WEBVTT 00:00:00.000 --> 00:00:02.400 align:middle line:90% [SQUEAKING] 00:00:02.400 --> 00:00:04.800 align:middle line:90% [RUSTLING] 00:00:04.800 --> 00:00:07.200 align:middle line:90% [CLICKING] 00:00:07.200 --> 00:00:11.612 align:middle line:90% 00:00:11.612 --> 00:00:13.570 align:middle line:84% GLENN ELLISON: Let me go ahead and get started. 00:00:13.570 --> 00:00:19.040 align:middle line:84% So today's topic is called strategic investment. 00:00:19.040 --> 00:00:21.920 align:middle line:84% So if you remember where we were last week 00:00:21.920 --> 00:00:25.710 align:middle line:84% is we were talking about the effect of entry on markets. 00:00:25.710 --> 00:00:28.580 align:middle line:84% And I made this big pitch that in some sense, 00:00:28.580 --> 00:00:30.140 align:middle line:84% entry is just critically important. 00:00:30.140 --> 00:00:32.432 align:middle line:84% And if you think about, what are firms' actual profits, 00:00:32.432 --> 00:00:34.920 align:middle line:84% firms' actual profits are just a rounding error. 00:00:34.920 --> 00:00:38.248 align:middle line:84% And so it's really the amount of market power 00:00:38.248 --> 00:00:40.790 align:middle line:84% you have relative to the number of firms in the market that's 00:00:40.790 --> 00:00:43.100 align:middle line:90% going to determine profits. 00:00:43.100 --> 00:00:48.290 align:middle line:84% This means that if you could affect entry, 00:00:48.290 --> 00:00:50.970 align:middle line:84% it could have a dramatic effect on how firms are doing. 00:00:50.970 --> 00:00:54.620 align:middle line:84% And I think in that previous lecture, 00:00:54.620 --> 00:00:56.940 align:middle line:84% it was always just that there was just fixed costs. 00:00:56.940 --> 00:00:59.660 align:middle line:84% Firms came in, paid the same-- paid some exogenous fixed costs, 00:00:59.660 --> 00:01:01.370 align:middle line:90% and then there was competition. 00:01:01.370 --> 00:01:03.950 align:middle line:84% There's a very, very large literature 00:01:03.950 --> 00:01:06.830 align:middle line:84% in economics about what can firms do 00:01:06.830 --> 00:01:08.910 align:middle line:90% to change the market structure. 00:01:08.910 --> 00:01:12.240 align:middle line:84% And that can either be actually changing the market structure 00:01:12.240 --> 00:01:14.940 align:middle line:84% to try to affect entry or just changing the market structure 00:01:14.940 --> 00:01:17.400 align:middle line:84% to try to affect what the competitive game looks 00:01:17.400 --> 00:01:20.700 align:middle line:84% like if you can't stop other firms from entering. 00:01:20.700 --> 00:01:23.700 align:middle line:84% And if you look at Tirole's book, 00:01:23.700 --> 00:01:26.190 align:middle line:84% the chapter on strategic entry deterrence 00:01:26.190 --> 00:01:28.000 align:middle line:90% is by far the largest chapter. 00:01:28.000 --> 00:01:31.860 align:middle line:84% And it's many more things written in the 35 years 00:01:31.860 --> 00:01:32.530 align:middle line:90% since then. 00:01:32.530 --> 00:01:35.880 align:middle line:84% So anyway, many of these models are 00:01:35.880 --> 00:01:38.850 align:middle line:90% the classic three-stage game. 00:01:38.850 --> 00:01:43.680 align:middle line:84% In the first stage, firm 1 chooses some investment k1. 00:01:43.680 --> 00:01:46.800 align:middle line:90% It pays some sunk cost c of k1. 00:01:46.800 --> 00:01:49.640 align:middle line:90% And. 00:01:49.640 --> 00:01:52.680 align:middle line:84% When I use the word investment, I mean investment very broadly. 00:01:52.680 --> 00:01:54.470 align:middle line:84% The important things about the investment 00:01:54.470 --> 00:01:58.850 align:middle line:84% is that it some sunk cost of making it, 00:01:58.850 --> 00:02:03.320 align:middle line:84% and it's credible that you're not going to be able to undo it. 00:02:03.320 --> 00:02:06.230 align:middle line:84% So, for instance building capacity, 00:02:06.230 --> 00:02:08.630 align:middle line:84% if you build a new factory, you've 00:02:08.630 --> 00:02:11.367 align:middle line:84% built the new factory you're not going to unbuild the factory. 00:02:11.367 --> 00:02:12.950 align:middle line:84% And so if you build a new factory that 00:02:12.950 --> 00:02:15.140 align:middle line:84% lets you produce a lower cost you've 00:02:15.140 --> 00:02:17.570 align:middle line:84% then committed yourself to being able to produce a lower 00:02:17.570 --> 00:02:19.310 align:middle line:90% cost in the future. 00:02:19.310 --> 00:02:24.170 align:middle line:84% If you advertise, you spend money advertising, 00:02:24.170 --> 00:02:26.900 align:middle line:84% you've changed the consumer's willingness 00:02:26.900 --> 00:02:28.400 align:middle line:84% to pay for your good and willingness 00:02:28.400 --> 00:02:29.970 align:middle line:84% to pay for other goods in the market, 00:02:29.970 --> 00:02:32.750 align:middle line:84% you can't unadvertise and get the money back 00:02:32.750 --> 00:02:35.960 align:middle line:84% and then say, OK, now the other firm's done something. 00:02:35.960 --> 00:02:37.640 align:middle line:84% I'll unadvertise, get my money back, 00:02:37.640 --> 00:02:39.820 align:middle line:84% and let the preferences go back to where they were. 00:02:39.820 --> 00:02:41.480 align:middle line:84% So anything that has that feature 00:02:41.480 --> 00:02:51.060 align:middle line:84% that you make some initial choice, it incurs a cost, 00:02:51.060 --> 00:02:54.460 align:middle line:84% and it's hard to undo, that's an investment. 00:02:54.460 --> 00:02:57.150 align:middle line:84% So for instance, like designing an iPhone 00:02:57.150 --> 00:03:01.950 align:middle line:84% with a lightning charger instead of a USB-C charger 00:03:01.950 --> 00:03:02.910 align:middle line:90% or something like that. 00:03:02.910 --> 00:03:05.050 align:middle line:90% You can't undesign the iPhone. 00:03:05.050 --> 00:03:07.410 align:middle line:84% You could design a new iPhone that costs you more money, 00:03:07.410 --> 00:03:09.368 align:middle line:84% but you've got the old ones out there that have 00:03:09.368 --> 00:03:13.050 align:middle line:90% that particular port in them. 00:03:13.050 --> 00:03:16.050 align:middle line:84% Firm 2 observes the investment level 00:03:16.050 --> 00:03:18.240 align:middle line:84% and then decides whether to enter. 00:03:18.240 --> 00:03:20.490 align:middle line:84% If it enters, there's a cost of E. 00:03:20.490 --> 00:03:23.110 align:middle line:84% I'm going to do this in the monopoly-going-to-duopoly case. 00:03:23.110 --> 00:03:25.610 align:middle line:84% Obviously, you could do all this with two firms, three firms 00:03:25.610 --> 00:03:27.030 align:middle line:90% going to n plus 1 firms. 00:03:27.030 --> 00:03:29.760 align:middle line:84% And then if firm 2 stays out of the market, 00:03:29.760 --> 00:03:31.980 align:middle line:90% firm 1 becomes a monopolist. 00:03:31.980 --> 00:03:35.850 align:middle line:84% It chooses some action, a1, like a price or a quantity, 00:03:35.850 --> 00:03:38.400 align:middle line:84% and then gets a profit that depends on the investment 00:03:38.400 --> 00:03:40.080 align:middle line:90% and on the action. 00:03:40.080 --> 00:03:43.770 align:middle line:84% And if firm 2 enters, they simultaneously 00:03:43.770 --> 00:03:44.850 align:middle line:90% choose these actions. 00:03:44.850 --> 00:03:47.460 align:middle line:84% And they earn profits that depend on the investment 00:03:47.460 --> 00:03:49.990 align:middle line:90% and on both actions. 00:03:49.990 --> 00:03:52.270 align:middle line:84% And I'm going to assume that this game that 00:03:52.270 --> 00:03:55.750 align:middle line:84% follows in second stage has well-behaved profit functions, 00:03:55.750 --> 00:03:58.690 align:middle line:84% and it has a unique Nash equilibrium, where the Nash 00:03:58.690 --> 00:04:01.650 align:middle line:90% equilibrium depends on k1. 00:04:01.650 --> 00:04:04.570 align:middle line:90% 00:04:04.570 --> 00:04:11.290 align:middle line:84% And you'll typically-- if the action is an investment that 00:04:11.290 --> 00:04:13.420 align:middle line:84% firm 1 takes in reducing its costs, 00:04:13.420 --> 00:04:15.010 align:middle line:90% the prices that they set are-- 00:04:15.010 --> 00:04:18.310 align:middle line:84% firm 1's action obviously reflects its costs. 00:04:18.310 --> 00:04:20.200 align:middle line:84% But then firm 2's action is also going 00:04:20.200 --> 00:04:22.630 align:middle line:84% to reflect firm 1's costs because firm 2 knows 00:04:22.630 --> 00:04:28.480 align:middle line:84% that firm 1's investment is going to reduce its cost, 00:04:28.480 --> 00:04:30.580 align:middle line:90% leading to change its price. 00:04:30.580 --> 00:04:34.010 align:middle line:84% There can also be direct effects of the investment on both firms. 00:04:34.010 --> 00:04:36.460 align:middle line:84% So in the advertising case, the advertising 00:04:36.460 --> 00:04:38.968 align:middle line:84% can be changing the demand for firm 2's product 00:04:38.968 --> 00:04:40.510 align:middle line:84% or the valuation of firm 2's product, 00:04:40.510 --> 00:04:44.560 align:middle line:84% as well as the valuation of firm 1's product. 00:04:44.560 --> 00:04:48.550 align:middle line:84% So in the early literature on this, 00:04:48.550 --> 00:04:51.220 align:middle line:84% Joe Bain described three different situations 00:04:51.220 --> 00:04:53.170 align:middle line:84% that you could see in an industry. 00:04:53.170 --> 00:04:55.820 align:middle line:84% The first one he described as blockaded entry. 00:04:55.820 --> 00:04:58.540 align:middle line:84% In the blockaded entry case, firms 00:04:58.540 --> 00:05:01.420 align:middle line:84% 2's entry costs are sufficiently high that firm 2 is never 00:05:01.420 --> 00:05:04.270 align:middle line:84% going to enter if firm 1 just ignores it. 00:05:04.270 --> 00:05:06.040 align:middle line:84% So in the blockaded entry case, firm 1 00:05:06.040 --> 00:05:09.165 align:middle line:84% just is-- essentially this is the monopoly model. 00:05:09.165 --> 00:05:09.790 align:middle line:90% You just go on. 00:05:09.790 --> 00:05:11.140 align:middle line:84% You choose the optimal investment, 00:05:11.140 --> 00:05:12.220 align:middle line:84% and then you earn the profits you 00:05:12.220 --> 00:05:14.012 align:middle line:84% get with the optimal investment, and firm 2 00:05:14.012 --> 00:05:16.510 align:middle line:90% is basically irrelevant. 00:05:16.510 --> 00:05:20.180 align:middle line:84% Second case he talks about is the deterred entry case. 00:05:20.180 --> 00:05:23.380 align:middle line:84% So suppose entry costs are a little bit lower 00:05:23.380 --> 00:05:24.880 align:middle line:84% than in the blockaded entry case, 00:05:24.880 --> 00:05:30.135 align:middle line:84% so that if firm 1 ignores firm 2, firm 2 will enter. 00:05:30.135 --> 00:05:31.510 align:middle line:84% But then he argues that there are 00:05:31.510 --> 00:05:35.860 align:middle line:84% going to be some cases where the firm can actively deter entry, 00:05:35.860 --> 00:05:38.650 align:middle line:84% where the firm can distort its investment in some way that 00:05:38.650 --> 00:05:41.500 align:middle line:84% reduces firm 2's profits, that firm 2 will no longer want 00:05:41.500 --> 00:05:45.310 align:middle line:84% to enter, and firm 1 remains a monopolist. 00:05:45.310 --> 00:05:48.580 align:middle line:84% And then the third case he called accommodated entry. 00:05:48.580 --> 00:05:50.740 align:middle line:84% Deterring entry is either impossible 00:05:50.740 --> 00:05:53.150 align:middle line:84% or it's just so costly you wouldn't want to do it. 00:05:53.150 --> 00:05:55.360 align:middle line:84% So the monopolist chooses its investment 00:05:55.360 --> 00:05:58.930 align:middle line:84% not to keep firm 2 out, but to change the nature of competition 00:05:58.930 --> 00:06:02.090 align:middle line:84% in some way that's going to earn it higher post-entry profits. 00:06:02.090 --> 00:06:08.832 align:middle line:84% And so we're going to focus on two main questions here. 00:06:08.832 --> 00:06:11.040 align:middle line:84% Essentially we're going to ignore the blockaded entry 00:06:11.040 --> 00:06:13.410 align:middle line:84% case except as a counterpoint and then 00:06:13.410 --> 00:06:16.200 align:middle line:84% ask, when do firms want to deter entry versus try 00:06:16.200 --> 00:06:17.680 align:middle line:90% to accommodate entry? 00:06:17.680 --> 00:06:21.990 align:middle line:84% And how do you use the investment, k, to do that? 00:06:21.990 --> 00:06:24.180 align:middle line:84% And obviously, the answer turns out 00:06:24.180 --> 00:06:26.370 align:middle line:84% to depend on how you use your investment. 00:06:26.370 --> 00:06:29.040 align:middle line:84% It depends on what type of investment it is 00:06:29.040 --> 00:06:30.670 align:middle line:90% and what type of game it is. 00:06:30.670 --> 00:06:33.510 align:middle line:84% And Fudenberg and Tirole have this paper from 1984 00:06:33.510 --> 00:06:36.030 align:middle line:84% that goes over a simple classification of types of games 00:06:36.030 --> 00:06:37.440 align:middle line:90% and types of investments. 00:06:37.440 --> 00:06:41.370 align:middle line:84% So I will do that before I discuss some applications. 00:06:41.370 --> 00:06:45.160 align:middle line:84% So your paper discusses two types of games. 00:06:45.160 --> 00:06:48.160 align:middle line:84% I think I've done this earlier, mentioned these terms earlier. 00:06:48.160 --> 00:06:51.450 align:middle line:84% We say that a game has strategic complements if the best response 00:06:51.450 --> 00:06:53.460 align:middle line:84% correspondences are upward sloping. 00:06:53.460 --> 00:06:55.290 align:middle line:84% And it has strategic substitutes, actually, 00:06:55.290 --> 00:06:57.790 align:middle line:84% if there are curves and if they're downward sloping. 00:06:57.790 --> 00:06:59.670 align:middle line:84% So if you remember the Hotelling model, 00:06:59.670 --> 00:07:00.970 align:middle line:90% you graph the best responses. 00:07:00.970 --> 00:07:02.290 align:middle line:90% It looks like this. 00:07:02.290 --> 00:07:05.150 align:middle line:84% The Cornell model, you graph the best responses. 00:07:05.150 --> 00:07:06.015 align:middle line:90% They look like that. 00:07:06.015 --> 00:07:10.570 align:middle line:90% 00:07:10.570 --> 00:07:12.130 align:middle line:90% Why was that? 00:07:12.130 --> 00:07:15.840 align:middle line:90% This is a simple theorem. 00:07:15.840 --> 00:07:19.090 align:middle line:84% You could do this more generally using increasing differences 00:07:19.090 --> 00:07:19.900 align:middle line:90% terminology. 00:07:19.900 --> 00:07:23.320 align:middle line:84% But a game is generally going to have strategic complements 00:07:23.320 --> 00:07:26.890 align:middle line:84% if the second cross partial, second derivative of profit 00:07:26.890 --> 00:07:30.190 align:middle line:84% with respect to ai and aj is positive and strategic 00:07:30.190 --> 00:07:33.370 align:middle line:84% substitutes if the second partial of profits with effect 00:07:33.370 --> 00:07:36.910 align:middle line:90% to ai and aj is negative. 00:07:36.910 --> 00:07:38.530 align:middle line:84% And as I said, you know, we usually 00:07:38.530 --> 00:07:41.140 align:middle line:84% think of price competition as being the complements case 00:07:41.140 --> 00:07:45.700 align:middle line:84% and quantity competition as being the substitutes case. 00:07:45.700 --> 00:07:50.410 align:middle line:84% But I would note here that the strategic complements and price 00:07:50.410 --> 00:07:54.840 align:middle line:84% competition is only there, has been there in all 00:07:54.840 --> 00:07:56.590 align:middle line:84% our previous analyses, because I've always 00:07:56.590 --> 00:07:59.320 align:middle line:84% been using linear demand curves that are separable. 00:07:59.320 --> 00:08:03.800 align:middle line:84% I'm using demand curves that look like D of pi, 00:08:03.800 --> 00:08:11.230 align:middle line:84% pj is 1/2 plus pj minus pi over 2t or something like that where 00:08:11.230 --> 00:08:15.460 align:middle line:84% there's no interaction between pi and pj, 00:08:15.460 --> 00:08:18.910 align:middle line:84% because if you look at the profit function in a price 00:08:18.910 --> 00:08:21.370 align:middle line:90% competition game, you take the-- 00:08:21.370 --> 00:08:24.040 align:middle line:84% I first differentiate with respect to pi. 00:08:24.040 --> 00:08:29.350 align:middle line:84% And I get pi minus c, dDi dpi plus D of pi. 00:08:29.350 --> 00:08:33.460 align:middle line:84% And then I differentiate this with respect to pj. 00:08:33.460 --> 00:08:37.610 align:middle line:84% And then I get-- this doesn't depend on pj, 00:08:37.610 --> 00:08:39.890 align:middle line:84% so I get p minus c, second derivative of demand 00:08:39.890 --> 00:08:43.940 align:middle line:84% with respect to pi and pj plus dDi, dpj. 00:08:43.940 --> 00:08:47.840 align:middle line:84% And we're used to this term, firm j increases its price, 00:08:47.840 --> 00:08:49.140 align:middle line:90% firm i's demand goes up. 00:08:49.140 --> 00:08:51.540 align:middle line:90% This is the positive term. 00:08:51.540 --> 00:08:56.040 align:middle line:84% But then there is this second term, pi minus c 00:08:56.040 --> 00:08:59.930 align:middle line:84% times this second derivative of Di with respect to pi and pj. 00:08:59.930 --> 00:09:07.920 align:middle line:84% So for instance, if I had an interaction in this demand 00:09:07.920 --> 00:09:10.510 align:middle line:84% curve, then this term would be negative. 