:49.520 align:middle line:90% given that state. 01:03:49.520 --> 01:03:52.130 align:middle line:84% Our skepticism of the Coase conjecture, in some sense, 01:03:52.130 --> 01:03:54.470 align:middle line:84% should translate to skepticism of some of those models. 01:03:54.470 --> 01:03:56.858 align:middle line:84% Or at least think about do we really 01:03:56.858 --> 01:03:58.400 align:middle line:84% think the firms lack commitment power 01:03:58.400 --> 01:04:01.010 align:middle line:84% and are always going to be maximizing period by period 01:04:01.010 --> 01:04:02.185 align:middle line:90% what's optimal for them? 01:04:02.185 --> 01:04:04.820 align:middle line:90% 01:04:04.820 --> 01:04:07.400 align:middle line:84% So then the final topic I want to talk about today, 01:04:07.400 --> 01:04:09.890 align:middle line:84% and I'll see if I have time to finish it, 01:04:09.890 --> 01:04:12.080 align:middle line:84% this is a more recent topic in IO. 01:04:12.080 --> 01:04:16.040 align:middle line:84% And if you wanted to do IO theory on monopoly pricing 01:04:16.040 --> 01:04:20.780 align:middle line:84% as a thesis topic, this would be the active area of IO monopoly 01:04:20.780 --> 01:04:22.790 align:middle line:90% pricing you could work on today. 01:04:22.790 --> 01:04:25.550 align:middle line:84% So I'll give an overview of information design. 01:04:25.550 --> 01:04:27.800 align:middle line:84% And obviously, I would say if you really want to study 01:04:27.800 --> 01:04:32.108 align:middle line:84% information design, Steven and Ian in the [14.]281 class do 01:04:32.108 --> 01:04:34.400 align:middle line:84% a lot of information design, and you would want to know 01:04:34.400 --> 01:04:36.692 align:middle line:84% information design at the level at which they teach it. 01:04:36.692 --> 01:04:42.360 align:middle line:84% But I think basics for this I'm going to cover here as well. 01:04:42.360 --> 01:04:45.230 align:middle line:84% So motivating lecture for this is 01:04:45.230 --> 01:04:48.950 align:middle line:84% that it is an interesting observation that when you look 01:04:48.950 --> 01:04:51.943 align:middle line:84% at goods, there are some goods for which you seem 01:04:51.943 --> 01:04:53.360 align:middle line:84% to get an awful lot of information 01:04:53.360 --> 01:04:54.950 align:middle line:84% about what you're buying and some for which 01:04:54.950 --> 01:04:56.690 align:middle line:84% you seem to get very little information about what you're 01:04:56.690 --> 01:04:57.535 align:middle line:90% buying. 01:04:57.535 --> 01:04:58.910 align:middle line:84% So an example of a good where you 01:04:58.910 --> 01:05:00.368 align:middle line:84% seem to get a lot of information is 01:05:00.368 --> 01:05:02.870 align:middle line:84% you want to buy a new laptop computer. 01:05:02.870 --> 01:05:04.620 align:middle line:84% You can get on the site and the firms that 01:05:04.620 --> 01:05:06.918 align:middle line:84% are selling you laptop computers, 01:05:06.918 --> 01:05:08.460 align:middle line:84% they're going to tell you exactly how 01:05:08.460 --> 01:05:09.940 align:middle line:90% many pixels are in the screen. 01:05:09.940 --> 01:05:12.030 align:middle line:84% They're going to tell you what processor 01:05:12.030 --> 01:05:18.440 align:middle line:84% is in it, what the processor speed is, what the battery 01:05:18.440 --> 01:05:22.040 align:middle line:84% life is, what the battery power is, how much the laptop weighs, 01:05:22.040 --> 01:05:23.153 align:middle line:90% what it's made out of. 01:05:23.153 --> 01:05:25.070 align:middle line:84% You can get a tremendous amount of information 01:05:25.070 --> 01:05:27.505 align:middle line:84% about every laptop that you're considering buying. 01:05:27.505 --> 01:05:28.880 align:middle line:84% I don't know how many of you have 01:05:28.880 --> 01:05:31.422 align:middle line:84% moved here recently and tried to buy a mattress in the United 01:05:31.422 --> 01:05:31.930 align:middle line:90% States. 01:05:31.930 --> 01:05:34.250 align:middle line:84% You try to buy a mattress in the United States 01:05:34.250 --> 01:05:37.070 align:middle line:84% and there are lots of these mattress stores 01:05:37.070 --> 01:05:38.660 align:middle line:90% and they sell really expensive. 01:05:38.660 --> 01:05:43.050 align:middle line:84% It's often like this mattress is $1,200 or this one is $300. 