00:09:10.510 --> 00:09:13.510 align:middle line:84% This term would be negative multiplied by this. 00:09:13.510 --> 00:09:16.200 align:middle line:84% It's no longer obvious you have strategic complements. 00:09:16.200 --> 00:09:20.040 align:middle line:84% So normally we think of price competition 00:09:20.040 --> 00:09:22.840 align:middle line:84% as complements and quantity competition as substitutes, 00:09:22.840 --> 00:09:24.635 align:middle line:90% but that's not always true. 00:09:24.635 --> 00:09:29.580 align:middle line:90% 00:09:29.580 --> 00:09:32.760 align:middle line:84% OK, second thing we're going to have to do in this model 00:09:32.760 --> 00:09:35.460 align:middle line:84% is classify types of investments. 00:09:35.460 --> 00:09:38.587 align:middle line:84% And the way the investment classification is going to work 00:09:38.587 --> 00:09:40.170 align:middle line:84% is we're going to classify investments 00:09:40.170 --> 00:09:43.890 align:middle line:84% based on their effect on the potential entrant. 00:09:43.890 --> 00:09:47.670 align:middle line:84% So in this model, I'm going to define this reduced form profit 00:09:47.670 --> 00:09:50.760 align:middle line:84% function to be the profit that firm I 00:09:50.760 --> 00:09:54.820 align:middle line:90% gets when firm 1 has chosen k1. 00:09:54.820 --> 00:10:01.010 align:middle line:84% And then we play the equilibrium action following k1. 00:10:01.010 --> 00:10:06.120 align:middle line:84% And so a simple definition, investment makes firm 1 tough, 00:10:06.120 --> 00:10:08.030 align:middle line:90% essentially if it hurts firm 2. 00:10:08.030 --> 00:10:11.900 align:middle line:84% If this reduced form profit function is decreasing in k, 00:10:11.900 --> 00:10:15.470 align:middle line:84% investment makes firm 1 soft if that reduced form 00:10:15.470 --> 00:10:19.430 align:middle line:90% profit is increasing in k. 00:10:19.430 --> 00:10:22.310 align:middle line:84% And you know, sometimes it's obvious 00:10:22.310 --> 00:10:23.550 align:middle line:90% which this is going to be. 00:10:23.550 --> 00:10:26.510 align:middle line:84% So for instance, if this is an investment in cost reduction, 00:10:26.510 --> 00:10:29.450 align:middle line:84% your intuition tells you if firm 1 has lower costs, that's 00:10:29.450 --> 00:10:31.940 align:middle line:84% going to be bad for firm 2 in equilibrium when 00:10:31.940 --> 00:10:33.510 align:middle line:90% the prices shake out. 00:10:33.510 --> 00:10:38.010 align:middle line:84% And so we will have d pi 2 dk1 less than 0, 00:10:38.010 --> 00:10:40.370 align:middle line:84% so cost reductions in investment, 00:10:40.370 --> 00:10:42.830 align:middle line:90% that makes you tough. 00:10:42.830 --> 00:10:46.550 align:middle line:84% If the investment is lobbying to lower the tax 00:10:46.550 --> 00:10:51.920 align:middle line:84% rates in an industry, if it's just like a profits tax, 00:10:51.920 --> 00:10:53.600 align:middle line:84% a percentage tax on profits, it won't 00:10:53.600 --> 00:10:55.250 align:middle line:90% affect any of the behavior. 00:10:55.250 --> 00:10:58.802 align:middle line:84% So if you lobby to get taxes reduced, a reduction in taxes 00:10:58.802 --> 00:11:00.510 align:middle line:84% isn't going to change any of the actions. 00:11:00.510 --> 00:11:02.860 align:middle line:84% And it's going to end up being good for both firms. 00:11:02.860 --> 00:11:07.140 align:middle line:84% So lobbying to lower tax rates would be an investment 00:11:07.140 --> 00:11:10.320 align:middle line:90% that makes you soft. 00:11:10.320 --> 00:11:13.710 align:middle line:84% When it's not obvious, it can help 00:11:13.710 --> 00:11:17.770 align:middle line:84% to write down what this profit function looks like. 00:11:17.770 --> 00:11:21.720 align:middle line:84% So if I think about differentiating, 00:11:21.720 --> 00:11:30.780 align:middle line:84% so I have pi 2 star of k1 is profit 00:11:30.780 --> 00:11:37.940 align:middle line:84% that firm 2 gets when the investment is k1. 00:11:37.940 --> 00:11:41.313 align:middle line:84% And we have a1 star of k1, a2 star of k1. 00:11:41.313 --> 00:11:46.170 align:middle line:90% 00:11:46.170 --> 00:11:48.660 align:middle line:84% OK, if I differentiate this with respect to k1, 00:11:48.660 --> 00:11:50.020 align:middle line:90% I'm going to get three terms. 00:11:50.020 --> 00:11:52.830 align:middle line:84% First, I get the derivative here, 00:11:52.830 --> 00:11:57.800 align:middle line:84% so I get the partial of pi 2 with respect to k1. 00:11:57.800 --> 00:12:02.480 align:middle line:84% Second, I get this effect that when I change k1, a1 changes. 00:12:02.480 --> 00:12:05.450 align:middle line:84% And that change in a1 affects pi 2. 00:12:05.450 --> 00:12:09.590 align:middle line:84% And so that's partial pi 2 with respect to a1 times 00:12:09.590 --> 00:12:13.430 align:middle line:84% partial a1 star with respect to k1. 00:12:13.430 --> 00:12:14.720 align:middle line:90% And then there's also-- 00:12:14.720 --> 00:12:16.553 align:middle line:84% you would think there would be a third term. 00:12:16.553 --> 00:12:18.350 align:middle line:84% When I change k1, it affects this. 00:12:18.350 --> 00:12:20.270 align:middle line:90% And that affects profits. 00:12:20.270 --> 00:12:27.590 align:middle line:84% But the third term would be partial pi 2 with respect 00:12:27.590 --> 00:12:35.840 align:middle line:84% to a2 evaluated at a2 star of k1 times 00:12:35.840 --> 00:12:46.520 align:middle line:84% derivative of a2 star with respect to k1 evaluated at k1. 00:12:46.520 --> 00:12:51.600 align:middle line:84% And if you think about this term here, that's 0. 00:12:51.600 --> 00:12:56.930 align:middle line:84% When you evaluate d pi 2 / da2 at a2 star of k1, a2 star of k1 00:12:56.930 --> 00:12:59.630 align:middle line:84% was chosen where the first order condition is 0. 00:12:59.630 --> 00:13:04.630 align:middle line:84% So there's no effect from k1 changes-- 00:13:04.630 --> 00:13:09.740 align:middle line:84% there's no third effect from k1 leads firm 2 to change a2. 00:13:09.740 --> 00:13:11.600 align:middle line:84% And that has an effect on profits. 00:13:11.600 --> 00:13:15.340 align:middle line:84% The marginal effect of changing a2 on profits is 0. 00:13:15.340 --> 00:13:18.010 align:middle line:84% So we really-- so this term is 0. 00:13:18.010 --> 00:13:19.610 align:middle line:90% And we just have two effects. 00:13:19.610 --> 00:13:23.410 align:middle line:84% We have what's called the direct effect where k1 directly 00:13:23.410 --> 00:13:27.640 align:middle line:84% affects firm 2's profits, and for the investment like cost 00:13:27.640 --> 00:13:30.820 align:middle line:90% reduction, this term is 0. 00:13:30.820 --> 00:13:33.220 align:middle line:84% Firm 1's cost reduction has no direct effect 00:13:33.220 --> 00:13:34.760 align:middle line:90% on firm 2's profits. 00:13:34.760 --> 00:13:38.440 align:middle line:84% And in the cost reduction case, it comes in through this term 00:13:38.440 --> 00:13:42.160 align:middle line:84% that when I invest more in cost reduction, 00:13:42.160 --> 00:13:44.810 align:middle line:90% I lower firm 1's price. 00:13:44.810 --> 00:13:45.700 align:middle line:90% So this is negative. 00:13:45.700 --> 00:13:49.810 align:middle line:84% When I lower my price, d pi 2 / da1 is positive. 00:13:49.810 --> 00:13:53.750 align:middle line:84% Firm 2 benefits when firm 1 sets a higher price. 00:13:53.750 --> 00:13:56.420 align:middle line:84% So the cost reduction has a negative strategic effect. 00:13:56.420 --> 00:13:59.260 align:middle line:84% It causes firm 1 to do something that's 00:13:59.260 --> 00:14:03.370 align:middle line:84% bad for firm 2, which is cut its price. 00:14:03.370 --> 00:14:05.950 align:middle line:84% In the lobbying to lower tax rates, 00:14:05.950 --> 00:14:09.020 align:middle line:84% there I was describing a thing that has a direct effect. 00:14:09.020 --> 00:14:10.480 align:middle line:84% If I lobbied lower tax rates and I 00:14:10.480 --> 00:14:13.210 align:middle line:84% get tax rates lowered, holding actions fixed, 00:14:13.210 --> 00:14:15.410 align:middle line:90% that raises firm 2's profits. 00:14:15.410 --> 00:14:18.970 align:middle line:84% And then if it's just a tax on profits, 00:14:18.970 --> 00:14:22.430 align:middle line:84% it doesn't affect equilibrium prices. 00:14:22.430 --> 00:14:25.365 align:middle line:84% So da1 / dk1 is 0, and there's no strategic effect. 00:14:25.365 --> 00:14:29.020 align:middle line:90% 00:14:29.020 --> 00:14:31.560 align:middle line:84% And so I have some examples on the other side. 00:14:31.560 --> 00:14:34.690 align:middle line:90% You can think about-- 00:14:34.690 --> 00:14:38.920 align:middle line:84% so for instance, suppose we had a competition on a line game. 00:14:38.920 --> 00:14:42.970 align:middle line:84% And firm 1 has the option to extend the line and move even 00:14:42.970 --> 00:14:45.460 align:middle line:84% further away out in this direction. 00:14:45.460 --> 00:14:46.725 align:middle line:90% What would that do? 00:14:46.725 --> 00:14:49.877 align:middle line:90% 00:14:49.877 --> 00:14:51.460 align:middle line:84% There, they're going to be-- and let's 00:14:51.460 --> 00:14:52.710 align:middle line:90% have it in the hotel and game. 00:14:52.710 --> 00:14:54.190 align:middle line:90% So firm 1 used to be here. 00:14:54.190 --> 00:14:56.320 align:middle line:84% And firm 1 is a chance to just move further away 00:14:56.320 --> 00:14:59.530 align:middle line:90% off the end of the beach. 00:14:59.530 --> 00:15:03.650 align:middle line:84% So moving further away, the direct effect is positive. 00:15:03.650 --> 00:15:07.780 align:middle line:84% If we hold the prices as fixed and firm 1 moves farther away, 00:15:07.780 --> 00:15:10.210 align:middle line:84% the midpoint of the line moves from here to here, 00:15:10.210 --> 00:15:13.860 align:middle line:84% so firm 2 is going to get more customers. 00:15:13.860 --> 00:15:18.630 align:middle line:84% And then the strategic effect, again, 00:15:18.630 --> 00:15:21.910 align:middle line:84% profit of firm 2 with respect to price of firm 1 is positive. 00:15:21.910 --> 00:15:24.030 align:middle line:84% Firm 2 likes 1 to set high prices. 00:15:24.030 --> 00:15:26.310 align:middle line:84% And the further firm 1 moves away, 00:15:26.310 --> 00:15:28.140 align:middle line:84% the higher will firm 1's price be. 00:15:28.140 --> 00:15:30.990 align:middle line:84% Therefore, this is positive, this is positive. 00:15:30.990 --> 00:15:33.610 align:middle line:84% So the direct effect in that model is positive. 00:15:33.610 --> 00:15:35.143 align:middle line:84% The strategic effect is positive. 00:15:35.143 --> 00:15:36.810 align:middle line:84% Moving away from firm 2 is an investment 00:15:36.810 --> 00:15:37.910 align:middle line:90% that makes firm 1 soft. 00:15:37.910 --> 00:15:40.875 align:middle line:90% 00:15:40.875 --> 00:15:43.990 align:middle line:84% And other things-- investments can be neither. 00:15:43.990 --> 00:15:46.947 align:middle line:84% So if you think about an advertising campaign, 00:15:46.947 --> 00:15:48.780 align:middle line:84% suppose you run an advertising campaign that 00:15:48.780 --> 00:15:50.430 align:middle line:84% will convince some number of customers 00:15:50.430 --> 00:15:52.020 align:middle line:84% to never leave your firm or to have 00:15:52.020 --> 00:15:56.190 align:middle line:84% a tremendous loyalty for your firm and not want to switch. 00:15:56.190 --> 00:15:59.290 align:middle line:84% The direct effect on your rival is negative. 00:15:59.290 --> 00:16:00.833 align:middle line:84% When those customers become captive, 00:16:00.833 --> 00:16:02.500 align:middle line:84% they're not going to consider the rival. 00:16:02.500 --> 00:16:04.625 align:middle line:84% Therefore, the rival is going to get fewer of them, 00:16:04.625 --> 00:16:05.830 align:middle line:90% holding prices fixed. 00:16:05.830 --> 00:16:08.400 align:middle line:90% That hurts the rival. 00:16:08.400 --> 00:16:11.910 align:middle line:84% But again, price competition, firm 2 00:16:11.910 --> 00:16:13.868 align:middle line:90% likes me charging high prices. 00:16:13.868 --> 00:16:15.660 align:middle line:84% And when I have these captive customers who 00:16:15.660 --> 00:16:17.118 align:middle line:84% won't consider switching, I'm going 00:16:17.118 --> 00:16:19.090 align:middle line:84% to price high to exploit the captive customers. 00:16:19.090 --> 00:16:22.030 align:middle line:84% I can't discriminate between the captive and non-captive people. 00:16:22.030 --> 00:16:24.390 align:middle line:84% Therefore, my price is going to be higher. 00:16:24.390 --> 00:16:25.980 align:middle line:84% So you get a negative direct effect, 00:16:25.980 --> 00:16:27.580 align:middle line:90% a positive strategic effect. 00:16:27.580 --> 00:16:29.735 align:middle line:90% It's not clear. 00:16:29.735 --> 00:16:31.110 align:middle line:84% It would depend on which of those 00:16:31.110 --> 00:16:32.490 align:middle line:84% is larger whether this is an investment that 00:16:32.490 --> 00:16:33.610 align:middle line:90% makes you tough or soft. 00:16:33.610 --> 00:16:34.860 align:middle line:90% And it could be neither. 00:16:34.860 --> 00:16:40.010 align:middle line:84% It could be sometimes soft in some ranges, tough in others. 00:16:40.010 --> 00:16:43.000 align:middle line:84% But just the tough, soft distinction is all about-- 00:16:43.000 --> 00:16:45.430 align:middle line:84% it's not really how it affects firm 1. 00:16:45.430 --> 00:16:46.665 align:middle line:90% It's how it affects firm 2. 00:16:46.665 --> 00:16:52.290 align:middle line:90% 00:16:52.290 --> 00:16:55.270 align:middle line:84% OK, so let me now go through Bain's cases in order, 00:16:55.270 --> 00:16:58.260 align:middle line:84% so first, this blockaded entry case. 00:16:58.260 --> 00:17:02.130 align:middle line:84% So blockaded entry case is the case where firm 2 just 00:17:02.130 --> 00:17:04.140 align:middle line:84% has very, very high entry costs and will never 00:17:04.140 --> 00:17:05.819 align:middle line:90% consider entering. 00:17:05.819 --> 00:17:09.030 align:middle line:84% So I'm going to write k1m and a1m 00:17:09.030 --> 00:17:12.130 align:middle line:84% for what a monopolist would do if there wasn't even a firm 2. 00:17:12.130 --> 00:17:15.060 align:middle line:84% It's just going to choose the a1 and the k1 00:17:15.060 --> 00:17:19.290 align:middle line:84% to maximize its monopoly profit function. 00:17:19.290 --> 00:17:21.780 align:middle line:84% And the blockaded entry case is when, 00:17:21.780 --> 00:17:26.190 align:middle line:84% if you choose this optimal monopoly investment, 00:17:26.190 --> 00:17:28.950 align:middle line:84% firm 2's equilibrium profits are less than e, 00:17:28.950 --> 00:17:31.150 align:middle line:90% so firm 2 is not going to enter. 00:17:31.150 --> 00:17:33.630 align:middle line:84% And so then this is-- really, it's not an entry model. 00:17:33.630 --> 00:17:36.100 align:middle line:84% Firm 1 just ignores potential entrant. 00:17:36.100 --> 00:17:37.530 align:middle line:90% It chooses k1m. 00:17:37.530 --> 00:17:38.555 align:middle line:90% That's all that happens. 00:17:38.555 --> 00:17:43.600 align:middle line:90% 00:17:43.600 --> 00:17:45.400 align:middle line:84% I mentioned that, because now let's 00:17:45.400 --> 00:17:47.733 align:middle line:84% think about what happens if you're going to deter entry. 00:17:47.733 --> 00:17:49.910 align:middle line:90% 00:17:49.910 --> 00:17:53.163 align:middle line:84% The optimal entry deterring investment 00:17:53.163 --> 00:17:55.080 align:middle line:84% is going-- you're going to solve this problem. 00:17:55.080 --> 00:17:58.490 align:middle line:84% I'm going to try to maximize over all possible investment 00:17:58.490 --> 00:18:01.190 align:middle line:84% levels and actions the profit that I'm 00:18:01.190 --> 00:18:05.900 align:middle line:84% going to get from choosing k1 and then choosing action a1. 00:18:05.900 --> 00:18:08.660 align:middle line:84% But I've got to do it subject to the constraint 00:18:08.660 --> 00:18:11.240 align:middle line:84% that firm 2's equilibrium profits 00:18:11.240 --> 00:18:16.290 align:middle line:84% after I've chosen k1 are at most e, so that firm 2 doesn't enter. 00:18:16.290 --> 00:18:19.075 align:middle line:84% So I'm choosing this same maximization problem, 00:18:19.075 --> 00:18:20.450 align:middle line:84% but I just put in this constraint 00:18:20.450 --> 00:18:26.780 align:middle line:84% that I have to ensure that firm 2's profit is less than e. 00:18:26.780 --> 00:18:31.100 align:middle line:84% And the entry deterrence case I'm looking at 00:18:31.100 --> 00:18:42.140 align:middle line:84% is suppose that pi 2 star of k1m is greater than e. 00:18:42.140 --> 00:18:44.820 align:middle line:84% So in this case, firm 2's profits, 00:18:44.820 --> 00:18:47.870 align:middle line:84% if I just behave as if firm 2 isn't there, 00:18:47.870 --> 00:18:49.800 align:middle line:90% firm 2 is going to enter. 00:18:49.800 --> 00:18:51.803 align:middle line:90% So this would be-- 00:18:51.803 --> 00:18:54.470 align:middle line:84% if this was less than e, I would be in the blockaded entry case. 00:18:54.470 --> 00:18:56.780 align:middle line:84% With this greater than e, this is when I 00:18:56.780 --> 00:18:58.220 align:middle line:90% need to be at this constraint. 00:18:58.220 --> 00:18:59.545 align:middle line:90% This constraint is binding. 