01:05:43.050 --> 01:05:46.610 align:middle line:84% And it's just impossible to figure out 01:05:46.610 --> 01:05:48.840 align:middle line:84% what makes mattress A different from mattress B. 01:05:48.840 --> 01:05:51.215 align:middle line:84% And you even think, OK, I'll go to four different stores, 01:05:51.215 --> 01:05:53.210 align:middle line:84% all of whom sell Sealy Posturepedic 01:05:53.210 --> 01:05:55.430 align:middle line:90% mattresses and comparison shop. 01:05:55.430 --> 01:05:58.400 align:middle line:84% And you'll see that this store has the Sealy Posturepedic 01:05:58.400 --> 01:06:00.890 align:middle line:84% Ruby and Diamond and Emerald, and this store 01:06:00.890 --> 01:06:04.550 align:middle line:84% has the Sealy Posturepedic Gold and Platinum and Bronze, 01:06:04.550 --> 01:06:08.090 align:middle line:84% and this one has this Sealy Posturepedic Cloud and Mountain 01:06:08.090 --> 01:06:09.800 align:middle line:90% and Valley or something. 01:06:09.800 --> 01:06:12.492 align:middle line:84% Just every different store, even if it's 01:06:12.492 --> 01:06:15.200 align:middle line:84% the same brand of mattress, they have all different names on them 01:06:15.200 --> 01:06:17.780 align:middle line:84% and no one's giving you any attribute. 01:06:17.780 --> 01:06:22.250 align:middle line:84% The coils in this one are 0.73 mils thick 01:06:22.250 --> 01:06:25.950 align:middle line:84% and the memory foam has this compression per unit thing. 01:06:25.950 --> 01:06:27.812 align:middle line:84% And given that you can't actually 01:06:27.812 --> 01:06:29.270 align:middle line:84% sleep on them in the store, it just 01:06:29.270 --> 01:06:31.550 align:middle line:84% seems like there's no information 01:06:31.550 --> 01:06:33.980 align:middle line:84% whatsoever to compare these matches and figure out 01:06:33.980 --> 01:06:37.250 align:middle line:84% do I want to pay $100 for this one instead of that one. 01:06:37.250 --> 01:06:38.960 align:middle line:84% Wholesale used cars are almost one 01:06:38.960 --> 01:06:42.540 align:middle line:84% of the most interesting examples of explicit lack of information. 01:06:42.540 --> 01:06:46.225 align:middle line:84% So you turn your car into a dealer. 01:06:46.225 --> 01:06:47.600 align:middle line:84% Let's suppose that you had a car. 01:06:47.600 --> 01:06:50.270 align:middle line:90% 01:06:50.270 --> 01:06:51.417 align:middle line:90% I have my Honda Civic. 01:06:51.417 --> 01:06:52.750 align:middle line:90% I decide I want to buy a Toyota. 01:06:52.750 --> 01:06:55.080 align:middle line:84% I turn in my old Honda Civic to the Toyota dealer 01:06:55.080 --> 01:06:56.500 align:middle line:90% when I buy my Toyota. 01:06:56.500 --> 01:06:59.318 align:middle line:84% And then my Toyota dealer doesn't want to have a Honda. 01:06:59.318 --> 01:07:01.860 align:middle line:84% So he takes and sells it at the wholesale level to some Honda 01:07:01.860 --> 01:07:05.820 align:middle line:84% dealer somewhere who wants to have a used Honda on their lot. 01:07:05.820 --> 01:07:08.100 align:middle line:84% You go to see these wholesale used car auctions 01:07:08.100 --> 01:07:09.390 align:middle line:90% and it's a remarkable thing. 01:07:09.390 --> 01:07:13.530 align:middle line:84% They have these giant parking lots in the middle of nowhere 01:07:13.530 --> 01:07:16.150 align:middle line:84% with some big building next to them. 01:07:16.150 --> 01:07:19.020 align:middle line:84% All the used car dealers show up with cars 01:07:19.020 --> 01:07:21.220 align:middle line:84% to sell and have their cars listed for sale. 01:07:21.220 --> 01:07:24.030 align:middle line:84% And then a bunch of people have the cars listed to buy. 01:07:24.030 --> 01:07:29.400 align:middle line:84% And the rules for the wholesale used car auction is they drive 01:07:29.400 --> 01:07:34.740 align:middle line:84% the 2013 Honda Civic, 63,000 miles, up there. 01:07:34.740 --> 01:07:38.400 align:middle line:84% All the buyers are allowed to look at the car from the inside. 01:07:38.400 --> 01:07:43.420 align:middle line:84% A guy who works at the auction house starts the engine. 01:07:43.420 --> 01:07:46.350 align:middle line:90% You can hear that the car runs. 01:07:46.350 --> 01:07:48.100 align:middle line:84% You can look at it from the outside. 01:07:48.100 --> 01:07:50.380 align:middle line:84% You may not open the hood and look under the hood. 01:07:50.380 --> 01:07:52.560 align:middle line:84% You may not look under it with the mirror 01:07:52.560 --> 01:07:55.480 align:middle line:84% and see whether the muffler is rusted out. 01:07:55.480 --> 01:07:59.670 align:middle line:84% You may not have your mechanic try out all the parts. 01:07:59.670 --> 01:08:01.330 align:middle line:90% The car is wheeled up there. 01:08:01.330 --> 01:08:03.790 align:middle line:84% Everything stays closed except for the guy from the dealer. 01:08:03.790 --> 01:08:06.310 align:middle line:90% He turns it on, runs the thing. 