00:18:59.545 --> 00:19:04.410 align:middle line:90% 00:19:04.410 --> 00:19:07.170 align:middle line:84% So anyway, how do you distort your investment 00:19:07.170 --> 00:19:10.290 align:middle line:90% and make k1ED differ from k1m? 00:19:10.290 --> 00:19:15.030 align:middle line:84% Very simple theorem says that if investment makes firm 1 tough, 00:19:15.030 --> 00:19:18.000 align:middle line:84% then the entry deterring level of investment 00:19:18.000 --> 00:19:19.800 align:middle line:90% is greater than k1m. 00:19:19.800 --> 00:19:21.990 align:middle line:84% And if investment makes firm 1 soft, 00:19:21.990 --> 00:19:25.595 align:middle line:84% then the entry deterring level of investment is less than k1m. 00:19:25.595 --> 00:19:32.500 align:middle line:90% 00:19:32.500 --> 00:19:36.450 align:middle line:84% An easy way to think about this is you just have-- 00:19:36.450 --> 00:19:39.060 align:middle line:84% I put my investment on this axis. 00:19:39.060 --> 00:19:42.210 align:middle line:84% And let me graph some profits on this axis. 00:19:42.210 --> 00:19:48.000 align:middle line:84% So let this be e, the level of profits 00:19:48.000 --> 00:19:51.270 align:middle line:84% that firm 2 needs to enter to come into the market. 00:19:51.270 --> 00:20:01.210 align:middle line:84% And so imagine-- you know, imagine this is k1m 00:20:01.210 --> 00:20:04.780 align:middle line:90% and this is pi 2 star of k1. 00:20:04.780 --> 00:20:08.100 align:middle line:84% So this is an investment that makes firm 1 tough, 00:20:08.100 --> 00:20:11.093 align:middle line:90% so investing, investing more-- 00:20:11.093 --> 00:20:13.960 align:middle line:90% 00:20:13.960 --> 00:20:18.160 align:middle line:90% investing more hurts firm 2. 00:20:18.160 --> 00:20:21.040 align:middle line:90% And again, if I also graph-- 00:20:21.040 --> 00:20:25.140 align:middle line:84% let's imagine pi 1 looks like this. 00:20:25.140 --> 00:20:29.250 align:middle line:90% 00:20:29.250 --> 00:20:33.490 align:middle line:84% pi 1 star of k1, it's maximized at k1 monopoly. 00:20:33.490 --> 00:20:36.030 align:middle line:84% So what I would like to do if firm 2 wasn't there 00:20:36.030 --> 00:20:38.730 align:middle line:84% is I would choose this level of investment. 00:20:38.730 --> 00:20:42.550 align:middle line:84% I would receive that level of profits. 00:20:42.550 --> 00:20:44.410 align:middle line:84% But I can't do that, because if I did, 00:20:44.410 --> 00:20:45.840 align:middle line:90% that, firm 2 would come in. 00:20:45.840 --> 00:20:47.620 align:middle line:90% So this is pi 1. 00:20:47.620 --> 00:20:52.420 align:middle line:84% We should call this pi 1 monopoly of k1. 00:20:52.420 --> 00:20:53.870 align:middle line:90% That's what I would like to do. 00:20:53.870 --> 00:20:54.890 align:middle line:90% I can't. 00:20:54.890 --> 00:20:56.030 align:middle line:90% What do I have to do? 00:20:56.030 --> 00:20:58.000 align:middle line:84% I have to choose a level of investment 00:20:58.000 --> 00:21:03.190 align:middle line:84% that is sufficient to make sure that pi 2 star is less than e. 00:21:03.190 --> 00:21:04.660 align:middle line:84% So what I would do in this example 00:21:04.660 --> 00:21:09.650 align:middle line:84% is this would be k1 entry deterrence. 00:21:09.650 --> 00:21:11.000 align:middle line:90% I just need to-- 00:21:11.000 --> 00:21:14.300 align:middle line:90% I need to move firm 2's-- 00:21:14.300 --> 00:21:16.400 align:middle line:84% I need to move firm 2's profits in whatever-- 00:21:16.400 --> 00:21:18.025 align:middle line:84% I need to move k1 in whatever direction 00:21:18.025 --> 00:21:20.330 align:middle line:90% moves firm 2's profits down. 00:21:20.330 --> 00:21:24.050 align:middle line:84% And so in this case, I distort my investment 00:21:24.050 --> 00:21:25.260 align:middle line:90% going in that direction. 00:21:25.260 --> 00:21:26.730 align:middle line:90% And so I'm over-investing. 00:21:26.730 --> 00:21:30.365 align:middle line:84% We'll call that I'm over-investing in k1 in order 00:21:30.365 --> 00:21:31.740 align:middle line:84% to keep firm 2 out of the market. 00:21:31.740 --> 00:21:35.540 align:middle line:84% So for instance, this would be the graph of I'm 00:21:35.540 --> 00:21:38.420 align:middle line:84% investing in a cost-reducing technology. 00:21:38.420 --> 00:21:41.270 align:middle line:84% And so if I-- if firm 2 wasn't there, 00:21:41.270 --> 00:21:44.543 align:middle line:84% I would just go ahead and do minimal investments in cost. 00:21:44.543 --> 00:21:46.460 align:middle line:84% Do this level of investment in cost reduction, 00:21:46.460 --> 00:21:47.970 align:middle line:90% I'd earn this profits. 00:21:47.970 --> 00:21:50.100 align:middle line:84% But then with this, if I have the cost 00:21:50.100 --> 00:21:53.000 align:middle line:84% that this investment produces, firm 2 is going to enter. 00:21:53.000 --> 00:21:55.490 align:middle line:84% So what I do is I do more cost reduction 00:21:55.490 --> 00:21:59.060 align:middle line:90% than I need to as a monopolist. 00:21:59.060 --> 00:22:03.080 align:middle line:84% I'm distorting my monopoly investment, 00:22:03.080 --> 00:22:05.430 align:middle line:84% therefore, I'm lowering my profits. 00:22:05.430 --> 00:22:07.538 align:middle line:84% But I do that because I need to do 00:22:07.538 --> 00:22:10.080 align:middle line:84% that to keep firm 2 out of the market and remain a monopolist 00:22:10.080 --> 00:22:12.570 align:middle line:84% and not be a duopoly, which in this case, I'm assuming 00:22:12.570 --> 00:22:13.700 align:middle line:90% is even worse for you. 00:22:13.700 --> 00:22:17.000 align:middle line:90% 00:22:17.000 --> 00:22:19.280 align:middle line:84% And if the investment was one that 00:22:19.280 --> 00:22:25.610 align:middle line:84% made firm 1 soft, if firm 2's profits instead 00:22:25.610 --> 00:22:28.250 align:middle line:84% looked like this, then obviously again, 00:22:28.250 --> 00:22:30.320 align:middle line:84% I would just distort down in this direction 00:22:30.320 --> 00:22:31.850 align:middle line:90% and take this point. 00:22:31.850 --> 00:22:35.900 align:middle line:84% So I'm always just distorting in whichever 00:22:35.900 --> 00:22:37.950 align:middle line:90% direction hurts firm 2. 00:22:37.950 --> 00:22:40.860 align:middle line:84% And so strategic entry deterrence is very simple. 00:22:40.860 --> 00:22:42.480 align:middle line:84% It's just about hurting your rival. 00:22:42.480 --> 00:22:43.585 align:middle line:90% That's all there is to it. 00:22:43.585 --> 00:22:49.450 align:middle line:90% 00:22:49.450 --> 00:22:54.520 align:middle line:84% As I drew this, I made firm 1's monopoly profits single peaked. 00:22:54.520 --> 00:22:56.880 align:middle line:84% When firm 1's monopoly profits are single peaked, 00:22:56.880 --> 00:22:59.680 align:middle line:84% you want to go as far away from-- 00:22:59.680 --> 00:23:03.400 align:middle line:84% you want to stay as close as you can to k1m and not distort more 00:23:03.400 --> 00:23:04.190 align:middle line:90% than you need to. 00:23:04.190 --> 00:23:05.857 align:middle line:84% So you would always go just to the point 00:23:05.857 --> 00:23:07.910 align:middle line:84% where the entry cost first hits e. 00:23:07.910 --> 00:23:13.030 align:middle line:90% 00:23:13.030 --> 00:23:14.475 align:middle line:90% Any questions on that? 00:23:14.475 --> 00:23:21.540 align:middle line:90% 00:23:21.540 --> 00:23:24.600 align:middle line:90% OK, so some examples. 00:23:24.600 --> 00:23:26.310 align:middle line:84% In the cost reduction game, I said, 00:23:26.310 --> 00:23:28.980 align:middle line:84% you know, firm 1 is going to over-invest in cost reduction 00:23:28.980 --> 00:23:31.860 align:middle line:90% to deter entry. 00:23:31.860 --> 00:23:33.990 align:middle line:84% Advertising to increase market size, so 00:23:33.990 --> 00:23:36.600 align:middle line:84% suppose you have a new product, you're 00:23:36.600 --> 00:23:38.700 align:middle line:84% advertising for the product, and advertising 00:23:38.700 --> 00:23:43.950 align:middle line:84% is going to raise demand for that product for everyone. 00:23:43.950 --> 00:23:48.210 align:middle line:84% And so if this advertising is going to make more people 00:23:48.210 --> 00:23:52.710 align:middle line:84% want whatever this product is, if that's going to help you 00:23:52.710 --> 00:23:55.410 align:middle line:84% and that's going to help rivals, if the rivals enter, 00:23:55.410 --> 00:23:58.680 align:middle line:84% then you're going to limit that advertising 00:23:58.680 --> 00:24:01.770 align:middle line:90% to avoid attracting entry. 00:24:01.770 --> 00:24:03.450 align:middle line:84% And so that's the case where investing 00:24:03.450 --> 00:24:10.590 align:middle line:84% in something that has spillovers to you or your rivals raises-- 00:24:10.590 --> 00:24:13.590 align:middle line:84% if it raises everyone's profits, then increasing 00:24:13.590 --> 00:24:16.065 align:middle line:84% the size of the market, then it's going to attract entry 00:24:16.065 --> 00:24:17.940 align:middle line:84% and therefore that investment makes you soft, 00:24:17.940 --> 00:24:20.550 align:middle line:84% you do less of it to deter entry. 00:24:20.550 --> 00:24:24.600 align:middle line:84% So, for instance, if I'm the Chipotle in Kendall Square and I 00:24:24.600 --> 00:24:27.600 align:middle line:84% just advertise "Burritos are great" more generally, 00:24:27.600 --> 00:24:30.840 align:middle line:84% could be a second burrito shop opens up in Kendall Square 00:24:30.840 --> 00:24:34.270 align:middle line:84% and therefore, everyone likes burritos more, whatever. 00:24:34.270 --> 00:24:37.620 align:middle line:84% So maybe it's not so plausible, but you just 00:24:37.620 --> 00:24:39.872 align:middle line:84% do less advertising, because the more you advertise, 00:24:39.872 --> 00:24:41.580 align:middle line:84% the bigger is the market, the more likely 00:24:41.580 --> 00:24:44.440 align:middle line:84% some other firm is going to come in. 00:24:44.440 --> 00:24:48.880 align:middle line:84% Again, differentiation-- if you can have some way that you can, 00:24:48.880 --> 00:24:53.140 align:middle line:84% either by designing your product or by advertising your product, 00:24:53.140 --> 00:24:55.390 align:middle line:84% make consumers have idiosyncratic preference 00:24:55.390 --> 00:25:00.360 align:middle line:84% for your product relative to rivals' products, 00:25:00.360 --> 00:25:04.530 align:middle line:84% differentiating also will raise the-- differentiation will-- 00:25:04.530 --> 00:25:06.990 align:middle line:84% in c plus t, if you increase t, it 00:25:06.990 --> 00:25:08.820 align:middle line:90% will raise the rivals' profits. 00:25:08.820 --> 00:25:12.680 align:middle line:84% Therefore, you want to avoid differentiating to deter entry. 00:25:12.680 --> 00:25:15.470 align:middle line:90% 00:25:15.470 --> 00:25:18.980 align:middle line:84% Like in a competition on a line model, 00:25:18.980 --> 00:25:21.200 align:middle line:84% it could be that the thing I want to do 00:25:21.200 --> 00:25:23.797 align:middle line:84% is locate exactly in the middle, so that if I locate exactly 00:25:23.797 --> 00:25:26.130 align:middle line:84% in the middle, then firm 2 has to be pretty close to me, 00:25:26.130 --> 00:25:27.570 align:middle line:90% and then we don't earn profits. 00:25:27.570 --> 00:25:29.665 align:middle line:84% That keeps the firm out of the market. 00:25:29.665 --> 00:25:33.670 align:middle line:90% 00:25:33.670 --> 00:25:35.540 align:middle line:90% Any questions? 00:25:35.540 --> 00:25:40.620 align:middle line:90% 00:25:40.620 --> 00:25:44.820 align:middle line:84% So accommodating entry, what's going 00:25:44.820 --> 00:25:47.730 align:middle line:84% to happen in this game is if you think about this, 00:25:47.730 --> 00:25:55.295 align:middle line:84% suppose I reduce e, and I make e a lot lower. 00:25:55.295 --> 00:25:58.240 align:middle line:90% 00:25:58.240 --> 00:26:00.730 align:middle line:84% If e is a lot lower, the only way 00:26:00.730 --> 00:26:05.920 align:middle line:84% I'm going to keep firm 2 out of the market is to reduce, 00:26:05.920 --> 00:26:09.100 align:middle line:84% is to increase k1 all the way up to this level. 00:26:09.100 --> 00:26:11.560 align:middle line:84% And then at this level, as I've drawn it, 00:26:11.560 --> 00:26:13.935 align:middle line:90% my profits are essentially 0. 00:26:13.935 --> 00:26:16.650 align:middle line:90% 00:26:16.650 --> 00:26:19.810 align:middle line:84% And so I'm going to have-- this was my monopoly profits. 00:26:19.810 --> 00:26:22.625 align:middle line:84% I'm going to also have, say, like duopoly profits curve. 00:26:22.625 --> 00:26:27.750 align:middle line:90% 00:26:27.750 --> 00:26:30.270 align:middle line:84% And if the amount by which I have 00:26:30.270 --> 00:26:37.800 align:middle line:84% to raise my investment in order to keep firm 2 out of the market 00:26:37.800 --> 00:26:41.610 align:middle line:84% is so large that it's going to make my-- 00:26:41.610 --> 00:26:45.622 align:middle line:84% I should probably have made this one lower everywhere. 00:26:45.622 --> 00:26:47.580 align:middle line:84% Duopoly profits are generally going to be lower 00:26:47.580 --> 00:26:48.770 align:middle line:90% than monopoly profits. 00:26:48.770 --> 00:26:52.140 align:middle line:90% So it might look like this. 00:26:52.140 --> 00:26:54.840 align:middle line:84% It may be that rather than go all the way out here and drive 00:26:54.840 --> 00:26:58.440 align:middle line:84% my monopoly profits to 0, I'm better off just 00:26:58.440 --> 00:27:01.260 align:middle line:84% going somewhere, doing a more moderate investment 00:27:01.260 --> 00:27:04.170 align:middle line:84% and admitting that firm 2 is going to be in the market 00:27:04.170 --> 00:27:06.400 align:middle line:90% and accommodate entry instead. 00:27:06.400 --> 00:27:11.130 align:middle line:84% So if the profits from deterring entry are sufficiently low, 00:27:11.130 --> 00:27:14.280 align:middle line:84% instead I may just go to an accommodation strategy 00:27:14.280 --> 00:27:17.970 align:middle line:84% where I instead choose k not to deter entry, 00:27:17.970 --> 00:27:19.980 align:middle line:84% but to do what's best possible for given 00:27:19.980 --> 00:27:21.455 align:middle line:90% that firm 2 is going to enter. 00:27:21.455 --> 00:27:24.740 align:middle line:90% 00:27:24.740 --> 00:27:28.500 align:middle line:84% So let me discuss the entry accommodation solution. 00:27:28.500 --> 00:27:30.770 align:middle line:90% I'm going to call that k1AC. 00:27:30.770 --> 00:27:36.410 align:middle line:84% That's going to be the argmax of overall k1 of pi 1 star of k1, 00:27:36.410 --> 00:27:38.980 align:middle line:84% where pi 1 star is the duopoly profit function. 00:27:38.980 --> 00:27:41.950 align:middle line:90% 00:27:41.950 --> 00:27:45.250 align:middle line:84% So in this example, let's suppose 00:27:45.250 --> 00:27:48.690 align:middle line:84% it looks like this, it's maximized here. 00:27:48.690 --> 00:27:51.710 align:middle line:90% I could have k1 accommodation. 00:27:51.710 --> 00:27:54.560 align:middle line:90% 00:27:54.560 --> 00:27:59.420 align:middle line:84% I'm instead maximizing the duopoly profits 00:27:59.420 --> 00:28:02.515 align:middle line:84% that I'm going to get if I choose investment k1. 00:28:02.515 --> 00:28:08.470 align:middle line:90% 00:28:08.470 --> 00:28:11.800 align:middle line:84% And so what Fudenberg will do is give you 00:28:11.800 --> 00:28:14.950 align:middle line:84% a little bit of thinking about, you know, when are you 00:28:14.950 --> 00:28:18.070 align:middle line:84% going to over-invest or under-invest 00:28:18.070 --> 00:28:20.120 align:middle line:90% in order to accommodate entry? 00:28:20.120 --> 00:28:21.800 align:middle line:84% How do you think about that decision? 00:28:21.800 --> 00:28:24.080 align:middle line:84% I said the deterrence one was simple. 00:28:24.080 --> 00:28:26.270 align:middle line:84% You just think about I want to hurt the other firm. 00:28:26.270 --> 00:28:27.860 align:middle line:90% If it's tough, I over-invest. 00:28:27.860 --> 00:28:29.720 align:middle line:90% If it's soft, I under-invest. 00:28:29.720 --> 00:28:32.850 align:middle line:90% How do we think about this? 00:28:32.850 --> 00:28:35.030 align:middle line:90% Well, so I need a benchmark. 00:28:35.030 --> 00:28:39.670 align:middle line:84% So as a point of comparison, they define k1OL-- 00:28:39.670 --> 00:28:42.520 align:middle line:90% OL stands for "open loop." 00:28:42.520 --> 00:28:48.190 align:middle line:84% That's a terminology that's disappeared from economics 00:28:48.190 --> 00:28:49.780 align:middle line:90% since they wrote the paper. 00:28:49.780 --> 00:28:53.830 align:middle line:84% So we define k1 open loop to be the level of investment 00:28:53.830 --> 00:28:57.700 align:middle line:84% that firm 1 would choose if entry was going to occur 00:28:57.700 --> 00:29:01.340 align:middle line:90% and firm 2 could not observe k1. 00:29:01.340 --> 00:29:07.340 align:middle line:84% So if firm 2 can't observe k1, then in equilibrium it 00:29:07.340 --> 00:29:09.440 align:middle line:84% must be the case that the partial of pi 1 00:29:09.440 --> 00:29:13.160 align:middle line:84% with respect to k1 evaluated at k1 open loop, 00:29:13.160 --> 00:29:17.810 align:middle line:84% and then the equilibrium actions that follow, is equal to 0. 00:29:17.810 --> 00:29:20.240 align:middle line:84% OK, so think about the equilibrium-- think 00:29:20.240 --> 00:29:23.240 align:middle line:84% about the game where firm 1 chooses k1, 00:29:23.240 --> 00:29:25.580 align:middle line:84% and then firm 2 doesn't observe k1 00:29:25.580 --> 00:29:27.440 align:middle line:84% before it makes its entry decision. 00:29:27.440 --> 00:29:29.750 align:middle line:84% In equilibrium, it correctly anticipates 00:29:29.750 --> 00:29:34.280 align:middle line:84% what the investment was, and then the actions by both firms 00:29:34.280 --> 00:29:37.760 align:middle line:90% reflect what the investment was. 00:29:37.760 --> 00:29:42.390 align:middle line:84% But so we get it's this partial is equal to 0 evaluated 00:29:42.390 --> 00:29:46.590 align:middle line:84% at k1 open loop, because it's almost as if firm 1 could 00:29:46.590 --> 00:29:50.010 align:middle line:84% go back and change, that k1 is not sunk, 00:29:50.010 --> 00:29:52.830 align:middle line:84% and firm 1 could go back and change k1 00:29:52.830 --> 00:29:55.998 align:middle line:84% after firm 2 has taken its action then, 00:29:55.998 --> 00:29:57.540 align:middle line:84% because if firm 2 doesn't observe it, 00:29:57.540 --> 00:29:59.500 align:middle line:84% it's as if you're choosing it afterwards. 00:29:59.500 --> 00:30:02.490 align:middle line:84% It's the same first order condition. 00:30:02.490 --> 00:30:04.070 align:middle line:90% OK? 00:30:04.070 --> 00:30:05.400 align:middle line:90% Is that definition OK? 00:30:05.400 --> 00:30:12.442 align:middle line:84% So anyway, we're going to think about strategic investment 00:30:12.442 --> 00:30:14.650 align:middle line:84% as being what do you do different from what you would 00:30:14.650 --> 00:30:16.042 align:middle line:90% do if k1 was unobservable. 00:30:16.042 --> 00:30:17.500 align:middle line:84% Because if k1 is unobservable, it's 00:30:17.500 --> 00:30:20.860 align:middle line:84% not affecting the other firm, the strategic use of it 00:30:20.860 --> 00:30:24.310 align:middle line:84% is doing something different because firm 2 sees it. 00:30:24.310 --> 00:30:27.730 align:middle line:84% And so first characterization theorem 00:30:27.730 --> 00:30:30.010 align:middle line:84% just highlights the strategic effect. 00:30:30.010 --> 00:30:32.990 align:middle line:84% Firms over-invest to accommodate entry, 00:30:32.990 --> 00:30:37.300 align:middle line:84% meaning they choose k1 bigger than the open loop value 00:30:37.300 --> 00:30:44.620 align:middle line:84% if and only if d pi 1 / da2, da2 star / dk1 is greater than 0. 00:30:44.620 --> 00:30:47.090 align:middle line:84% OK, and the proof of this is straightforward. 00:30:47.090 --> 00:30:49.660 align:middle line:84% I again, just write out here what is-- now 00:30:49.660 --> 00:30:54.130 align:middle line:84% I'm going to write out what is pi 1 star. 00:30:54.130 --> 00:30:55.420 align:middle line:90% It's that. 00:30:55.420 --> 00:30:58.810 align:middle line:84% And then think about what is the derivative of pi 1 00:30:58.810 --> 00:31:04.005 align:middle line:84% star with respect to k1 when I evaluate it at k1 open loop? 00:31:04.005 --> 00:31:06.520 align:middle line:90% 00:31:06.520 --> 00:31:11.560 align:middle line:84% First you get a partial, partial pi 1 with respect to-- 00:31:11.560 --> 00:31:15.140 align:middle line:84% partial pi 1 with respect to k1 evaluated at k1 open loop. 00:31:15.140 --> 00:31:19.000 align:middle line:84% That's 0 by the definition of k1 open loop. 00:31:19.000 --> 00:31:23.530 align:middle line:84% Second thing you get is d pi 1 / da1, da1 / star dk1 again 00:31:23.530 --> 00:31:27.410 align:middle line:84% with this d pi 1 / da1 evaluated at a1 star of k1. 00:31:27.410 --> 00:31:30.910 align:middle line:84% This is 0 by the envelope theorem again. 00:31:30.910 --> 00:31:37.660 align:middle line:84% So d pi 1 star / dk1 is positive if this strategic effect 00:31:37.660 --> 00:31:40.570 align:middle line:84% is positive, and it's negative if the strategic effect is 00:31:40.570 --> 00:31:41.420 align:middle line:90% negative. 00:31:41.420 --> 00:31:45.610 align:middle line:84% So the strategic effect is how does the investment affect a2, 00:31:45.610 --> 00:31:50.230 align:middle line:84% and how is your profit affected by a2. 00:31:50.230 --> 00:31:53.200 align:middle line:84% Does k1 change the other firm's action 00:31:53.200 --> 00:31:55.160 align:middle line:84% in a way that's good or bad for you? 00:31:55.160 --> 00:31:56.960 align:middle line:90% That's the strategic effect. 00:31:56.960 --> 00:32:00.670 align:middle line:84% And so notice that if I evaluate that derivative, 00:32:00.670 --> 00:32:05.740 align:middle line:84% if the total derivative of pi 1 star with respect to k1 00:32:05.740 --> 00:32:10.120 align:middle line:84% evaluated at k1 open loop is greater than 0 and profit is 00:32:10.120 --> 00:32:16.960 align:middle line:84% single peaked, then that's going to imply that k1AC is strictly 00:32:16.960 --> 00:32:18.940 align:middle line:90% greater than k1 open loop. 00:32:18.940 --> 00:32:23.560 align:middle line:84% It's just a case where we have a profit function we 00:32:23.560 --> 00:32:25.760 align:middle line:90% know evaluated at k1 open loop. 00:32:25.760 --> 00:32:29.247 align:middle line:84% This pi 1 star is increasing, so the maximum 00:32:29.247 --> 00:32:30.330 align:middle line:90% is somewhere to the right. 00:32:30.330 --> 00:32:34.660 align:middle line:90% 00:32:34.660 --> 00:32:37.060 align:middle line:84% So one way to think about strategic investment 00:32:37.060 --> 00:32:39.590 align:middle line:84% is just to do these calculations for yourself. 00:32:39.590 --> 00:32:45.898 align:middle line:84% How would this investment affect firm 2's action? 00:32:45.898 --> 00:32:47.690 align:middle line:84% And then what would I think of that action? 00:32:47.690 --> 00:32:53.440 align:middle line:84% So for instance, cost reduction, I'm in a price competition game, 00:32:53.440 --> 00:32:57.190 align:middle line:84% and I'm thinking of investing in cost reduction. 00:32:57.190 --> 00:33:00.030 align:middle line:84% If I reduce my costs in a price competition game, 00:33:00.030 --> 00:33:02.530 align:middle line:84% that's going to lead the other firm to lower its equilibrium 00:33:02.530 --> 00:33:05.180 align:middle line:84% price because he knows I'm going to lower my price. 00:33:05.180 --> 00:33:08.860 align:middle line:84% And so da2 / dk1 becomes negative, more investment, 00:33:08.860 --> 00:33:10.330 align:middle line:90% lower rival price. 00:33:10.330 --> 00:33:13.220 align:middle line:84% And then my profit is increasing in the rival's price. 00:33:13.220 --> 00:33:15.980 align:middle line:84% So this is positive, so positive times negative. 00:33:15.980 --> 00:33:18.220 align:middle line:84% This is a negative strategic effect. 00:33:18.220 --> 00:33:22.270 align:middle line:84% So in the strategic entry accommodation game, 00:33:22.270 --> 00:33:25.780 align:middle line:84% you're going to underinvest in cost reduction, 00:33:25.780 --> 00:33:27.430 align:middle line:84% meaning you're going to do less of it 00:33:27.430 --> 00:33:31.055 align:middle line:84% than you would do if you could do secret cost reduction. 00:33:31.055 --> 00:33:32.680 align:middle line:84% So if I could do secret cost reduction, 00:33:32.680 --> 00:33:34.700 align:middle line:84% I would do more because it really 00:33:34.700 --> 00:33:38.070 align:middle line:84% does benefit me profit wise to have a lower cost. 00:33:38.070 --> 00:33:40.730 align:middle line:84% But I know that if my rival sees me investing in the cost 00:33:40.730 --> 00:33:42.880 align:middle line:84% reduction, he's just going to lower his price, 00:33:42.880 --> 00:33:44.660 align:middle line:84% so I'm going to eat up a lot of the benefits I was getting 00:33:44.660 --> 00:33:46.020 align:middle line:90% from the cost reduction. 00:33:46.020 --> 00:33:48.740 align:middle line:84% So what we'll see firms is underinvesting 00:33:48.740 --> 00:33:49.875 align:middle line:90% in cost reduction. 00:33:49.875 --> 00:33:52.250 align:middle line:84% You could do a calculation and say, why doesn't that firm 00:33:52.250 --> 00:33:53.490 align:middle line:90% adopt this technology? 00:33:53.490 --> 00:33:56.880 align:middle line:84% That technology would clearly pay for itself. 00:33:56.880 --> 00:33:59.330 align:middle line:84% And the answer is they don't do it for strategic reasons 00:33:59.330 --> 00:34:01.910 align:middle line:84% because if their rival knew they had that lower 00:34:01.910 --> 00:34:04.130 align:middle line:84% cost manufacturing technology, the rival would just 00:34:04.130 --> 00:34:05.040 align:middle line:90% charge a lower price. 00:34:05.040 --> 00:34:05.873 align:middle line:90% They'd be worse off. 00:34:05.873 --> 00:34:10.042 align:middle line:90% 00:34:10.042 --> 00:34:11.600 align:middle line:90% Is that example pretty clear? 00:34:11.600 --> 00:34:15.219 align:middle line:90% 00:34:15.219 --> 00:34:16.545 align:middle line:90% OK. 00:34:16.545 --> 00:34:18.670 align:middle line:84% Fudenberg and Tirole give the second theorem noting 00:34:18.670 --> 00:34:23.590 align:middle line:84% that one way to think about strategic entry accommodation 00:34:23.590 --> 00:34:28.060 align:middle line:84% is in terms of, what's the effect of the investment 00:34:28.060 --> 00:34:28.850 align:middle line:90% on your rival? 00:34:28.850 --> 00:34:30.110 align:middle line:90% What is the rival? 00:34:30.110 --> 00:34:31.489 align:middle line:90% What does that do to you? 00:34:31.489 --> 00:34:35.440 align:middle line:84% They note that it also can be related to the type of game 00:34:35.440 --> 00:34:36.620 align:middle line:90% and the type of investment. 00:34:36.620 --> 00:34:38.030 align:middle line:90% So here's a theorem. 00:34:38.030 --> 00:34:40.270 align:middle line:84% Let's, for instance, initially consider 00:34:40.270 --> 00:34:43.300 align:middle line:84% investments where there's no direct effect on firm 00:34:43.300 --> 00:34:46.060 align:middle line:90% 2 like a cost reduction. 00:34:46.060 --> 00:34:51.639 align:middle line:84% Let's assume that the game is at least somewhat symmetric, weakly 00:34:51.639 --> 00:34:52.610 align:middle line:90% symmetric, I'd call it. 00:34:52.610 --> 00:34:55.960 align:middle line:84% So profit-- changing the sign of d pi i / daj 00:34:55.960 --> 00:34:58.430 align:middle line:90% is the same as d pi j / dai. 00:34:58.430 --> 00:35:01.025 align:middle line:84% So any game like price competition or quantity 00:35:01.025 --> 00:35:02.650 align:middle line:84% competition or whatever will have this. 00:35:02.650 --> 00:35:04.990 align:middle line:84% Just the way that firm 1's action 00:35:04.990 --> 00:35:06.610 align:middle line:84% affects firm 2 is the same as the way 00:35:06.610 --> 00:35:09.100 align:middle line:90% firm 2's action affects firm 1. 00:35:09.100 --> 00:35:11.110 align:middle line:84% If you had some bizarre game where 00:35:11.110 --> 00:35:14.720 align:middle line:84% firm 1 was choosing a price and firm 2 was choosing a quantity, 00:35:14.720 --> 00:35:16.190 align:middle line:90% then that might not be true. 00:35:16.190 --> 00:35:20.510 align:middle line:84% But most of the games we do are fully symmetric. 00:35:20.510 --> 00:35:25.040 align:middle line:84% And so the theorem is that firms overinvest to accommodate entry 00:35:25.040 --> 00:35:28.340 align:middle line:84% if the game has in strategic complements and investment 00:35:28.340 --> 00:35:32.480 align:middle line:84% makes them soft, or if the game has 00:35:32.480 --> 00:35:36.500 align:middle line:84% strategic substitutes and investment makes firm 1 tough. 00:35:36.500 --> 00:35:43.130 align:middle line:84% So let me start with the strategic substitutes case. 00:35:43.130 --> 00:35:48.530 align:middle line:84% In the strategic substitutes case, what this says is-- 00:35:48.530 --> 00:35:52.100 align:middle line:84% so in strategic substitutes-- so I did put it over here. 00:35:52.100 --> 00:35:58.430 align:middle line:84% So in the substitutes case, if the investment makes you tough, 00:35:58.430 --> 00:35:59.496 align:middle line:90% you overinvest. 00:35:59.496 --> 00:36:02.391 align:middle line:90% 00:36:02.391 --> 00:36:05.016 align:middle line:84% And if you have something that makes you soft, you underinvest. 00:36:05.016 --> 00:36:07.560 align:middle line:90% 00:36:07.560 --> 00:36:11.610 align:middle line:84% This is the same thing we had in-- this 00:36:11.610 --> 00:36:15.825 align:middle line:84% is the same thing we had in the entry deterrence case. 00:36:15.825 --> 00:36:17.880 align:middle line:84% When you're in deter entry, you're always 00:36:17.880 --> 00:36:19.030 align:middle line:90% trying to hurt your rivals. 00:36:19.030 --> 00:36:21.130 align:middle line:84% So if the investment makes you tough, you do it. 00:36:21.130 --> 00:36:23.110 align:middle line:84% If it makes you soft, you don't do it. 00:36:23.110 --> 00:36:26.820 align:middle line:84% What this says in games with strategic substitutes, 00:36:26.820 --> 00:36:30.510 align:middle line:84% accommodation of entry is just like deterrence of entry. 00:36:30.510 --> 00:36:32.230 align:middle line:84% In a game with strategic substitutes, 00:36:32.230 --> 00:36:36.440 align:middle line:84% what you always want to do is just hurt your rival. 00:36:36.440 --> 00:36:42.950 align:middle line:84% But things turn around in games with strategic complements. 00:36:42.950 --> 00:36:46.310 align:middle line:84% In games with strategic complements, 00:36:46.310 --> 00:36:49.220 align:middle line:90% you want to help your rival? 00:36:49.220 --> 00:36:51.470 align:middle line:84% So actually, when I was describing the cost reduction, 00:36:51.470 --> 00:36:54.580 align:middle line:84% I did the cost reduction here, I said in the cost reduction game, 00:36:54.580 --> 00:36:56.410 align:middle line:84% to deter entry, you don't want to do it, 00:36:56.410 --> 00:36:58.210 align:middle line:84% but to accommodate entry, you don't 00:36:58.210 --> 00:37:02.380 align:middle line:84% invest because it's just going to make your rival lower prices. 00:37:02.380 --> 00:37:04.930 align:middle line:84% But the interesting observation here 00:37:04.930 --> 00:37:09.473 align:middle line:84% is that sometimes I describe Cournot versus Bertrand 00:37:09.473 --> 00:37:11.390 align:middle line:84% as just two different assumptions people make. 00:37:11.390 --> 00:37:12.410 align:middle line:90% They're similar models. 00:37:12.410 --> 00:37:14.020 align:middle line:90% Profits decline. 00:37:14.020 --> 00:37:16.362 align:middle line:84% This notes that a lot of the conclusions we're 00:37:16.362 --> 00:37:18.070 align:middle line:84% going to make about strategic investments 00:37:18.070 --> 00:37:21.820 align:middle line:84% would just completely flip if we had a quantity competition 00:37:21.820 --> 00:37:24.730 align:middle line:84% game versus if we had a price competition game. 00:37:24.730 --> 00:37:26.890 align:middle line:84% So in the quantity competition game-- 00:37:26.890 --> 00:37:31.000 align:middle line:84% let me go back to my previous first order condition. 00:37:31.000 --> 00:37:34.930 align:middle line:84% What's the strategic effect of reducing my cost? 00:37:34.930 --> 00:37:39.080 align:middle line:84% In a quantity competition game, if I reduce my cost, 00:37:39.080 --> 00:37:42.130 align:middle line:84% it's a commitment that I'm going to produce a lot. 00:37:42.130 --> 00:37:44.470 align:middle line:84% And in Cournot with those downward-sloping reaction 00:37:44.470 --> 00:37:46.180 align:middle line:84% curves, if I'm going to produce a lot, 00:37:46.180 --> 00:37:48.310 align:middle line:84% my rival is going to produce less. 00:37:48.310 --> 00:37:52.160 align:middle line:84% And so by investing in cost reduction, 00:37:52.160 --> 00:37:54.380 align:middle line:90% I make his quantity lower. 00:37:54.380 --> 00:37:56.430 align:middle line:90% So this is negative. 00:37:56.430 --> 00:38:01.120 align:middle line:84% And then if his quantity is higher, that's bad for me. 00:38:01.120 --> 00:38:02.200 align:middle line:90% So this is also negative. 00:38:02.200 --> 00:38:03.575 align:middle line:84% So I have negative times negative 00:38:03.575 --> 00:38:07.890 align:middle line:84% is a positive number, which means that I want to overinvest 00:38:07.890 --> 00:38:09.540 align:middle line:90% in cost reduction. 00:38:09.540 --> 00:38:13.200 align:middle line:84% So if you look at firms that would be competing in a Cournot 00:38:13.200 --> 00:38:15.630 align:middle line:84% market, the conclusion you'd get is 00:38:15.630 --> 00:38:18.060 align:middle line:84% that if they were considering cost reduction-- if you look 00:38:18.060 --> 00:38:20.640 align:middle line:84% at price competing firms, we're going 00:38:20.640 --> 00:38:23.220 align:middle line:84% to expect that there will be these cost-reducing investments 00:38:23.220 --> 00:38:25.345 align:middle line:84% that we don't understand why firm 1 isn't taking it 00:38:25.345 --> 00:38:27.000 align:middle line:84% because it would be privately optimal, 00:38:27.000 --> 00:38:28.267 align:middle line:90% holding everything else fixed. 00:38:28.267 --> 00:38:29.850 align:middle line:84% In a quantity competition game, you're 00:38:29.850 --> 00:38:32.790 align:middle line:84% going to see firms having overinvested in cost reduction. 00:38:32.790 --> 00:38:36.360 align:middle line:84% And you're going to calculate that that cost reduction 00:38:36.360 --> 00:38:38.350 align:middle line:84% technology is not paying for itself. 00:38:38.350 --> 00:38:42.120 align:middle line:84% It's making less money with the cost-reducing technology 00:38:42.120 --> 00:38:43.270 align:middle line:90% than the amount. 00:38:43.270 --> 00:38:45.130 align:middle line:84% It's actually saving on production. 00:38:45.130 --> 00:38:48.060 align:middle line:84% But it does it because of the strategic effect. 00:38:48.060 --> 00:38:51.150 align:middle line:84% That is, in a Cournot game, committing 00:38:51.150 --> 00:38:54.340 align:middle line:84% to be tough forces your rival to stay at a smaller scale. 