01:08:06.310 --> 01:08:08.430 align:middle line:84% They then auction off in 45 seconds 01:08:08.430 --> 01:08:09.900 align:middle line:90% and move on to the next car. 01:08:09.900 --> 01:08:12.630 align:middle line:84% So basically, they provide no information whatsoever 01:08:12.630 --> 01:08:16.830 align:middle line:84% on the car that's being sold other than Honda Civic 2013, 01:08:16.830 --> 01:08:18.010 align:middle line:90% 6,300 miles. 01:08:18.010 --> 01:08:19.630 align:middle line:90% Here's what the engine turns on. 01:08:19.630 --> 01:08:20.800 align:middle line:90% Here's what it sounds like. 01:08:20.800 --> 01:08:22.633 align:middle line:84% This is what it looks like from the outside. 01:08:22.633 --> 01:08:24.720 align:middle line:90% And then the next one is sold. 01:08:24.720 --> 01:08:27.250 align:middle line:84% And it's clearly an intentional design 01:08:27.250 --> 01:08:30.023 align:middle line:84% to provide zero information to the buyers. 01:08:30.023 --> 01:08:33.310 align:middle line:90% 01:08:33.310 --> 01:08:35.560 align:middle line:84% These examples, these are ones where 01:08:35.560 --> 01:08:38.290 align:middle line:84% we think of as these are clearly pure monopoly pricing 01:08:38.290 --> 01:08:43.000 align:middle line:84% decisions that somehow the mattress firm, the laptop firm, 01:08:43.000 --> 01:08:44.800 align:middle line:84% the used car firms, they've decided 01:08:44.800 --> 01:08:46.540 align:middle line:84% that providing this information to their consumers 01:08:46.540 --> 01:08:49.040 align:middle line:84% is what maximizes their profits, because they're controlling 01:08:49.040 --> 01:08:51.790 align:middle line:90% the information flow. 01:08:51.790 --> 01:08:53.560 align:middle line:84% There are other examples, now, I guess, 01:08:53.560 --> 01:08:55.510 align:middle line:84% which has [INAUDIBLE] another side, which 01:08:55.510 --> 01:08:59.950 align:middle line:84% is that there are now plenty of intermediaries that make money 01:08:59.950 --> 01:09:01.810 align:middle line:84% by providing information about products 01:09:01.810 --> 01:09:04.180 align:middle line:84% and helping consumers find products they want. 01:09:04.180 --> 01:09:08.750 align:middle line:84% So, for instance, Yelp helps people find restaurants. 01:09:08.750 --> 01:09:11.620 align:middle line:84% Amazon sells many things that are not Amazon products, 01:09:11.620 --> 01:09:15.050 align:middle line:84% helping you to find what's the best phone for me to buy? 01:09:15.050 --> 01:09:19.600 align:middle line:84% What's the best tent for me to buy? 01:09:19.600 --> 01:09:21.100 align:middle line:84% What's the best shirt for me to buy? 01:09:21.100 --> 01:09:22.725 align:middle line:84% What are the best pair of hiking boots? 01:09:22.725 --> 01:09:23.319 align:middle line:90% Whatever. 01:09:23.319 --> 01:09:25.720 align:middle line:84% Amazon is selling these products that they don't make 01:09:25.720 --> 01:09:29.069 align:middle line:84% and just providing you information about them. 01:09:29.069 --> 01:09:30.990 align:middle line:84% One concern that you can think about, 01:09:30.990 --> 01:09:32.359 align:middle line:90% what does Amazon want you to do? 01:09:32.359 --> 01:09:36.270 align:middle line:84% Amazon wants you to buy products from Amazon. 01:09:36.270 --> 01:09:38.189 align:middle line:84% Amazon therefore wants to provide you 01:09:38.189 --> 01:09:40.620 align:middle line:84% with consumer surplus when you're buying these third party 01:09:40.620 --> 01:09:42.620 align:middle line:84% products, that you buy the third party products, 01:09:42.620 --> 01:09:43.840 align:middle line:90% you get consumer surplus. 01:09:43.840 --> 01:09:45.720 align:middle line:84% You keep coming back to Amazon as a place 01:09:45.720 --> 01:09:48.933 align:middle line:90% to start your shopping trips. 01:09:48.933 --> 01:09:50.350 align:middle line:84% What information problem they have 01:09:50.350 --> 01:09:51.939 align:middle line:84% is that let's suppose they provide you 01:09:51.939 --> 01:09:54.940 align:middle line:84% with some number of information about the good. 01:09:54.940 --> 01:09:56.440 align:middle line:84% If they provide you with information 01:09:56.440 --> 01:09:58.000 align:middle line:84% that lets you figure out, yes, this 01:09:58.000 --> 01:10:01.220 align:middle line:84% is the perfect pair of hiking boots for me. 01:10:01.220 --> 01:10:03.320 align:middle line:84% Then the hiking boot company knowing 01:10:03.320 --> 01:10:07.530 align:middle line:84% that Amazon is selling it only to the consumers for whom it's 01:10:07.530 --> 01:10:09.530 align:middle line:84% the perfect product will set a really high price 01:10:09.530 --> 01:10:11.750 align:middle line:84% and the consumer surplus will be low. 01:10:11.750 --> 01:10:13.580 align:middle line:84% So in some sense, Amazon has this problem 01:10:13.580 --> 01:10:15.410 align:middle line:84% of if they provide information that 01:10:15.410 --> 01:10:19.340 align:middle line:84% lets you figure out exactly how good the product is for you 01:10:19.340 --> 01:10:22.190 align:middle line:84% and what is, in some sense, trying to maximize 01:10:22.190 --> 01:10:25.250 align:middle line:84% the profit maximizing set of information, 01:10:25.250 --> 01:10:27.650 align:middle line:84% then it may be the monopolist then just raises the price 01:10:27.650 --> 01:10:29.870 align:middle line:84% and extracts all that surplus and then 01:10:29.870 --> 01:10:31.177 align:middle line:90% the consumers get no surplus. 01:10:31.177 --> 01:10:33.260 align:middle line:84% And then Amazon isn't the place to start shopping, 01:10:33.260 --> 01:10:35.810 align:middle line:84% because it's the place where, yes, you found what you wanted, 01:10:35.810 --> 01:10:37.790 align:middle line:84% but yes, you also paid a super premium price, 01:10:37.790 --> 01:10:42.740 align:middle line:84% and therefore you didn't get any surplus anyway. 01:10:42.740 --> 01:10:44.330 align:middle line:90% So these kinds of-- 01:10:44.330 --> 01:10:47.630 align:middle line:84% I think it's the platform application and the developments 01:10:47.630 --> 01:10:50.630 align:middle line:84% in theory of this growth of what we call information design. 01:10:50.630 --> 01:10:54.020 align:middle line:84% Thinking about how would a firm that 01:10:54.020 --> 01:10:56.000 align:middle line:84% controls information about products 01:10:56.000 --> 01:10:58.400 align:middle line:84% want to provide that information either to maximize 01:10:58.400 --> 01:11:00.230 align:middle line:84% its profit, selling them directly, 01:11:00.230 --> 01:11:03.930 align:middle line:84% or to maximize the consumer surplus of the consumers who 01:11:03.930 --> 01:11:07.032 align:middle line:84% it's not selling them-- who are-- if it's an intermediary, 01:11:07.032 --> 01:11:09.240 align:middle line:84% it wants to maximize the consumer surplus rather than 01:11:09.240 --> 01:11:12.180 align:middle line:84% the profit, or it might want to do something in between. 01:11:12.180 --> 01:11:16.170 align:middle line:84% So to think about an example for illustration, 01:11:16.170 --> 01:11:18.280 align:middle line:84% let's assume that we have a new product. 01:11:18.280 --> 01:11:23.850 align:middle line:84% Maybe this new product is Clover has a new fried cauliflower 01:11:23.850 --> 01:11:26.700 align:middle line:84% sandwich with some kind of Chinese sauce on it 01:11:26.700 --> 01:11:32.360 align:middle line:84% or Indo-Chinese fried cauliflower sandwich, which 01:11:32.360 --> 01:11:34.010 align:middle line:90% I think they do have. 01:11:34.010 --> 01:11:37.250 align:middle line:84% So suppose that fully informed consumers, if they really 01:11:37.250 --> 01:11:39.080 align:middle line:84% understood what the sandwich tasted like, 01:11:39.080 --> 01:11:41.720 align:middle line:84% would have values v distributed uniform 0, 1. 01:11:41.720 --> 01:11:45.680 align:middle line:84% Or maybe in this case, uniform 0, $18. 01:11:45.680 --> 01:11:49.470 align:middle line:84% Suppose that the monopolist has a constant marginal cost of 1/3. 01:11:49.470 --> 01:11:50.980 align:middle line:90% So $6 in this example. 01:11:50.980 --> 01:11:54.250 align:middle line:90% 01:11:54.250 --> 01:11:56.710 align:middle line:84% What are you going to do if consumers are fully informed? 01:11:56.710 --> 01:11:58.630 align:middle line:90% They have values uniform 0, 1. 01:11:58.630 --> 01:12:00.850 align:middle line:90% You have a cost of 1/3. 01:12:00.850 --> 01:12:05.380 align:middle line:84% You maximize p minus 1/3 times 1 minus p. 01:12:05.380 --> 01:12:07.600 align:middle line:90% Monopoly price is 2/3. 01:12:07.600 --> 01:12:11.540 align:middle line:84% So we get this situation of your cost is 1/3. 01:12:11.540 --> 01:12:13.300 align:middle line:90% You set a price of 2/3. 01:12:13.300 --> 01:12:15.460 align:middle line:90% 1/3 of the people buy. 01:12:15.460 --> 01:12:19.580 align:middle line:84% So if I multiply by 18, I have a cost of $6. 01:12:19.580 --> 01:12:21.760 align:middle line:90% I set the price at $12. 01:12:21.760 --> 01:12:23.300 align:middle line:84% 1/3 of the people buy my sandwich. 01:12:23.300 --> 01:12:24.440 align:middle line:90% I make $6 each. 01:12:24.440 --> 01:12:25.455 align:middle line:90% I make $2 in profit. 01:12:25.455 --> 01:12:28.710 align:middle line:90% 01:12:28.710 --> 01:12:31.110 align:middle line:84% So that's clearly one thing that Clover 01:12:31.110 --> 01:12:34.410 align:middle line:84% could do is just tell everybody everything about the sandwich, 01:12:34.410 --> 01:12:36.240 align:middle line:84% somehow convey to them what it tastes like. 01:12:36.240 --> 01:12:38.573 align:middle line:84% I don't know quite how they do that, but they convey it. 01:12:38.573 --> 01:12:39.750 align:middle line:90% People get these values. 