00:38:54.340 --> 00:38:55.340 align:middle line:90% And that's good for you. 00:38:55.340 --> 00:39:02.720 align:middle line:90% 00:39:02.720 --> 00:39:05.840 align:middle line:84% So anyway, I think the thing is in games 00:39:05.840 --> 00:39:09.500 align:middle line:84% with strategic substitutes, what you're always trying to do 00:39:09.500 --> 00:39:11.920 align:middle line:84% is take investments that hurt your rival. 00:39:11.920 --> 00:39:14.750 align:middle line:90% 00:39:14.750 --> 00:39:18.440 align:middle line:84% And what you're doing in a game with strategic complements 00:39:18.440 --> 00:39:20.840 align:middle line:84% is you're trying to help your rival. 00:39:20.840 --> 00:39:23.795 align:middle line:84% So anything you can do that's good for your rival 00:39:23.795 --> 00:39:26.630 align:middle line:84% is good for you as well or at least 00:39:26.630 --> 00:39:27.850 align:middle line:90% that's the strategic effect. 00:39:27.850 --> 00:39:34.805 align:middle line:90% 00:39:34.805 --> 00:39:36.930 align:middle line:84% And I'm just phrasing this in the discussion of how 00:39:36.930 --> 00:39:41.310 align:middle line:84% does k1AC compare to k1OL, which is the amount of investment you 00:39:41.310 --> 00:39:46.570 align:middle line:84% would be doing in the oligopoly if it was unobservable. 00:39:46.570 --> 00:39:49.220 align:middle line:90% 00:39:49.220 --> 00:39:50.640 align:middle line:90% Put a proof on the next slide. 00:39:50.640 --> 00:39:53.120 align:middle line:84% I think I won't take the time to go through this 00:39:53.120 --> 00:39:56.430 align:middle line:84% because sometimes I find it doesn't-- 00:39:56.430 --> 00:39:57.790 align:middle line:90% it's a simple calculation. 00:39:57.790 --> 00:39:59.623 align:middle line:84% I don't know that it's all that informative. 00:39:59.623 --> 00:40:01.265 align:middle line:84% But basically, the argument is just-- 00:40:01.265 --> 00:40:03.940 align:middle line:90% 00:40:03.940 --> 00:40:07.380 align:middle line:90% You know, think about how-- 00:40:07.380 --> 00:40:09.435 align:middle line:90% think about how a1 star of k1-- 00:40:09.435 --> 00:40:15.245 align:middle line:90% 00:40:15.245 --> 00:40:17.300 align:middle line:90% what I'm trying to sign is this. 00:40:17.300 --> 00:40:19.830 align:middle line:84% I know this is the strategic effect. 00:40:19.830 --> 00:40:24.170 align:middle line:84% I'm going to overinvest if k1 moves a2 in a direction 00:40:24.170 --> 00:40:25.250 align:middle line:90% that I like. 00:40:25.250 --> 00:40:27.710 align:middle line:84% And I'm going to underinvest if k1 moves a2 00:40:27.710 --> 00:40:29.430 align:middle line:90% in a direction I don't like. 00:40:29.430 --> 00:40:32.000 align:middle line:84% So this was the strategic effect. 00:40:32.000 --> 00:40:35.300 align:middle line:84% And I'm just trying to think about how do I 00:40:35.300 --> 00:40:36.950 align:middle line:90% sign this strategic effect? 00:40:36.950 --> 00:40:41.510 align:middle line:84% In particular, how do I find out what da2 star / dk1 is? 00:40:41.510 --> 00:40:44.840 align:middle line:84% Is k1 going to move a2 in a direction I like 00:40:44.840 --> 00:40:48.120 align:middle line:84% or in the direction I don't like? 00:40:48.120 --> 00:40:49.740 align:middle line:90% And how do I do that? 00:40:49.740 --> 00:40:54.280 align:middle line:84% I note that a2 star satisfies this for every k1. 00:40:54.280 --> 00:40:56.760 align:middle line:84% This is the first order condition 00:40:56.760 --> 00:40:59.940 align:middle line:84% defining firm 2's best response function 00:40:59.940 --> 00:41:02.110 align:middle line:90% or defining equilibrium. 00:41:02.110 --> 00:41:06.030 align:middle line:84% That partial of pi 2 with respect to a2 evaluated 00:41:06.030 --> 00:41:08.710 align:middle line:90% at a2 star of k1 is equal to 0. 00:41:08.710 --> 00:41:11.910 align:middle line:84% So what I just do is I take this thing. 00:41:11.910 --> 00:41:15.060 align:middle line:84% I differentiate it with respect to k1. 00:41:15.060 --> 00:41:17.760 align:middle line:90% It gives me these three terms. 00:41:17.760 --> 00:41:22.770 align:middle line:84% And then when I look at this expression d pi 00:41:22.770 --> 00:41:26.080 align:middle line:84% 2 / dk1 was 0 with no direct effect. 00:41:26.080 --> 00:41:29.750 align:middle line:90% So this term drops out. 00:41:29.750 --> 00:41:33.390 align:middle line:84% d squared pi 2 da1 / da2, this is the thing 00:41:33.390 --> 00:41:36.440 align:middle line:84% that's whether you have strategic complements 00:41:36.440 --> 00:41:39.710 align:middle line:90% or strategic substitutes. 00:41:39.710 --> 00:41:42.030 align:middle line:90% This term here is negative. 00:41:42.030 --> 00:41:44.210 align:middle line:90% And so anyway, you-- 00:41:44.210 --> 00:41:46.740 align:middle line:90% I'm trying to sign da2 star dk1. 00:41:46.740 --> 00:41:47.990 align:middle line:90% I move this to the other side. 00:41:47.990 --> 00:41:51.270 align:middle line:84% And the sign of da2 star / dk1 is the sign of this. 00:41:51.270 --> 00:41:53.720 align:middle line:84% And that depends on complements or substitutes 00:41:53.720 --> 00:41:58.390 align:middle line:90% and the soft or tough. 00:41:58.390 --> 00:42:05.110 align:middle line:90% 00:42:05.110 --> 00:42:07.450 align:middle line:84% Anyway, you can look at the calculations afterwards. 00:42:07.450 --> 00:42:10.540 align:middle line:90% But in some sense, that's-- 00:42:10.540 --> 00:42:12.655 align:middle line:84% the reason why is you're just thinking about what 00:42:12.655 --> 00:42:14.290 align:middle line:90% it is that's going to-- 00:42:14.290 --> 00:42:16.540 align:middle line:84% are you going to move the other firm's action since it 00:42:16.540 --> 00:42:17.930 align:middle line:90% hurts you or helps you? 00:42:17.930 --> 00:42:19.420 align:middle line:84% And that just turns out to depend 00:42:19.420 --> 00:42:22.600 align:middle line:90% whether if it's complements-- 00:42:22.600 --> 00:42:26.120 align:middle line:84% like it's helping-- firm 2 is helping you and vice versa. 00:42:26.120 --> 00:42:28.960 align:middle line:84% And it substitutes-- helping firm 2 is hurting you 00:42:28.960 --> 00:42:29.700 align:middle line:90% and vice versa. 00:42:29.700 --> 00:42:33.080 align:middle line:90% 00:42:33.080 --> 00:42:37.070 align:middle line:84% But I think the main thing to take away 00:42:37.070 --> 00:42:39.650 align:middle line:84% is this intuition for what you want 00:42:39.650 --> 00:42:42.740 align:middle line:84% to do in entry accommodation games is that in Cournot games, 00:42:42.740 --> 00:42:44.990 align:middle line:84% you're always trying to still hurt your rival like you 00:42:44.990 --> 00:42:46.610 align:middle line:90% do to deter entry. 00:42:46.610 --> 00:42:49.250 align:middle line:84% But in games with strategic complements, things turn around, 00:42:49.250 --> 00:42:50.625 align:middle line:84% and you should always be thinking 00:42:50.625 --> 00:42:52.940 align:middle line:84% about I want to aid my rival and distort my investment 00:42:52.940 --> 00:42:54.400 align:middle line:90% in that direction. 00:42:54.400 --> 00:42:55.850 align:middle line:84% And I just want to do that quickly 00:42:55.850 --> 00:42:57.380 align:middle line:84% because I'll just go through now. 00:42:57.380 --> 00:43:00.920 align:middle line:84% I've got like five different examples of classic arguments 00:43:00.920 --> 00:43:06.060 align:middle line:84% to think about, what do firms do for strategic reasons? 00:43:06.060 --> 00:43:09.300 align:middle line:84% So the first is a model of learning by doing. 00:43:09.300 --> 00:43:14.370 align:middle line:84% So we start with a firm that's a monopolist. 00:43:14.370 --> 00:43:18.050 align:middle line:90% And it has some costs c0. 00:43:18.050 --> 00:43:22.220 align:middle line:84% And it sells some quantity in the first period, q11, 00:43:22.220 --> 00:43:26.390 align:middle line:84% where I guess the upper one refers to the-- 00:43:26.390 --> 00:43:27.170 align:middle line:90% I don't know. 00:43:27.170 --> 00:43:28.860 align:middle line:84% I guess it refers to the first period. 00:43:28.860 --> 00:43:31.940 align:middle line:84% So it's firm 1 in the first period. 00:43:31.940 --> 00:43:35.690 align:middle line:84% After firm 1 produced in the first period, 00:43:35.690 --> 00:43:39.080 align:middle line:84% firm 2 observes firm 1's production level 00:43:39.080 --> 00:43:42.793 align:middle line:84% and then chooses whether to enter at a cost of E. 00:43:42.793 --> 00:43:44.960 align:middle line:84% The critical assumption of a learning by doing model 00:43:44.960 --> 00:43:49.850 align:middle line:84% is that the firm 1's marginal cost in the second period 00:43:49.850 --> 00:43:53.540 align:middle line:84% is going to depend on how much it produced in the first period. 00:43:53.540 --> 00:43:58.200 align:middle line:84% So c of q11 is the period 2 marginal cost. 00:43:58.200 --> 00:44:00.650 align:middle line:84% And so we're assuming that the more you produce 00:44:00.650 --> 00:44:03.380 align:middle line:84% in the first period, the lower is your marginal cost 00:44:03.380 --> 00:44:05.210 align:middle line:90% in the second period. 00:44:05.210 --> 00:44:08.810 align:middle line:84% So just by producing more, you get more experience producing. 00:44:08.810 --> 00:44:10.400 align:middle line:84% Every unit you produce, you figure out 00:44:10.400 --> 00:44:12.600 align:middle line:84% things that improve your production process. 00:44:12.600 --> 00:44:16.410 align:middle line:90% You make your costs lower. 00:44:16.410 --> 00:44:21.270 align:middle line:84% So in this model, we can think of overproducing. 00:44:21.270 --> 00:44:24.780 align:middle line:90% Overproducing is just choosing-- 00:44:24.780 --> 00:44:28.020 align:middle line:84% I would normally choose q11 if I was-- if there was-- 00:44:28.020 --> 00:44:32.760 align:middle line:84% in the blockaded entry case, I would choose q11 to be 00:44:32.760 --> 00:44:45.330 align:middle line:84% the argmax over q11 of pi 1 of q11, 00:44:45.330 --> 00:44:48.340 align:middle line:90% pi 11 plus discount factor-- 00:44:48.340 --> 00:44:50.070 align:middle line:84% I don't think I do discount factors-- 00:44:50.070 --> 00:44:57.540 align:middle line:90% a pi 12 monopoly of q11 q12. 00:44:57.540 --> 00:45:00.570 align:middle line:84% So my first period profit just depends on the quantity 00:45:00.570 --> 00:45:02.280 align:middle line:90% I produce in the first period. 00:45:02.280 --> 00:45:04.680 align:middle line:84% And in the second period, if I'm still a monopolist, 00:45:04.680 --> 00:45:06.480 align:middle line:84% my profit is going to depend both on what 00:45:06.480 --> 00:45:08.867 align:middle line:84% I produced in the first period and what 00:45:08.867 --> 00:45:10.950 align:middle line:84% I produced in the second period with the first one 00:45:10.950 --> 00:45:13.750 align:middle line:84% here because what I produce in the first period 00:45:13.750 --> 00:45:16.630 align:middle line:84% affects my cost in the second period. 00:45:16.630 --> 00:45:20.830 align:middle line:90% And so this is what I would do. 00:45:20.830 --> 00:45:23.060 align:middle line:84% I think I called this q1 monopoly. 00:45:23.060 --> 00:45:27.070 align:middle line:84% This is what you would do in the blockaded entry case is just 00:45:27.070 --> 00:45:29.780 align:middle line:90% produce this level of output. 00:45:29.780 --> 00:45:34.840 align:middle line:84% And so you can think of if I choose q11 greater than q1m, 00:45:34.840 --> 00:45:41.020 align:middle line:84% that's an investment in reducing costs. 00:45:41.020 --> 00:45:45.470 align:middle line:90% 00:45:45.470 --> 00:45:47.080 align:middle line:84% It's an investment in reducing costs 00:45:47.080 --> 00:45:49.060 align:middle line:84% that I make by just overproducing 00:45:49.060 --> 00:45:50.630 align:middle line:90% in the first period. 00:45:50.630 --> 00:45:52.580 align:middle line:84% What's the cost of the investment? 00:45:52.580 --> 00:46:03.490 align:middle line:84% The cost of the investment is pi 11 of q1m minus pi 11 of q11. 00:46:03.490 --> 00:46:05.620 align:middle line:90% If I produce this-- 00:46:05.620 --> 00:46:08.620 align:middle line:84% if I produce some extra amount, it's 00:46:08.620 --> 00:46:10.640 align:middle line:84% going to lower my first period profits. 00:46:10.640 --> 00:46:13.240 align:middle line:90% The amount by which it lower-- 00:46:13.240 --> 00:46:15.600 align:middle line:84% so this is in some sense the investment cost. 00:46:15.600 --> 00:46:21.190 align:middle line:90% 00:46:21.190 --> 00:46:23.380 align:middle line:84% So producing much in the first period 00:46:23.380 --> 00:46:25.420 align:middle line:84% is a strategic investment, an investment 00:46:25.420 --> 00:46:27.820 align:middle line:84% that I spend something and lost profits. 00:46:27.820 --> 00:46:30.360 align:middle line:84% I gain something in having reduced my future costs. 00:46:30.360 --> 00:46:33.220 align:middle line:90% 00:46:33.220 --> 00:46:38.000 align:middle line:90% So-- oh, I put a B on here. 00:46:38.000 --> 00:46:40.750 align:middle line:84% So let's suppose that there's no firm two. 00:46:40.750 --> 00:46:45.610 align:middle line:84% Then the blockaded entry level would be q1B. 00:46:45.610 --> 00:46:48.682 align:middle line:84% If I need to deter entry, and I want to deter entry, 00:46:48.682 --> 00:46:50.890 align:middle line:84% then what I'm going to do is I'm going to overproduce 00:46:50.890 --> 00:46:53.140 align:middle line:90% because investment makes-- 00:46:53.140 --> 00:46:54.430 align:middle line:90% it reduces my cost. 00:46:54.430 --> 00:46:55.250 align:middle line:90% It makes me tough. 00:46:55.250 --> 00:46:57.440 align:middle line:84% I want to hurt firm two, so I'm going to choose. 00:46:57.440 --> 00:46:58.810 align:middle line:90% q11 greater than q1B. 00:46:58.810 --> 00:47:02.270 align:middle line:90% 00:47:02.270 --> 00:47:06.890 align:middle line:90% In the entry accommodation case. 00:47:06.890 --> 00:47:10.130 align:middle line:84% I'm either going to want to underproduce or overproduce, 00:47:10.130 --> 00:47:12.680 align:middle line:84% depending on whether I'm playing a price competition game 00:47:12.680 --> 00:47:14.600 align:middle line:90% or a quantity competition game. 00:47:14.600 --> 00:47:16.820 align:middle line:84% So if I'm playing a price competition game 00:47:16.820 --> 00:47:18.440 align:middle line:84% where now suddenly I want to be nice, 00:47:18.440 --> 00:47:21.740 align:middle line:84% in price competition games, I'm going to distort downward-- 00:47:21.740 --> 00:47:25.280 align:middle line:90% have q11 less than a1 open loop. 00:47:25.280 --> 00:47:28.040 align:middle line:84% And I'm going to distort downward 00:47:28.040 --> 00:47:29.987 align:middle line:84% because I want to keep my costs high, 00:47:29.987 --> 00:47:32.570 align:middle line:84% and I want firm two to know that I'm going to have high costs, 00:47:32.570 --> 00:47:34.127 align:middle line:84% that firm two, when it enters, is 00:47:34.127 --> 00:47:35.960 align:middle line:84% going to also charge a high price because it 00:47:35.960 --> 00:47:37.500 align:middle line:90% knows I have high costs. 00:47:37.500 --> 00:47:39.900 align:middle line:84% So in the accommodation game, I want to be nice. 00:47:39.900 --> 00:47:42.690 align:middle line:84% I do that by underinvesting in cost reduction. 00:47:42.690 --> 00:47:47.510 align:middle line:84% Note, also, that we're going to typically have-- 00:47:47.510 --> 00:47:50.600 align:middle line:84% q1 open loop is going to typically 00:47:50.600 --> 00:47:52.880 align:middle line:90% be less than q1 blockaded. 00:47:52.880 --> 00:48:00.950 align:middle line:90% 00:48:00.950 --> 00:48:03.540 align:middle line:84% So to deter entry, I'm going to produce more than I 00:48:03.540 --> 00:48:05.820 align:middle line:90% get with blockaded entry. 00:48:05.820 --> 00:48:08.250 align:middle line:84% And then to accommodate, I'm going to produce less 00:48:08.250 --> 00:48:10.590 align:middle line:90% than the open loop value. 00:48:10.590 --> 00:48:14.770 align:middle line:84% And why do I say the open loop value is typically 00:48:14.770 --> 00:48:17.470 align:middle line:84% going to be less than the blockaded value? 00:48:17.470 --> 00:48:22.420 align:middle line:84% When I-- in the first period, I'm investing in cost reduction. 00:48:22.420 --> 00:48:25.120 align:middle line:84% I'm investing in cost reduction in part 00:48:25.120 --> 00:48:27.670 align:middle line:84% because I'm then going to have lower cost in every unit I 00:48:27.670 --> 00:48:30.760 align:middle line:90% sell in the second period. 00:48:30.760 --> 00:48:33.470 align:middle line:84% In the blockaded case, that's what I'm taking into account. 00:48:33.470 --> 00:48:35.762 align:middle line:84% How many units am I going to sell in the second period? 00:48:35.762 --> 00:48:37.240 align:middle line:84% What's my cost savings going to be? 00:48:37.240 --> 00:48:38.980 align:middle line:84% In the open loop case, it's going to be-- 00:48:38.980 --> 00:48:40.870 align:middle line:84% I'm going to be in a duopoly in the second period, 00:48:40.870 --> 00:48:42.885 align:middle line:84% so I'm only going to be serving half the market. 00:48:42.885 --> 00:48:44.260 align:middle line:84% When I'm serving half the market, 00:48:44.260 --> 00:48:46.700 align:middle line:84% I'm going to sell fewer units in the second period. 00:48:46.700 --> 00:48:50.060 align:middle line:84% Therefore, I invest less in cost reduction in the first period. 