01:12:39.750 --> 01:12:41.610 align:middle line:84% Maybe they give free sandwiches to everyone, 01:12:41.610 --> 01:12:43.980 align:middle line:90% so they all learn their values. 01:12:43.980 --> 01:12:46.588 align:middle line:84% They learn their v's and then they monopoly price 01:12:46.588 --> 01:12:47.630 align:middle line:90% relative to those values. 01:12:47.630 --> 01:12:52.730 align:middle line:90% 01:12:52.730 --> 01:12:55.250 align:middle line:90% What else could you do? 01:12:55.250 --> 01:13:00.923 align:middle line:84% Imagine that consumers learn their values. 01:13:00.923 --> 01:13:04.340 align:middle line:90% 01:13:04.340 --> 01:13:06.350 align:middle line:84% Consumers have to learn their values somewhere. 01:13:06.350 --> 01:13:07.950 align:middle line:84% I know there's a new Clover sandwich. 01:13:07.950 --> 01:13:10.010 align:middle line:84% How would I know whether I like it or not? 01:13:10.010 --> 01:13:12.830 align:middle line:84% I know that I'm rational, so I know that v 01:13:12.830 --> 01:13:15.823 align:middle line:90% is distributed uniform 0, 18. 01:13:15.823 --> 01:13:17.990 align:middle line:84% But I don't know whether I'm one of the people who's 01:13:17.990 --> 01:13:21.440 align:middle line:84% going to this sandwich or one who's not going to the sandwich. 01:13:21.440 --> 01:13:23.480 align:middle line:84% Imagine one thing Clover could do 01:13:23.480 --> 01:13:27.392 align:middle line:84% is provide no information whatsoever about the sandwich. 01:13:27.392 --> 01:13:29.100 align:middle line:84% Don't even tell you it's got cauliflower. 01:13:29.100 --> 01:13:30.683 align:middle line:84% Don't even tell you what the spice is. 01:13:30.683 --> 01:13:33.320 align:middle line:84% Say, this is the new Clover sandwich. 01:13:33.320 --> 01:13:34.790 align:middle line:84% If you're a rational, you're going 01:13:34.790 --> 01:13:37.645 align:middle line:84% to know that, OK, there's a new sandwich out there. 01:13:37.645 --> 01:13:40.970 align:middle line:90% 01:13:40.970 --> 01:13:42.950 align:middle line:84% My value must be somewhere between 0 and 18 01:13:42.950 --> 01:13:44.450 align:middle line:90% for that sandwich. 01:13:44.450 --> 01:13:49.540 align:middle line:90% I don't know what that value is. 01:13:49.540 --> 01:13:51.610 align:middle line:90% Or uniform 0, 1. 01:13:51.610 --> 01:13:55.380 align:middle line:84% So here I would think on average that sandwich is worth $9 to me 01:13:55.380 --> 01:13:55.880 align:middle line:90% not knowing. 01:13:55.880 --> 01:13:57.172 align:middle line:90% It's just in a brown paper bag. 01:13:57.172 --> 01:13:58.030 align:middle line:90% They've kept. 01:13:58.030 --> 01:14:01.450 align:middle line:90% It's a secret sandwich. 01:14:01.450 --> 01:14:05.140 align:middle line:84% My value is 1/2 or 9 in that example. 01:14:05.140 --> 01:14:07.330 align:middle line:84% So what the monopolist can do is sell 01:14:07.330 --> 01:14:09.130 align:middle line:84% these sandwiches in the brown paper bags 01:14:09.130 --> 01:14:11.970 align:middle line:90% to everyone at a price of 9. 01:14:11.970 --> 01:14:14.880 align:middle line:84% And so instead of selling it to the-- 01:14:14.880 --> 01:14:17.400 align:middle line:84% I had the previous example, I'd sell it to 1/3 of the people 01:14:17.400 --> 01:14:23.550 align:middle line:84% at a price of 12 and earn 1/3 times 12 minus 6 is $2 profit. 01:14:23.550 --> 01:14:28.320 align:middle line:84% Now everyone is willing to buy the sandwich in the paper bag. 01:14:28.320 --> 01:14:30.235 align:middle line:84% They're only willing to pay $9 for it 01:14:30.235 --> 01:14:31.860 align:middle line:84% because they don't know whether they're 01:14:31.860 --> 01:14:33.735 align:middle line:84% someone who likes the sandwich or someone who 01:14:33.735 --> 01:14:34.780 align:middle line:90% hates the sandwich. 01:14:34.780 --> 01:14:39.190 align:middle line:84% But I get 1/3 times 9 minus 6 is 3. 01:14:39.190 --> 01:14:42.705 align:middle line:84% So I've made more money by hiding what my product is. 01:14:42.705 --> 01:14:45.728 align:middle line:90% 01:14:45.728 --> 01:14:48.780 align:middle line:84% And how I made more money by hiding what my product is, 01:14:48.780 --> 01:14:52.380 align:middle line:84% well, social surplus has gone down. 01:14:52.380 --> 01:14:56.240 align:middle line:84% There are all these people with low values for the sandwich, 01:14:56.240 --> 01:14:58.200 align:middle line:90% values below 6. 