00:48:50.060 --> 00:48:54.940 align:middle line:84% So I do less overproduction relative to what maximizes this 00:48:54.940 --> 00:48:58.030 align:middle line:84% because I know that I'm only going to be-- 00:48:58.030 --> 00:49:00.488 align:middle line:84% I'm going to be selling fewer units of t equals 2 anyway. 00:49:00.488 --> 00:49:02.030 align:middle line:84% I don't value cost reduction as much. 00:49:02.030 --> 00:49:04.770 align:middle line:84% I go closer to just the static monopoly profit. 00:49:04.770 --> 00:49:10.520 align:middle line:90% 00:49:10.520 --> 00:49:14.600 align:middle line:84% Now, if we were playing in a Cournot game, in a Cournot game, 00:49:14.600 --> 00:49:18.800 align:middle line:84% I'm going to choose q11 greater than a1 open loop. 00:49:18.800 --> 00:49:22.370 align:middle line:84% So this would be in a Bertrand kind 00:49:22.370 --> 00:49:25.670 align:middle line:90% of game or a Hotelling game. 00:49:25.670 --> 00:49:30.795 align:middle line:84% In a Cournot game, what I'm going to know is that Cournot-- 00:49:30.795 --> 00:49:34.550 align:middle line:90% 00:49:34.550 --> 00:49:39.800 align:middle line:84% I'm going to know that q1 open loop is less than a1 blockaded 00:49:39.800 --> 00:49:42.980 align:middle line:90% less than q1 entry deterrence. 00:49:42.980 --> 00:49:48.110 align:middle line:90% And then q1 accommodation-- 00:49:48.110 --> 00:49:49.610 align:middle line:84% I don't know where it's going to be. 00:49:49.610 --> 00:49:52.490 align:middle line:90% It could be here. 00:49:52.490 --> 00:49:54.150 align:middle line:90% It could be here. 00:49:54.150 --> 00:49:57.860 align:middle line:84% We know it's bigger than q1 open loop. 00:49:57.860 --> 00:50:00.960 align:middle line:84% But q1 loop is less than q1 blockaded. 00:50:00.960 --> 00:50:03.860 align:middle line:84% So it could be that you overinvest, but still not as 00:50:03.860 --> 00:50:05.460 align:middle line:90% much as you would do here. 00:50:05.460 --> 00:50:07.335 align:middle line:84% Or it could be that you have to produce a ton 00:50:07.335 --> 00:50:09.752 align:middle line:84% to keep the other firm out, but keeping the other firm out 00:50:09.752 --> 00:50:10.830 align:middle line:90% is really important. 00:50:10.830 --> 00:50:13.700 align:middle line:84% It could be you're producing up here, or even more-- 00:50:13.700 --> 00:50:17.570 align:middle line:90% even up here or something. 00:50:17.570 --> 00:50:18.440 align:middle line:90% I guess this is-- 00:50:18.440 --> 00:50:19.523 align:middle line:90% it's you're investing to-- 00:50:19.523 --> 00:50:21.800 align:middle line:90% 00:50:21.800 --> 00:50:22.880 align:middle line:90% sorry. 00:50:22.880 --> 00:50:25.250 align:middle line:84% Yeah. you're investing just to reduce his profits 00:50:25.250 --> 00:50:28.760 align:middle line:90% to reduce his output, I guess. 00:50:28.760 --> 00:50:31.320 align:middle line:84% This is reducing his output all the way to zero. 00:50:31.320 --> 00:50:33.512 align:middle line:84% So if you're reducing his output all the way to zero 00:50:33.512 --> 00:50:35.220 align:middle line:84% would be the biggest benefit you can get. 00:50:35.220 --> 00:50:37.345 align:middle line:84% So this one would have a smaller benefit than this. 00:50:37.345 --> 00:50:41.540 align:middle line:84% So I guess it would be either here or here. 00:50:41.540 --> 00:50:45.200 align:middle line:84% But in the competition game, I'm just overinvesting, 00:50:45.200 --> 00:50:47.300 align:middle line:84% but I'm overinvesting, recognizing that it's still 00:50:47.300 --> 00:50:48.217 align:middle line:90% going to be a duopoly. 00:50:48.217 --> 00:50:51.190 align:middle line:84% So I'm overinvesting relative to the somewhat lower level. 00:50:51.190 --> 00:50:57.600 align:middle line:90% 00:50:57.600 --> 00:51:00.810 align:middle line:84% Leverage theory-- so there's this old classic literature 00:51:00.810 --> 00:51:05.970 align:middle line:84% in IO about, if you're a firm, and you're a monopolist, 00:51:05.970 --> 00:51:07.297 align:middle line:90% do you get one monopoly? 00:51:07.297 --> 00:51:09.630 align:middle line:84% It's like-- there's a big Chicago school of thought that 00:51:09.630 --> 00:51:11.940 align:middle line:84% says if you have one monopoly, you 00:51:11.940 --> 00:51:14.680 align:middle line:84% can only earn one monopoly profit from it. 00:51:14.680 --> 00:51:16.940 align:middle line:90% And so if you can only get one-- 00:51:16.940 --> 00:51:18.930 align:middle line:90% you would never do something-- 00:51:18.930 --> 00:51:21.570 align:middle line:84% you would never use your monopoly power in one market 00:51:21.570 --> 00:51:23.745 align:middle line:84% to try to get extra profits in a second market. 00:51:23.745 --> 00:51:25.620 align:middle line:84% Because if you're trying to use your monopoly 00:51:25.620 --> 00:51:27.078 align:middle line:84% to get profit in the second market, 00:51:27.078 --> 00:51:28.380 align:middle line:90% the only way you can do that-- 00:51:28.380 --> 00:51:30.297 align:middle line:84% Like, if all I can do is sell people something 00:51:30.297 --> 00:51:31.920 align:middle line:90% they don't like as much-- 00:51:31.920 --> 00:51:32.790 align:middle line:90% like, if I'm Apple. 00:51:32.790 --> 00:51:36.150 align:middle line:84% I try to monopolize the iPhone charger market. 00:51:36.150 --> 00:51:37.830 align:middle line:84% People don't like iPhone chargers 00:51:37.830 --> 00:51:41.120 align:middle line:84% as much as they would have liked USB C chargers. 00:51:41.120 --> 00:51:43.500 align:middle line:84% Yes, I can charge them more by making them all 00:51:43.500 --> 00:51:44.560 align:middle line:90% buy chargers for me. 00:51:44.560 --> 00:51:46.283 align:middle line:90% But anything that they-- 00:51:46.283 --> 00:51:48.200 align:middle line:84% anything that I'm making selling the chargers, 00:51:48.200 --> 00:51:49.930 align:middle line:84% I could have just charged them more for the phone 00:51:49.930 --> 00:51:52.138 align:middle line:84% to begin with and let them buy the charger they want. 00:51:52.138 --> 00:51:53.440 align:middle line:90% So all I'm doing is just-- 00:51:53.440 --> 00:51:55.390 align:middle line:84% I'm just getting one monopoly profit. 00:51:55.390 --> 00:51:57.190 align:middle line:84% And you can't extend your monopoly power 00:51:57.190 --> 00:51:59.572 align:middle line:84% to a second market because anything people pay extra 00:51:59.572 --> 00:52:01.030 align:middle line:84% in the second market, they're going 00:52:01.030 --> 00:52:03.718 align:middle line:84% to treat as a cost of buying your initial product. 00:52:03.718 --> 00:52:06.010 align:middle line:84% And they're just going to pay less for the base product 00:52:06.010 --> 00:52:08.690 align:middle line:84% because you've tried to make them buy a second product. 00:52:08.690 --> 00:52:14.340 align:middle line:84% So that's the Chicago school critique that says, 00:52:14.340 --> 00:52:16.680 align:middle line:84% no, you can't tie two markets together 00:52:16.680 --> 00:52:18.940 align:middle line:90% and get any more profits. 00:52:18.940 --> 00:52:20.910 align:middle line:90% And then so-- 00:52:20.910 --> 00:52:24.960 align:middle line:84% Mike Whinston has this paper, classic paper, saying sometimes 00:52:24.960 --> 00:52:28.210 align:middle line:84% you will want to try to extend profits to a second market. 00:52:28.210 --> 00:52:31.068 align:middle line:84% And this comes from a strategic investment-- 00:52:31.068 --> 00:52:33.360 align:middle line:84% either strategic entry accommodation or strategic entry 00:52:33.360 --> 00:52:35.070 align:middle line:90% deterrence motive. 00:52:35.070 --> 00:52:38.610 align:middle line:84% So suppose you have a firm, firm one. 00:52:38.610 --> 00:52:42.390 align:middle line:84% It's a monopolist in good A. And suppose that there's some-- 00:52:42.390 --> 00:52:46.930 align:middle line:84% suppose that Samsung and whatever has disappeared. 00:52:46.930 --> 00:52:49.380 align:middle line:90% So now there's only iPhones. 00:52:49.380 --> 00:52:51.990 align:middle line:84% So firm one is an iPhone monopolist, 00:52:51.990 --> 00:52:53.940 align:middle line:84% but it does compete with firm two 00:52:53.940 --> 00:52:58.050 align:middle line:84% in producing chargers or earbuds or something like that. 00:52:58.050 --> 00:53:01.720 align:middle line:84% Suppose we have a homogeneous-- simple, homogeneous model. 00:53:01.720 --> 00:53:03.630 align:middle line:84% All consumers get value v from good 00:53:03.630 --> 00:53:06.930 align:middle line:84% And Then there's a demand for good B 00:53:06.930 --> 00:53:16.340 align:middle line:84% that depends on prices for good one and prices for good two. 00:53:16.340 --> 00:53:18.417 align:middle line:90% 00:53:18.417 --> 00:53:20.000 align:middle line:84% So I'm just thinking, in this example, 00:53:20.000 --> 00:53:21.417 align:middle line:84% this has to be something where you 00:53:21.417 --> 00:53:24.980 align:middle line:84% can use good B without good A. So I guess the phone 00:53:24.980 --> 00:53:26.800 align:middle line:84% and the charger does not work here. 00:53:26.800 --> 00:53:33.020 align:middle line:90% 00:53:33.020 --> 00:53:33.520 align:middle line:90% OK. 00:53:33.520 --> 00:53:35.680 align:middle line:90% So suppose these are-- 00:53:35.680 --> 00:53:39.200 align:middle line:84% this is taking us back in the day. 00:53:39.200 --> 00:53:41.410 align:middle line:84% Suppose that Apple's the iPhone monopolist. 00:53:41.410 --> 00:53:43.127 align:middle line:90% Good B is headphones. 00:53:43.127 --> 00:53:44.710 align:middle line:84% But suppose that headphones could also 00:53:44.710 --> 00:53:47.350 align:middle line:84% be used to listen to an MP3 player, which 00:53:47.350 --> 00:53:49.000 align:middle line:84% was a separate object that existed 00:53:49.000 --> 00:53:51.040 align:middle line:84% that you could use to listen to music on, not 00:53:51.040 --> 00:53:52.090 align:middle line:90% through your phone. 00:53:52.090 --> 00:53:54.067 align:middle line:84% So you could just have a separate thing 00:53:54.067 --> 00:53:56.400 align:middle line:84% that looked like a phone that you carried in your pocket 00:53:56.400 --> 00:54:00.340 align:middle line:84% that you could play music on and listen to music on it. 00:54:00.340 --> 00:54:00.840 align:middle line:90% So anyway. 00:54:00.840 --> 00:54:02.800 align:middle line:84% So there's a demand for headphones 00:54:02.800 --> 00:54:05.748 align:middle line:84% even if you don't have a-- even if you can't afford a phone, 00:54:05.748 --> 00:54:07.290 align:middle line:84% you just have headphones that you use 00:54:07.290 --> 00:54:09.780 align:middle line:90% to listen to music otherwise. 00:54:09.780 --> 00:54:10.350 align:middle line:90% Anyway. 00:54:10.350 --> 00:54:12.040 align:middle line:90% So firm one has this choice. 00:54:12.040 --> 00:54:14.515 align:middle line:84% It can design its products so that they cannot be used 00:54:14.515 --> 00:54:17.140 align:middle line:84% separately, or it can design it so that you just plug and play. 00:54:17.140 --> 00:54:20.130 align:middle line:84% You can use anybody's earphones with your iPhone. 00:54:20.130 --> 00:54:23.380 align:middle line:84% So imagine first firm one makes a design decision. 00:54:23.380 --> 00:54:25.980 align:middle line:84% Then firm two chooses whether to enter and compete 00:54:25.980 --> 00:54:26.890 align:middle line:90% with the monopolist. 00:54:26.890 --> 00:54:28.765 align:middle line:84% And then we either get a monopoly or duopoly. 00:54:28.765 --> 00:54:31.730 align:middle line:90% 00:54:31.730 --> 00:54:37.940 align:middle line:84% In the-- if firm two comes in, it's going to set a price for-- 00:54:37.940 --> 00:54:39.860 align:middle line:84% firm two is going to set a price for good B 00:54:39.860 --> 00:54:42.140 align:middle line:84% and doesn't sell good A. And then 00:54:42.140 --> 00:54:46.280 align:middle line:84% firm one is either going to be only selling 00:54:46.280 --> 00:54:49.880 align:middle line:84% a bundle of a phone with earbuds, 00:54:49.880 --> 00:54:52.530 align:middle line:84% or it's going to be selling the two products separately. 00:54:52.530 --> 00:54:53.938 align:middle line:90% You can buy the phone from it. 00:54:53.938 --> 00:54:55.730 align:middle line:84% You can buy the earbuds from it separately, 00:54:55.730 --> 00:54:59.090 align:middle line:84% but they're sold at two separate prices. 00:54:59.090 --> 00:55:01.880 align:middle line:84% And we're going to think about in this model, 00:55:01.880 --> 00:55:06.020 align:middle line:84% would Apple design its product so that every phone comes 00:55:06.020 --> 00:55:09.230 align:middle line:84% with earbuds included, so that no one would then ever 00:55:09.230 --> 00:55:11.080 align:middle line:90% buy earphones from anyone else? 00:55:11.080 --> 00:55:13.860 align:middle line:90% 00:55:13.860 --> 00:55:20.370 align:middle line:84% So absent any strategic effects, if you can't prevent entry, 00:55:20.370 --> 00:55:23.610 align:middle line:84% you can't alter p2, firm one would never 00:55:23.610 --> 00:55:25.650 align:middle line:90% tie the products together. 00:55:25.650 --> 00:55:30.990 align:middle line:84% Because if you sell this bundle at price p1 AB, 00:55:30.990 --> 00:55:35.370 align:middle line:84% you could instead just sell good A to everyone at price v, 00:55:35.370 --> 00:55:39.980 align:middle line:84% and then sell the bundle at price p1 AB minus v. 00:55:39.980 --> 00:55:45.750 align:middle line:90% And if you do this, every-- 00:55:45.750 --> 00:55:48.420 align:middle line:84% once-- with this separate pricing strategy, 00:55:48.420 --> 00:55:50.550 align:middle line:84% everybody in the world is going to buy the iPhone 00:55:50.550 --> 00:55:58.300 align:middle line:84% at price v. And sales of good B are unchanged. 00:55:58.300 --> 00:56:02.240 align:middle line:84% Because anybody who-- people are making this choice. 00:56:02.240 --> 00:56:03.340 align:middle line:90% Do I buy the Apple bundle? 00:56:03.340 --> 00:56:06.610 align:middle line:84% Or do I just not have a phone and just buy earphones 00:56:06.610 --> 00:56:08.440 align:middle line:90% to listen to my MP3 player? 00:56:08.440 --> 00:56:12.010 align:middle line:84% Anybody-- you sell exactly the same-- 00:56:12.010 --> 00:56:14.830 align:middle line:90% anyone who values the iPhone-- 00:56:14.830 --> 00:56:19.780 align:middle line:84% anyone who values the Apple earbuds at more than this 00:56:19.780 --> 00:56:21.110 align:middle line:90% is going to buy from Apple. 00:56:21.110 --> 00:56:23.500 align:middle line:84% Anyone who buys from less than this is going to now-- 00:56:23.500 --> 00:56:25.400 align:middle line:84% is going to buy it not from Apple. 00:56:25.400 --> 00:56:27.910 align:middle line:84% And so you sell exactly the same number 00:56:27.910 --> 00:56:31.730 align:middle line:84% of units of earphones at these unbundled prices. 00:56:31.730 --> 00:56:34.150 align:middle line:84% But you sell more phones because there are some people who 00:56:34.150 --> 00:56:37.512 align:middle line:84% would like the phone as a standalone, 00:56:37.512 --> 00:56:39.220 align:middle line:84% but they really like the outside earbuds, 00:56:39.220 --> 00:56:41.803 align:middle line:84% so that they buy-- they don't buy the phone when they can only 00:56:41.803 --> 00:56:45.440 align:middle line:84% buy the phone bundled with the earbuds. 00:56:45.440 --> 00:56:49.323 align:middle line:84% So in a nonstrategic model, you would never 00:56:49.323 --> 00:56:51.740 align:middle line:84% tie these two products together because it's always better 00:56:51.740 --> 00:56:53.448 align:middle line:84% to let people buy what they want and just 00:56:53.448 --> 00:56:55.050 align:middle line:84% have the two of them sold separately. 00:56:55.050 --> 00:56:57.710 align:middle line:90% 00:56:57.710 --> 00:57:00.710 align:middle line:84% But think about tying the product-- 00:57:00.710 --> 00:57:03.170 align:middle line:84% tying the product-- it's a strategic investment 00:57:03.170 --> 00:57:04.760 align:middle line:90% that makes firm one tough. 00:57:04.760 --> 00:57:07.250 align:middle line:90% It hurts firms two. 00:57:07.250 --> 00:57:08.250 align:middle line:90% Why is that? 00:57:08.250 --> 00:57:10.970 align:middle line:90% Well, because it's as if-- 00:57:10.970 --> 00:57:13.580 align:middle line:84% you've got this profit you're earning on good A-- 00:57:13.580 --> 00:57:15.620 align:middle line:90% v minus CA. 00:57:15.620 --> 00:57:21.380 align:middle line:84% But you earn that profit on good A only if you sell good B. 00:57:21.380 --> 00:57:24.920 align:middle line:84% So it's as if your opportunity cost of selling B 00:57:24.920 --> 00:57:27.290 align:middle line:84% is reduced from the cost of producing 00:57:27.290 --> 00:57:32.280 align:middle line:84% good B to the cost of good B minus this opportunity cost. 00:57:32.280 --> 00:57:35.490 align:middle line:84% So every time I sell good B, I'm selling good A-- 00:57:35.490 --> 00:57:37.520 align:middle line:84% every time I sell good B, I'm selling good A 00:57:37.520 --> 00:57:38.740 align:middle line:90% and earning this profit. 00:57:38.740 --> 00:57:40.980 align:middle line:84% So it's like, this is my opportunity-- that's 00:57:40.980 --> 00:57:43.910 align:middle line:84% my actual cost of selling good B. 00:57:43.910 --> 00:57:49.340 align:middle line:84% And so this is like bundling the goods together. 00:57:49.340 --> 00:57:51.030 align:middle line:90% You can think of it as a cost. 00:57:51.030 --> 00:57:55.540 align:middle line:84% It's like a cost reduction on good B. Or it's similar-- 00:57:55.540 --> 00:57:59.370 align:middle line:84% it's identical to a cost reduction algebraically. 