01:14:58.200 --> 01:14:59.910 align:middle line:84% That sandwich should not be sold to them. 01:14:59.910 --> 01:15:01.790 align:middle line:84% There's this inefficient production 01:15:01.790 --> 01:15:04.760 align:middle line:84% that there are people who don't even 01:15:04.760 --> 01:15:07.820 align:middle line:84% like the sandwich it's being produced and sold to. 01:15:07.820 --> 01:15:11.600 align:middle line:84% But I've eliminated the monopoly deadweight loss 01:15:11.600 --> 01:15:14.330 align:middle line:84% that came from these people with values between 6 and 12 01:15:14.330 --> 01:15:17.160 align:middle line:84% not getting the sandwich because monopoly pricing at 12. 01:15:17.160 --> 01:15:21.650 align:middle line:84% And so because I'm extracting all the consumer surplus, 01:15:21.650 --> 01:15:23.930 align:middle line:84% I've driven consumer surplus to 0, 01:15:23.930 --> 01:15:26.640 align:middle line:84% total social surplus has gone down. 01:15:26.640 --> 01:15:30.080 align:middle line:84% But because consumer surplus went down even faster 01:15:30.080 --> 01:15:35.750 align:middle line:84% than social surplus went up, suppressing information 01:15:35.750 --> 01:15:39.530 align:middle line:84% is transferred surplus and the monopolist is better off. 01:15:39.530 --> 01:15:41.210 align:middle line:84% So this is just an extreme example 01:15:41.210 --> 01:15:45.330 align:middle line:84% where I hide information about my product, my profits go up, 01:15:45.330 --> 01:15:48.580 align:middle line:84% and it goes up because I've made consumers homogeneous. 01:15:48.580 --> 01:15:51.940 align:middle line:84% By making them homogeneous, I can extract all their surplus. 01:15:51.940 --> 01:15:55.953 align:middle line:84% And that's socially wasteful, but it's 01:15:55.953 --> 01:15:57.870 align:middle line:84% a transfer from the consumer to the monopolist 01:15:57.870 --> 01:15:59.495 align:middle line:84% and it makes the monopolist better off. 01:15:59.495 --> 01:16:03.816 align:middle line:90% 01:16:03.816 --> 01:16:08.030 align:middle line:84% So anyway, that's one thing Clover could do here. 01:16:08.030 --> 01:16:11.750 align:middle line:84% Second observation is that you can also have examples like this 01:16:11.750 --> 01:16:17.890 align:middle line:84% where providing less than full information 01:16:17.890 --> 01:16:20.260 align:middle line:84% actually makes the consumer surplus go up. 01:16:20.260 --> 01:16:21.360 align:middle line:90% So here's an example. 01:16:21.360 --> 01:16:25.600 align:middle line:84% Again, think about Amazon or Yelp trying 01:16:25.600 --> 01:16:29.190 align:middle line:84% to sell through Clover, provide information 01:16:29.190 --> 01:16:31.650 align:middle line:84% about Clover sandwiches to people, 01:16:31.650 --> 01:16:39.830 align:middle line:84% and give the consumers more surplus by using the platform. 01:16:39.830 --> 01:16:43.040 align:middle line:84% So suppose what happens here is that-- 01:16:43.040 --> 01:16:47.510 align:middle line:84% so think of Yelp as the firm designing the information 01:16:47.510 --> 01:16:48.560 align:middle line:90% structure. 01:16:48.560 --> 01:16:52.070 align:middle line:84% And suppose that Yelp, because they've read all your reviews 01:16:52.070 --> 01:16:54.770 align:middle line:84% that you've ever written of restaurants in the past, 01:16:54.770 --> 01:16:59.540 align:middle line:84% they know how much you're going to like the Clover sandwich. 01:16:59.540 --> 01:17:01.400 align:middle line:84% They've got enough things of people 01:17:01.400 --> 01:17:04.110 align:middle line:84% who both ate this sandwich and had the same taste as you 01:17:04.110 --> 01:17:09.110 align:middle line:84% at all previous restaurants, that Yelp knows whether or not 01:17:09.110 --> 01:17:11.570 align:middle line:84% you're going to like the sandwich. 01:17:11.570 --> 01:17:13.463 align:middle line:84% Because in some sense, they implicitly 01:17:13.463 --> 01:17:15.505 align:middle line:84% know what it tastes like relative to your tastes. 01:17:15.505 --> 01:17:18.300 align:middle line:90% 01:17:18.300 --> 01:17:19.800 align:middle line:84% And so suppose what Yelp does is it 01:17:19.800 --> 01:17:23.310 align:middle line:84% commits to this partial information policy. 01:17:23.310 --> 01:17:28.185 align:middle line:84% If your value is between 0 and 1/3 or 0 and $6, it tells you, 01:17:28.185 --> 01:17:31.050 align:middle line:84% you are someone who is going to have a low value for that 01:17:31.050 --> 01:17:32.870 align:middle line:90% sandwich. 01:17:32.870 --> 01:17:36.450 align:middle line:84% If your value is between 1/3 and 7/9, 01:17:36.450 --> 01:17:40.670 align:middle line:84% so that's between $6 and $14, it's going to you, 01:17:40.670 --> 01:17:43.410 align:middle line:84% you have a medium value for that sandwich. 01:17:43.410 --> 01:17:46.410 align:middle line:84% And if your value is between $14 and $18, it's going to tell you, 01:17:46.410 --> 01:17:48.370 align:middle line:84% you have a high value for that sandwich. 01:17:48.370 --> 01:17:50.850 align:middle line:84% So it's like it's taken the sandwich, 01:17:50.850 --> 01:17:53.980 align:middle line:84% it's wrapped it in a paper bag and not told you what it is, 01:17:53.980 --> 01:17:56.790 align:middle line:84% but it says, based on what we know about you, 01:17:56.790 --> 01:17:58.795 align:middle line:84% your value for this sandwich is low. 01:17:58.795 --> 01:18:00.420 align:middle line:84% Your value for that sandwich is medium. 01:18:00.420 --> 01:18:02.760 align:middle line:84% Your value of that sandwich is high. 01:18:02.760 --> 01:18:04.380 align:middle line:84% And then it sends you into Clover 01:18:04.380 --> 01:18:08.490 align:middle line:84% with that knowledge of you only can buy from Clover through Yelp 01:18:08.490 --> 01:18:09.780 align:middle line:90% in this example. 01:18:09.780 --> 01:18:12.450 align:middle line:84% I mean, it's more plausible with Amazon. 01:18:12.450 --> 01:18:15.750 align:middle line:84% And so all you're provided with is the information of my value 01:18:15.750 --> 01:18:18.690 align:middle line:84% is low, my value is medium, or my value is high. 01:18:18.690 --> 01:18:23.640 align:middle line:84% So what does Clover do when it's in this world? 01:18:23.640 --> 01:18:27.210 align:middle line:84% Well, the people who are told the value is low think that 01:18:27.210 --> 01:18:33.410 align:middle line:90% the value is 1/6 or 3. 01:18:33.410 --> 01:18:36.350 align:middle line:84% Those who think that their value is medium think their value is 01:18:36.350 --> 01:18:40.740 align:middle line:84% 5/9, halfway between 3/9 and 7/9. 01:18:40.740 --> 01:18:43.910 align:middle line:84% And those who are told the value is high in expectation think 01:18:43.910 --> 01:18:47.000 align:middle line:84% their utility is going to be 8/9. 01:18:47.000 --> 01:18:49.770 align:middle line:84% So Clover is facing a strange population 01:18:49.770 --> 01:18:54.880 align:middle line:84% where we now have, instead of having consumers 01:18:54.880 --> 01:18:59.770 align:middle line:84% with values uniform on 0 to 18, we now just have 3 point masses. 01:18:59.770 --> 01:19:04.270 align:middle line:84% We have a point mass of 1/3 on people who think it's worth $3 01:19:04.270 --> 01:19:05.800 align:middle line:90% to them. 01:19:05.800 --> 01:19:10.480 align:middle line:84% We have a point mass that's this big. 01:19:10.480 --> 01:19:13.930 align:middle line:84% 4/9 of the people who think the thing is worth 10. 01:19:13.930 --> 01:19:17.950 align:middle line:84% And then we have a smaller point mass 2/9 who think the thing is 01:19:17.950 --> 01:19:18.873 align:middle line:90% worth 16. 01:19:18.873 --> 01:19:22.330 align:middle line:90% 01:19:22.330 --> 01:19:24.930 align:middle line:84% So what does Clover do if it's faced with this demand? 01:19:24.930 --> 01:19:26.680 align:middle line:84% There are some people who think it's worth 01:19:26.680 --> 01:19:31.450 align:middle line:84% 3, some who think it's worth 10, some who think it's worth 16. 01:19:31.450 --> 01:19:34.630 align:middle line:84% Well, obviously, you're going to price the good either 01:19:34.630 --> 01:19:38.530 align:middle line:90% at 10 or 16 in this example. 01:19:38.530 --> 01:19:41.410 align:middle line:84% Because everyone has a value of-- if I price a value 01:19:41.410 --> 01:19:42.910 align:middle line:84% more than 10, only these people buy, 01:19:42.910 --> 01:19:44.740 align:middle line:90% so I might as well charge 16. 01:19:44.740 --> 01:19:46.850 align:middle line:90% So I basically have two choices. 01:19:46.850 --> 01:19:52.060 align:middle line:84% I can price it at 10 and earn 10 to the minus 6 and sell it 01:19:52.060 --> 01:19:54.800 align:middle line:90% to 2/3 of the people. 01:19:54.800 --> 01:20:01.670 align:middle line:84% Or I can price it at 16 and get 16 minus 6 and then sell it 01:20:01.670 --> 01:20:02.815 align:middle line:90% to 2/9 of the people. 01:20:02.815 --> 01:20:06.990 align:middle line:90% 01:20:06.990 --> 01:20:11.040 align:middle line:84% And I think the way I did these numbers, selling 01:20:11.040 --> 01:20:15.180 align:middle line:84% to both groups at 10 is better for them than selling 01:20:15.180 --> 01:20:17.920 align:middle line:90% to only the high types at 16. 01:20:17.920 --> 01:20:24.070 align:middle line:84% So what happens is I get them to sell the sandwich to both 01:20:24.070 --> 01:20:27.500 align:middle line:90% the medium and the high types. 