00:57:59.370 --> 00:58:01.620 align:middle line:84% So therefore, if you want to tie-- 00:58:01.620 --> 00:58:07.470 align:middle line:84% if you want to prevent any firm from entering market B, 00:58:07.470 --> 00:58:10.470 align:middle line:84% I tie the goods together to deter entry. 00:58:10.470 --> 00:58:12.240 align:middle line:84% Because tying the goods together is 00:58:12.240 --> 00:58:15.400 align:middle line:84% going to reduce the number of headphones that can get sold. 00:58:15.400 --> 00:58:18.570 align:middle line:84% And by reducing the number of headphones that can get sold, 00:58:18.570 --> 00:58:20.190 align:middle line:84% I may keep firm two out of the market 00:58:20.190 --> 00:58:22.920 align:middle line:84% entirely and have a monopoly on the phone-- on both of them 00:58:22.920 --> 00:58:23.880 align:middle line:90% together. 00:58:23.880 --> 00:58:28.350 align:middle line:84% So this is a story where if you think about the entry decision, 00:58:28.350 --> 00:58:30.895 align:middle line:84% you do have an incentive to tie goods together. 00:58:30.895 --> 00:58:32.520 align:middle line:84% And by tying the goods together, you're 00:58:32.520 --> 00:58:33.690 align:middle line:84% earning higher profits than you would 00:58:33.690 --> 00:58:34.930 align:middle line:90% earn not tying them together. 00:58:34.930 --> 00:58:37.860 align:middle line:84% So in some sense, you're going beyond your monopoly profits 00:58:37.860 --> 00:58:40.790 align:middle line:84% in firm one and also gaining profits on market two. 00:58:40.790 --> 00:58:44.100 align:middle line:90% 00:58:44.100 --> 00:58:47.687 align:middle line:84% In the entry accommodation game, this 00:58:47.687 --> 00:58:49.270 align:middle line:84% is an investment that makes you tough. 00:58:49.270 --> 00:58:52.980 align:middle line:84% So if you're in price competition in market B, 00:58:52.980 --> 00:58:57.970 align:middle line:84% you would not want to make this tie because it hurts firm two. 00:58:57.970 --> 00:59:01.050 align:middle line:84% And in a price competition game, you want to help your rivals. 00:59:01.050 --> 00:59:04.590 align:middle line:84% So if you are competing in a price competition game in market 00:59:04.590 --> 00:59:06.810 align:middle line:84% B, then in the entry accommodate-- 00:59:06.810 --> 00:59:09.400 align:middle line:84% then you would not want to tie the products together. 00:59:09.400 --> 00:59:12.580 align:middle line:84% And the leverage theory of tying doesn't make any sense. 00:59:12.580 --> 00:59:17.190 align:middle line:84% You can't be tying them in order to get extra profits. 00:59:17.190 --> 00:59:20.260 align:middle line:90% You would always only tie-- 00:59:20.260 --> 00:59:22.260 align:middle line:84% in some sense, you're going to have an incentive 00:59:22.260 --> 00:59:24.540 align:middle line:84% to not tie them because not tying them 00:59:24.540 --> 00:59:27.270 align:middle line:84% strategically gets firm two to price higher. 00:59:27.270 --> 00:59:30.090 align:middle line:84% Now, if somehow this was an example where you had a Cournot 00:59:30.090 --> 00:59:34.890 align:middle line:84% market for good B, if good B was being sold in a Cournot market, 00:59:34.890 --> 00:59:40.350 align:middle line:84% then you may want to also tie products in order-- 00:59:40.350 --> 00:59:42.735 align:middle line:84% so it's like, I'm doing something or other. 00:59:42.735 --> 00:59:44.110 align:middle line:84% If you buy this from me, you also 00:59:44.110 --> 00:59:46.010 align:middle line:90% have to buy your coal from me. 00:59:46.010 --> 00:59:48.500 align:middle line:84% And then coal was produced in a Cournot market, 00:59:48.500 --> 00:59:51.760 align:middle line:84% I might want to tie goods together in order-- 00:59:51.760 --> 00:59:55.180 align:middle line:84% for the same reason, to be mean to you to get you to produce 00:59:55.180 --> 00:59:57.030 align:middle line:90% less coal so that we're-- 00:59:57.030 --> 00:59:59.800 align:middle line:90% 00:59:59.800 --> 01:00:02.550 align:middle line:84% that I'm hurting you in the accommodation game. 01:00:02.550 --> 01:00:08.770 align:middle line:90% 01:00:08.770 --> 01:00:13.150 align:middle line:84% But I think it does suggest that it's the leverage 01:00:13.150 --> 01:00:16.030 align:middle line:84% to deter entry is an easy story to tell, 01:00:16.030 --> 01:00:18.070 align:middle line:84% and leverage to accommodate entry, 01:00:18.070 --> 01:00:20.720 align:middle line:84% or the tying to accommodate entry, 01:00:20.720 --> 01:00:23.820 align:middle line:84% you're going to have to come up with strategic substitutes. 01:00:23.820 --> 01:00:25.570 align:middle line:84% I think it's a strategic substitutes game, 01:00:25.570 --> 01:00:26.970 align:middle line:90% and that's going to be-- 01:00:26.970 --> 01:00:28.905 align:middle line:84% not apply to many things that we think of. 01:00:28.905 --> 01:00:31.470 align:middle line:90% 01:00:31.470 --> 01:00:33.480 align:middle line:90% So limit pricing. 01:00:33.480 --> 01:00:37.410 align:middle line:84% Yeah, the next two stories I'm going to say is-- 01:00:37.410 --> 01:00:39.210 align:middle line:84% again, as I've said, many things can 01:00:39.210 --> 01:00:42.640 align:middle line:84% be thought of as an investment in these previous stories. 01:00:42.640 --> 01:00:44.860 align:middle line:84% An investment is anything that's sunk. 01:00:44.860 --> 01:00:47.713 align:middle line:84% And once it's sunk, you can't easily undo it. 01:00:47.713 --> 01:00:49.380 align:middle line:84% And it's going to affect the competition 01:00:49.380 --> 01:00:51.870 align:middle line:90% after the rival enters. 01:00:51.870 --> 01:00:55.660 align:middle line:84% Can we think of prices as an entry-deterring device? 01:00:55.660 --> 01:00:57.510 align:middle line:84% So the classic idea of limit pricing 01:00:57.510 --> 01:01:00.720 align:middle line:84% is that you have a firm facing entry. 01:01:00.720 --> 01:01:03.340 align:middle line:84% And the thought in the limit pricing literature was that way. 01:01:03.340 --> 01:01:04.923 align:middle line:84% You had this idea that firms are going 01:01:04.923 --> 01:01:06.780 align:middle line:84% to set low prices because they want 01:01:06.780 --> 01:01:09.550 align:middle line:84% to keep rivals from entering and competing with them. 01:01:09.550 --> 01:01:12.513 align:middle line:84% And there's a natural critique of that idea, this limit pricing 01:01:12.513 --> 01:01:13.680 align:middle line:90% idea, that says, well, wait. 01:01:13.680 --> 01:01:14.590 align:middle line:90% How would that work? 01:01:14.590 --> 01:01:16.673 align:middle line:84% Sure, you can set your prices low so that everyone 01:01:16.673 --> 01:01:18.060 align:middle line:90% sees that your prices are low. 01:01:18.060 --> 01:01:19.858 align:middle line:84% But they know that if you enter, we're 01:01:19.858 --> 01:01:22.150 align:middle line:84% just going to play the static Nash equilibrium anyways. 01:01:22.150 --> 01:01:25.260 align:middle line:84% The fact that you wasted money in all the previous periods 01:01:25.260 --> 01:01:29.820 align:middle line:84% by pricing below the monopoly price, it's just-- 01:01:29.820 --> 01:01:33.190 align:middle line:84% it's a sunk loss, but it just-- it has 01:01:33.190 --> 01:01:34.440 align:middle line:90% nothing to do with the future. 01:01:34.440 --> 01:01:36.107 align:middle line:84% If it has nothing to do with the future, 01:01:36.107 --> 01:01:38.490 align:middle line:84% that can't-- you can't deter entry by setting low prices 01:01:38.490 --> 01:01:41.040 align:middle line:84% to avoid getting people to come in and compete with you. 01:01:41.040 --> 01:01:45.270 align:middle line:84% And so there are several different parts 01:01:45.270 --> 01:01:48.210 align:middle line:84% of the IO literature that are discussed, stories 01:01:48.210 --> 01:01:50.850 align:middle line:84% that would account for firms wanting 01:01:50.850 --> 01:01:53.760 align:middle line:84% to price below the static monopoly profit 01:01:53.760 --> 01:01:56.250 align:middle line:90% in order to keep firms out. 01:01:56.250 --> 01:01:58.650 align:middle line:90% One of them is a classic-- 01:01:58.650 --> 01:02:02.370 align:middle line:84% is Milgrom and Roberts, 1982, noted 01:02:02.370 --> 01:02:06.630 align:middle line:84% that you could have limit pricing if firm two doesn't 01:02:06.630 --> 01:02:12.170 align:middle line:84% know firm one's costs, and prices are a signal of costs. 01:02:12.170 --> 01:02:15.650 align:middle line:84% So this is going to be a classic signaling game. 01:02:15.650 --> 01:02:19.250 align:middle line:84% Firm one has two possible cost levels, C lower bar 01:02:19.250 --> 01:02:20.780 align:middle line:90% or C upper bar. 01:02:20.780 --> 01:02:23.240 align:middle line:84% Assume that firm one knows its costs. 01:02:23.240 --> 01:02:26.780 align:middle line:84% Firm two doesn't know firm one's costs. 01:02:26.780 --> 01:02:29.720 align:middle line:84% Firm two has some prior that there's probability mu 01:02:29.720 --> 01:02:33.740 align:middle line:84% that its a high-cost firm, and 1 minus mu that firm one's 01:02:33.740 --> 01:02:35.180 align:middle line:90% a low-cost firm. 01:02:35.180 --> 01:02:38.330 align:middle line:84% Firm 2 is going to have some separate costs, C2, 01:02:38.330 --> 01:02:40.010 align:middle line:90% if it comes in and competes. 01:02:40.010 --> 01:02:43.400 align:middle line:84% It knows C2, but it doesn't know whether firm one has a better 01:02:43.400 --> 01:02:44.910 align:middle line:90% or worse technology than it is. 01:02:44.910 --> 01:02:48.170 align:middle line:84% So it doesn't know if firm one is a lower cost 01:02:48.170 --> 01:02:53.050 align:middle line:84% firm than it is, or firm one is a higher cost firm than it is. 01:02:53.050 --> 01:02:56.020 align:middle line:84% And so we're going to think of a two-period model, where firm 01:02:56.020 --> 01:02:59.530 align:middle line:90% one chooses p11 at t equals 1. 01:02:59.530 --> 01:03:02.650 align:middle line:84% Firm two observes the price that firm one is charging, 01:03:02.650 --> 01:03:05.230 align:middle line:84% and then it chooses whether to enter or not 01:03:05.230 --> 01:03:08.170 align:middle line:84% and has some fixed entry cost, E. 01:03:08.170 --> 01:03:13.000 align:middle line:84% And then we either get a monopoly, 01:03:13.000 --> 01:03:17.380 align:middle line:84% where firm one produces and earns monopoly profits that 01:03:17.380 --> 01:03:19.870 align:middle line:90% depend on its actual cost, C1. 01:03:19.870 --> 01:03:24.170 align:middle line:84% Or we get a duopoly, where the profits depend on the costs C1 01:03:24.170 --> 01:03:24.670 align:middle line:90% and C2. 01:03:24.670 --> 01:03:28.070 align:middle line:90% 01:03:28.070 --> 01:03:30.433 align:middle line:84% And in this model, Milgrom and Roberts 01:03:30.433 --> 01:03:31.850 align:middle line:84% notes there are two different ways 01:03:31.850 --> 01:03:35.780 align:middle line:84% we can think of where we can get firm one in equilibrium 01:03:35.780 --> 01:03:41.270 align:middle line:84% to set a price lower than it would set in the blockaded entry 01:03:41.270 --> 01:03:43.300 align:middle line:90% case. 01:03:43.300 --> 01:03:47.770 align:middle line:84% Or in this case, it's simply the profit-maximizing case. 01:03:47.770 --> 01:03:49.620 align:middle line:90% And why can that be? 01:03:49.620 --> 01:03:52.360 align:middle line:90% So there are two cases. 01:03:52.360 --> 01:04:06.040 align:middle line:84% Case one is E, such that firm two won't enter 01:04:06.040 --> 01:04:07.516 align:middle line:90% if it gets no information. 01:04:07.516 --> 01:04:11.290 align:middle line:90% 01:04:11.290 --> 01:04:14.185 align:middle line:84% So imagine that firm two-- it's unsure about where 01:04:14.185 --> 01:04:17.185 align:middle line:90% the firm one's costs are. 01:04:17.185 --> 01:04:19.560 align:middle line:84% We're going to typically think of if firm one has costs C 01:04:19.560 --> 01:04:21.690 align:middle line:84% lower bar, firm two is going to get low profits 01:04:21.690 --> 01:04:23.320 align:middle line:90% and wish it hadn't entered. 01:04:23.320 --> 01:04:25.350 align:middle line:84% If C is C upper bar, then firm two 01:04:25.350 --> 01:04:28.200 align:middle line:84% is going to earn high profits and be happy that it entered. 01:04:28.200 --> 01:04:30.420 align:middle line:84% But imagine that the balance between those two 01:04:30.420 --> 01:04:34.440 align:middle line:84% is that if firm two doesn't get any new information, 01:04:34.440 --> 01:04:36.690 align:middle line:84% so it still has this prior mu 1 minus mu, 01:04:36.690 --> 01:04:39.930 align:middle line:84% it's going to stay out of the market. 01:04:39.930 --> 01:04:42.900 align:middle line:84% Well, if firm two would stay out, 01:04:42.900 --> 01:04:45.300 align:middle line:84% if it gets no new information, we 01:04:45.300 --> 01:04:48.300 align:middle line:90% can get a pooling equilibrium. 01:04:48.300 --> 01:04:50.160 align:middle line:84% And the pooling equilibrium-- firm 01:04:50.160 --> 01:04:53.160 align:middle line:84% one is setting a low price at t equals 1, regardless of 01:04:53.160 --> 01:04:54.760 align:middle line:90% whether its cost is. 01:04:54.760 --> 01:04:57.300 align:middle line:84% And then firm two stays out if it sees this price 01:04:57.300 --> 01:05:00.050 align:middle line:90% and enters otherwise. 01:05:00.050 --> 01:05:04.670 align:middle line:84% So you imagine there are these two different cost levels or two 01:05:04.670 --> 01:05:06.290 align:middle line:90% different price levels. 01:05:06.290 --> 01:05:12.410 align:middle line:84% P1 lower bar would be the argmax of firm one's profits. 01:05:12.410 --> 01:05:16.340 align:middle line:84% If it's at price P1 and its cost to C lower bar. 01:05:16.340 --> 01:05:22.250 align:middle line:84% And P1 upper bar would be the argmax of pi 1-- 01:05:22.250 --> 01:05:23.780 align:middle line:90% these are monopoly profits-- 01:05:23.780 --> 01:05:26.830 align:middle line:90% at P1 C upper bar. 01:05:26.830 --> 01:05:31.570 align:middle line:84% So these two different profit level-- two different price 01:05:31.570 --> 01:05:36.700 align:middle line:84% levels that you would set if you had low cost or high cost. 01:05:36.700 --> 01:05:40.660 align:middle line:84% But if you imagine-- if we tried to have an equilibrium where 01:05:40.660 --> 01:05:42.880 align:middle line:84% firm one set this price when its cost was low 01:05:42.880 --> 01:05:47.410 align:middle line:84% and this price when its cost was high, if you set this price, 01:05:47.410 --> 01:05:50.380 align:middle line:84% firm two is going to realize that you have high cost. 01:05:50.380 --> 01:05:52.360 align:middle line:84% And then firm two is going to enter. 01:05:52.360 --> 01:05:54.730 align:middle line:84% So what we can get is this pooling equilibrium, 01:05:54.730 --> 01:06:01.660 align:middle line:84% where we have p11 equals p1 lower bar for all C. 01:06:01.660 --> 01:06:05.650 align:middle line:84% So if your cost is low, you set price C lower bar. 01:06:05.650 --> 01:06:06.990 align:middle line:90% If your cost is high-- 01:06:06.990 --> 01:06:07.490 align:middle line:90% sorry. 01:06:07.490 --> 01:06:10.460 align:middle line:84% If you set-- price is low, you set price P1 lower bar. 01:06:10.460 --> 01:06:14.670 align:middle line:84% But if your cost is high, you also set price P1 lower bar. 01:06:14.670 --> 01:06:17.130 align:middle line:84% And so you're pricing low because 01:06:17.130 --> 01:06:20.040 align:middle line:90% by pricing low, firm two-- 01:06:20.040 --> 01:06:22.140 align:middle line:84% firm two knows that firm one price is low 01:06:22.140 --> 01:06:24.100 align:middle line:90% even if it's cost is high. 01:06:24.100 --> 01:06:28.020 align:middle line:84% But by doing that, it prevents firm two from learning anything. 01:06:28.020 --> 01:06:30.160 align:middle line:84% And by preventing firm to from learning anything, 01:06:30.160 --> 01:06:31.505 align:middle line:90% firm two isn't going to enter. 01:06:31.505 --> 01:06:38.860 align:middle line:90% 01:06:38.860 --> 01:06:42.710 align:middle line:84% So kind of a familiar argument how signaling models typically 01:06:42.710 --> 01:06:43.210 align:middle line:90% work. 01:06:43.210 --> 01:06:45.085 align:middle line:84% That is, firm two is going to infer something 01:06:45.085 --> 01:06:46.340 align:middle line:90% from your equilibrium play. 01:06:46.340 --> 01:06:48.160 align:middle line:84% So we can have this pooling equilibrium 01:06:48.160 --> 01:06:51.267 align:middle line:84% where we all charge a low price regardless of our costs, 01:06:51.267 --> 01:06:53.350 align:middle line:84% and that prevents firm two from learning anything. 01:06:53.350 --> 01:06:57.810 align:middle line:84% That gives you limit pricing to keep firm two out. 01:06:57.810 --> 01:07:01.175 align:middle line:84% The other thing you can get-- and this would be the case two. 01:07:01.175 --> 01:07:05.360 align:middle line:90% 01:07:05.360 --> 01:07:14.360 align:middle line:84% Case two is E, such that firm two enters 01:07:14.360 --> 01:07:16.630 align:middle line:90% if it gets no new info. 01:07:16.630 --> 01:07:19.400 align:middle line:90% 01:07:19.400 --> 01:07:22.760 align:middle line:84% In this case, clearly, we can't have this pooling equilibrium 01:07:22.760 --> 01:07:23.300 align:middle line:90% anymore. 01:07:23.300 --> 01:07:25.550 align:middle line:84% Because in this pooling equilibrium, 01:07:25.550 --> 01:07:27.590 align:middle line:84% if this was how people played the game, 01:07:27.590 --> 01:07:30.690 align:middle line:84% then all the-- then firm two wouldn't learn anything. 