01:20:27.500 --> 01:20:29.810 align:middle line:84% And if you look at what I've done, 01:20:29.810 --> 01:20:32.450 align:middle line:84% I've worked out the calculations here. 01:20:32.450 --> 01:20:35.120 align:middle line:84% I've gotten them to sell the sandwich. 01:20:35.120 --> 01:20:38.780 align:middle line:84% Instead of having the monopoly price of 12, 01:20:38.780 --> 01:20:40.970 align:middle line:84% the monopoly price was sell it at 12 and everyone 01:20:40.970 --> 01:20:44.810 align:middle line:90% with a value above 12 buys. 01:20:44.810 --> 01:20:46.790 align:middle line:84% Here I've gotten them to sell the good 01:20:46.790 --> 01:20:51.270 align:middle line:90% at a price of 10 instead of 12. 01:20:51.270 --> 01:20:53.340 align:middle line:90% And so by getting them to-- 01:20:53.340 --> 01:20:55.950 align:middle line:84% by some sense, creating this extra mass of consumers 01:20:55.950 --> 01:20:58.155 align:middle line:84% who think it's worth 10, I've shifted down 01:20:58.155 --> 01:21:00.030 align:middle line:84% where the monopoly price is and gotten clover 01:21:00.030 --> 01:21:02.520 align:middle line:84% to sell the good at 10 instead of 12. 01:21:02.520 --> 01:21:06.960 align:middle line:84% And it turns out, if you work out the algebra here, 01:21:06.960 --> 01:21:13.620 align:middle line:84% the profit of the monopolist is, I believe, lower. 01:21:13.620 --> 01:21:14.340 align:middle line:90% Nope. 01:21:14.340 --> 01:21:16.030 align:middle line:90% So actually, sorry, it's here. 01:21:16.030 --> 01:21:18.750 align:middle line:84% So consumer surplus is higher than it 01:21:18.750 --> 01:21:19.900 align:middle line:90% is under monopoly pricing. 01:21:19.900 --> 01:21:22.358 align:middle line:84% And actually, in this example, the profit of the monopolist 01:21:22.358 --> 01:21:23.460 align:middle line:90% is also higher. 01:21:23.460 --> 01:21:26.530 align:middle line:84% So this was the no information treatment. 01:21:26.530 --> 01:21:29.170 align:middle line:84% This was the full information treatment. 01:21:29.170 --> 01:21:34.230 align:middle line:84% This one I've actually bumped up consumer surplus and profits 01:21:34.230 --> 01:21:37.470 align:middle line:84% relative to the monopolist is done. 01:21:37.470 --> 01:21:41.040 align:middle line:84% Anyway, so I am out of time now, so why don't I leave it there? 01:21:41.040 --> 01:21:46.940 align:middle line:84% I will do my final slides on information design next time. 01:21:46.940 --> 01:21:52.080 align:middle line:84% And I should say jumping ahead, as I said, 01:21:52.080 --> 01:21:54.780 align:middle line:84% I'm going to do theory on Wednesdays and empirical work 01:21:54.780 --> 01:21:57.055 align:middle line:84% on Mondays for the immediate future. 01:21:57.055 --> 01:21:58.680 align:middle line:84% So in Monday's class, I'm going to talk 01:21:58.680 --> 01:22:01.200 align:middle line:84% about two empirical papers on monopoly pricing, 01:22:01.200 --> 01:22:04.530 align:middle line:84% Chevalier-Goolsbee and Gentzkow-Shapiro. 01:22:04.530 --> 01:22:07.080 align:middle line:90% So be useful for you to-- 01:22:07.080 --> 01:22:08.850 align:middle line:84% I will discuss what's in the papers, 01:22:08.850 --> 01:22:11.635 align:middle line:84% but probably useful for you to try to look at the papers 01:22:11.635 --> 01:22:13.260 align:middle line:84% beforehand, try to read what's in them, 01:22:13.260 --> 01:22:15.420 align:middle line:84% get some sense for reading IO papers. 01:22:15.420 --> 01:22:18.090 align:middle line:84% I will say that Gentzkow-Shapiro is a bit difficult paper 01:22:18.090 --> 01:22:20.700 align:middle line:84% to read if you haven't done a lot of econometrics. 01:22:20.700 --> 01:22:23.280 align:middle line:84% We are covering the demand estimation 01:22:23.280 --> 01:22:24.867 align:middle line:90% in week four of this course. 01:22:24.867 --> 01:22:26.700 align:middle line:84% And in the first three weeks of this course, 01:22:26.700 --> 01:22:28.825 align:middle line:84% you may be reading papers on demand estimation that 01:22:28.825 --> 01:22:29.710 align:middle line:90% are harder to read. 01:22:29.710 --> 01:22:31.320 align:middle line:84% I would just in the first three weeks 01:22:31.320 --> 01:22:34.230 align:middle line:84% skim through them if you haven't done GMM estimation and things 01:22:34.230 --> 01:22:37.250 align:middle line:84% and have trouble reading those papers. 01:22:37.250 --> 01:22:37.870 align:middle line:90% OK. 01:22:37.870 --> 01:22:38.530 align:middle line:90% Thanks. 01:22:38.530 --> 01:22:41.540 align:middle line:84% Hope you enjoy the course, and I will see you on Monday. 01:22:41.540 --> 01:22:48.000 align:middle line:90%