01:07:30.690 --> 01:07:32.700 align:middle line:84% But then firm two would always enter. 01:07:32.700 --> 01:07:34.310 align:middle line:84% And so if you're a high-cost firm, 01:07:34.310 --> 01:07:36.920 align:middle line:84% why are you charging a low price when it doesn't affect entry 01:07:36.920 --> 01:07:37.980 align:middle line:90% in the future? 01:07:37.980 --> 01:07:40.350 align:middle line:84% That's, again, this here relevant entry case. 01:07:40.350 --> 01:07:43.910 align:middle line:84% But so we can get this equilibrium where we have P1-- 01:07:43.910 --> 01:07:46.940 align:middle line:90% 01:07:46.940 --> 01:07:48.890 align:middle line:84% I'll call it the limit pricing equilibrium. 01:07:48.890 --> 01:07:58.190 align:middle line:84% P1 limit pricing of C1 is less than P1 lower bar. 01:07:58.190 --> 01:08:02.000 align:middle line:90% And P1 limit pricing of C1-- 01:08:02.000 --> 01:08:08.720 align:middle line:84% sorry-- of C lower bar and of C upper bar equals P1 upper bar. 01:08:08.720 --> 01:08:12.350 align:middle line:84% So we can get this equilibrium where if your cost is high, 01:08:12.350 --> 01:08:14.160 align:middle line:84% you just admit your cost is high. 01:08:14.160 --> 01:08:15.500 align:middle line:90% Set a high price. 01:08:15.500 --> 01:08:19.740 align:middle line:84% Firm two realize you have a high price and high cost and comes in 01:08:19.740 --> 01:08:21.600 align:middle line:84% and enters and competes with you. 01:08:21.600 --> 01:08:27.149 align:middle line:84% But if your cost is low, you charge something super low, 01:08:27.149 --> 01:08:30.510 align:middle line:84% even lower than P1 lower bar, in some sense 01:08:30.510 --> 01:08:35.399 align:middle line:90% to prove too firm two that-- 01:08:35.399 --> 01:08:38.609 align:middle line:84% prove to firm two that you are the low-cost firm. 01:08:38.609 --> 01:08:41.439 align:middle line:84% And why might you need to charge something lower than this? 01:08:41.439 --> 01:08:44.623 align:middle line:84% Because it could be that if you just charge P1 lower bar, 01:08:44.623 --> 01:08:46.290 align:middle line:84% that wouldn't be possible in equilibrium 01:08:46.290 --> 01:08:47.915 align:middle line:84% because then the C upper bar firm would 01:08:47.915 --> 01:08:49.740 align:middle line:90% pretend to be a low-cost firm. 01:08:49.740 --> 01:08:54.810 align:middle line:84% But by pricing way, way below, even below my static best 01:08:54.810 --> 01:08:58.500 align:middle line:84% response price, I set a price so low that a high-cost firm 01:08:58.500 --> 01:08:59.590 align:middle line:90% wouldn't copy me. 01:08:59.590 --> 01:09:04.060 align:middle line:84% And then it's a credible signal that my cost really is low. 01:09:04.060 --> 01:09:07.000 align:middle line:84% So we can get these two different types of equilibria-- 01:09:07.000 --> 01:09:08.620 align:middle line:84% the pooling equilibria, where it's 01:09:08.620 --> 01:09:15.100 align:middle line:84% the high-cost guys who price below their static optimum. 01:09:15.100 --> 01:09:16.990 align:middle line:84% And then the separating equilibria, 01:09:16.990 --> 01:09:18.790 align:middle line:84% where the low-cost guys are pricing 01:09:18.790 --> 01:09:20.850 align:middle line:84% below their static optimum to signal their costs. 01:09:20.850 --> 01:09:24.430 align:middle line:90% 01:09:24.430 --> 01:09:27.130 align:middle line:84% So that's a cost-signaling argument 01:09:27.130 --> 01:09:31.000 align:middle line:84% for why firms might want-- and it's plausible 01:09:31.000 --> 01:09:33.370 align:middle line:84% that the firm in the market knows more about their costs 01:09:33.370 --> 01:09:35.479 align:middle line:90% than the other firm does. 01:09:35.479 --> 01:09:37.870 align:middle line:84% So you price low to signal your costs. 01:09:37.870 --> 01:09:39.790 align:middle line:84% And that is why we see firms pricing 01:09:39.790 --> 01:09:41.630 align:middle line:84% low to try to keep out competition. 01:09:41.630 --> 01:09:48.790 align:middle line:90% 01:09:48.790 --> 01:09:49.300 align:middle line:90% OK. 01:09:49.300 --> 01:09:51.092 align:middle line:84% Second kind of story I was going to tell us 01:09:51.092 --> 01:09:56.320 align:middle line:84% what Fudenberg and Tirole call a signal-jamming story. 01:09:56.320 --> 01:09:59.710 align:middle line:84% In signal-jamming models, firms one and two 01:09:59.710 --> 01:10:02.650 align:middle line:84% are both symmetrically uninformed about some state 01:10:02.650 --> 01:10:04.450 align:middle line:90% of nature. 01:10:04.450 --> 01:10:07.450 align:middle line:84% But, again, signal jamming can be motivated by a desire 01:10:07.450 --> 01:10:10.270 align:middle line:90% to deter entry or induce exit. 01:10:10.270 --> 01:10:16.660 align:middle line:84% And it, again, can look like pricing low either to try to-- 01:10:16.660 --> 01:10:19.150 align:middle line:84% pricing low either to prevent firm two from entering, 01:10:19.150 --> 01:10:22.097 align:middle line:84% or to force firm two to exit a market. 01:10:22.097 --> 01:10:23.930 align:middle line:84% So I thought I would do this one just like-- 01:10:23.930 --> 01:10:28.190 align:middle line:84% I've been doing everything in terms of avoiding entry. 01:10:28.190 --> 01:10:30.730 align:middle line:84% I figured I'd do this one in terms of forcing exit. 01:10:30.730 --> 01:10:33.620 align:middle line:84% But some of the same stories can be told both kinds of games. 01:10:33.620 --> 01:10:35.080 align:middle line:90% So here's my model. 01:10:35.080 --> 01:10:40.077 align:middle line:84% So firms one and two choose quantities q11 and q21 01:10:40.077 --> 01:10:40.910 align:middle line:90% in the first period. 01:10:40.910 --> 01:10:44.730 align:middle line:84% So I'm doing this in a Cournot model. 01:10:44.730 --> 01:10:47.730 align:middle line:84% There is a market demand that determines some price. 01:10:47.730 --> 01:10:52.860 align:middle line:84% And the market demand is theta minus q1 plus q2, where theta is 01:10:52.860 --> 01:10:54.690 align:middle line:90% an unknown state of the world. 01:10:54.690 --> 01:10:57.840 align:middle line:84% So neither of the firms knows how big the market 01:10:57.840 --> 01:10:58.830 align:middle line:90% is for their product. 01:10:58.830 --> 01:10:59.920 align:middle line:90% It's a new product. 01:10:59.920 --> 01:11:01.420 align:middle line:90% They've just both invented it. 01:11:01.420 --> 01:11:03.930 align:middle line:84% They start-- they produce some number of quantities, 01:11:03.930 --> 01:11:06.270 align:middle line:84% dump them on a Cournot market, and then there's 01:11:06.270 --> 01:11:09.710 align:middle line:84% some market-clearing price for the good. 01:11:09.710 --> 01:11:12.740 align:middle line:84% And let's assume that theta is unknown, 01:11:12.740 --> 01:11:14.530 align:middle line:90% and it has some common prior. 01:11:14.530 --> 01:11:19.010 align:middle line:84% If theta is distributed normal-- mean mu variance sigma squared. 01:11:19.010 --> 01:11:21.515 align:middle line:84% So it could be that demand for the products is very high. 01:11:21.515 --> 01:11:23.515 align:middle line:84% It could be demand for the products is very low. 01:11:23.515 --> 01:11:26.160 align:middle line:90% 01:11:26.160 --> 01:11:28.140 align:middle line:84% And so then on the observability, 01:11:28.140 --> 01:11:30.960 align:middle line:84% we're going to assume that after firms choose their first period 01:11:30.960 --> 01:11:34.540 align:middle line:84% quantities, they know what quantity they produced. 01:11:34.540 --> 01:11:37.590 align:middle line:84% They don't know what quantity their rival produced. 01:11:37.590 --> 01:11:42.310 align:middle line:84% But they do observe what the market-clearing price was. 01:11:42.310 --> 01:11:44.200 align:middle line:84% And so they observe market-clearing price, 01:11:44.200 --> 01:11:45.993 align:middle line:90% but not the rival's quantity. 01:11:45.993 --> 01:11:48.160 align:middle line:84% And then I'm going to assume that after they observe 01:11:48.160 --> 01:11:50.530 align:middle line:84% the market price, firm two has this this option to exit 01:11:50.530 --> 01:11:53.170 align:middle line:84% the market, and by scrapping its firm 01:11:53.170 --> 01:11:56.273 align:middle line:84% earn some value, K. And then if firm two-- 01:11:56.273 --> 01:11:57.940 align:middle line:84% depending on what firm two does, we then 01:11:57.940 --> 01:11:59.950 align:middle line:84% end up with firm two exiting in their monopoly 01:11:59.950 --> 01:12:03.222 align:middle line:84% in the second period, or firm two stays in, 01:12:03.222 --> 01:12:05.430 align:middle line:84% and there's duopoly competition in the second period. 01:12:05.430 --> 01:12:09.490 align:middle line:90% 01:12:09.490 --> 01:12:10.490 align:middle line:90% This is a misspelling. 01:12:10.490 --> 01:12:13.030 align:middle line:84% So the equilibrium is going to have 01:12:13.030 --> 01:12:16.600 align:middle line:84% q11 greater than the static best response to q21. 01:12:16.600 --> 01:12:19.780 align:middle line:84% So the idea is going to be in this model, 01:12:19.780 --> 01:12:22.540 align:middle line:84% firm one is going to what we call signal jam 01:12:22.540 --> 01:12:26.500 align:middle line:84% and overproduce in the first period driving down the market 01:12:26.500 --> 01:12:28.470 align:middle line:90% price. 01:12:28.470 --> 01:12:31.770 align:middle line:84% And in equilibrium-- firm two always 01:12:31.770 --> 01:12:34.090 align:middle line:84% knows what firm one is doing in equilibrium. 01:12:34.090 --> 01:12:38.130 align:middle line:84% So in equilibrium, firm two correctly anticipates q11. 01:12:38.130 --> 01:12:45.180 align:middle line:84% It can always then infer that it knows that theta minus q11 plus 01:12:45.180 --> 01:12:48.850 align:middle line:90% q21 equals P1. 01:12:48.850 --> 01:12:50.910 align:middle line:84% So it can always infer that theta 01:12:50.910 --> 01:12:57.450 align:middle line:90% equals P1 plus q11 plus q21. 01:12:57.450 --> 01:12:59.140 align:middle line:84% So I guess this is the way it infers it. 01:12:59.140 --> 01:13:00.900 align:middle line:84% It's like after the first period, 01:13:00.900 --> 01:13:03.270 align:middle line:90% it sees the first period price. 01:13:03.270 --> 01:13:05.670 align:middle line:84% It knows what quantity it produced. 01:13:05.670 --> 01:13:06.600 align:middle line:90% It anticipates. 01:13:06.600 --> 01:13:09.250 align:middle line:84% It knows the equilibrium, q11 star. 01:13:09.250 --> 01:13:10.800 align:middle line:84% And so it now believes that theta 01:13:10.800 --> 01:13:13.770 align:middle line:90% equals this with probability 1. 01:13:13.770 --> 01:13:19.900 align:middle line:84% And so result is that this equilibrium has 01:13:19.900 --> 01:13:24.060 align:middle line:84% to have overproduction at t equals 1 by firm one. 01:13:24.060 --> 01:13:24.970 align:middle line:90% Why is that? 01:13:24.970 --> 01:13:27.330 align:middle line:84% Well, because suppose that firm one 01:13:27.330 --> 01:13:31.760 align:middle line:84% was just playing the static best response to q21. 01:13:31.760 --> 01:13:34.970 align:middle line:84% Then firm one could increase its output with no first order 01:13:34.970 --> 01:13:36.950 align:middle line:90% loss and profits. 01:13:36.950 --> 01:13:39.470 align:middle line:84% And there's a one for one relationship 01:13:39.470 --> 01:13:42.530 align:middle line:84% between the more I produce in the first period 01:13:42.530 --> 01:13:48.055 align:middle line:84% and what firm two thinks theta is going to be. 01:13:48.055 --> 01:13:58.160 align:middle line:90% 01:13:58.160 --> 01:13:59.510 align:middle line:90% Because if I produce-- 01:13:59.510 --> 01:14:05.300 align:middle line:84% imagine that I produce q11 greater than q1 star. 01:14:05.300 --> 01:14:08.030 align:middle line:84% Firm two is thinking I produce q1 star. 01:14:08.030 --> 01:14:13.100 align:middle line:84% So when I produce q11, P1 gets driven down. 01:14:13.100 --> 01:14:15.920 align:middle line:84% And so when P1 gets driven down, P1 01:14:15.920 --> 01:14:18.740 align:middle line:84% is going to be actually the true P1. 01:14:18.740 --> 01:14:20.675 align:middle line:84% Well, it's going to be the-- sorry. 01:14:20.675 --> 01:14:24.890 align:middle line:84% What I'm going to get is their estimated theta, theta hat, 01:14:24.890 --> 01:14:31.400 align:middle line:84% is going to be the true theta minus q11 minus q1 star. 01:14:31.400 --> 01:14:39.390 align:middle line:84% So if I overproduce, firm two is going to miss estimate theta 2 01:14:39.390 --> 01:14:41.220 align:middle line:84% and think theta 2 is lower than it really 01:14:41.220 --> 01:14:43.980 align:middle line:84% is because the price was lower than it should 01:14:43.980 --> 01:14:45.850 align:middle line:90% have been given this q11 star. 01:14:45.850 --> 01:14:49.210 align:middle line:84% So firm two is going to think theta is lower. 01:14:49.210 --> 01:14:53.020 align:middle line:84% And so because firm two thinks theta is lower than it really 01:14:53.020 --> 01:14:55.210 align:middle line:84% is, firm two is going to be more likely to exit 01:14:55.210 --> 01:14:57.380 align:middle line:90% after the first period. 01:14:57.380 --> 01:14:59.540 align:middle line:84% But then if firm two is more likely to exit 01:14:59.540 --> 01:15:04.320 align:middle line:84% in the first period, that's good for me, and so I want to do it. 01:15:04.320 --> 01:15:06.670 align:middle line:84% So why would I not do this in equilibrium? 01:15:06.670 --> 01:15:08.670 align:middle line:84% The only reason I would not do it in equilibrium 01:15:08.670 --> 01:15:14.400 align:middle line:84% is choosing q11 greater than q1 star has a first order 01:15:14.400 --> 01:15:15.870 align:middle line:90% cost to me. 01:15:15.870 --> 01:15:18.370 align:middle line:84% And the only way it's going to have a first order cost to me 01:15:18.370 --> 01:15:22.493 align:middle line:84% is if q11 is greater than the static best response to q21. 01:15:22.493 --> 01:15:25.110 align:middle line:90% 01:15:25.110 --> 01:15:29.020 align:middle line:84% So what happened is we have this equilibrium where I overproduce. 01:15:29.020 --> 01:15:30.398 align:middle line:90% He knows I'm overproducing. 01:15:30.398 --> 01:15:32.190 align:middle line:84% He correctly infers what's going to happen, 01:15:32.190 --> 01:15:33.670 align:middle line:90% given the overproduction. 01:15:33.670 --> 01:15:36.660 align:middle line:84% But the overproduction by me is necessary to have me not 01:15:36.660 --> 01:15:39.600 align:middle line:84% have an incentive to overproduce by even more. 01:15:39.600 --> 01:15:41.970 align:middle line:84% And so we get this signal-jamming equilibrium 01:15:41.970 --> 01:15:45.767 align:middle line:84% where firm one produces more than it looks 01:15:45.767 --> 01:15:47.850 align:middle line:84% like the static best response in the first period, 01:15:47.850 --> 01:15:49.380 align:middle line:90% drives prices down. 01:15:49.380 --> 01:15:53.070 align:middle line:84% And it drives prices down in an attempt to get firm two to exit. 01:15:53.070 --> 01:15:55.230 align:middle line:84% And notice, there is a funny efficiency 01:15:55.230 --> 01:15:57.660 align:middle line:84% thing behind that story is in equilibrium, 01:15:57.660 --> 01:16:01.050 align:middle line:84% firm two anticipates theta correctly, 01:16:01.050 --> 01:16:03.270 align:middle line:84% undoes my signal jamming, and only 01:16:03.270 --> 01:16:05.990 align:middle line:84% exits when it would exit efficiently anyway. 01:16:05.990 --> 01:16:07.800 align:middle line:84% So this signal-jamming is, in some sense, 01:16:07.800 --> 01:16:09.870 align:middle line:84% good for social welfare relative to the world 01:16:09.870 --> 01:16:13.300 align:middle line:84% where all the cues were available-- were observable. 01:16:13.300 --> 01:16:15.330 align:middle line:84% Because if the cues were observable, 01:16:15.330 --> 01:16:17.230 align:middle line:84% you'd get exactly the same entry. 01:16:17.230 --> 01:16:18.690 align:middle line:84% Except here, we get this low price, 01:16:18.690 --> 01:16:20.190 align:middle line:84% and then we still get the same entry 01:16:20.190 --> 01:16:22.150 align:middle line:90% you'd get with observable cues. 01:16:22.150 --> 01:16:27.800 align:middle line:84% So this unobservability is just socially a good thing. 01:16:27.800 --> 01:16:31.010 align:middle line:84% I guess, one more classic paper-- 01:16:31.010 --> 01:16:33.410 align:middle line:90% Aghion and Bolton, 1987-- 01:16:33.410 --> 01:16:36.770 align:middle line:84% discusses the question of, can firms 01:16:36.770 --> 01:16:40.670 align:middle line:84% deter entry by getting their customers 01:16:40.670 --> 01:16:42.720 align:middle line:90% to sign long-term contracts? 01:16:42.720 --> 01:16:44.090 align:middle line:90% Like, if I could get-- 01:16:44.090 --> 01:16:46.730 align:middle line:84% if I'm trying to deter entry by rival cell phone 01:16:46.730 --> 01:16:49.453 align:middle line:84% companies in Boston, could I just get 01:16:49.453 --> 01:16:50.870 align:middle line:84% everyone who's currently in Boston 01:16:50.870 --> 01:16:53.780 align:middle line:84% to sign like, say, my cell phone contracts 01:16:53.780 --> 01:16:55.010 align:middle line:90% have a five-year commitment. 01:16:55.010 --> 01:16:56.130 align:middle line:90% You sign up with me. 01:16:56.130 --> 01:16:57.360 align:middle line:90% It's a five-year commitment. 01:16:57.360 --> 01:17:00.920 align:middle line:84% You have to pay me an early termination fee of $2,000 if you 01:17:00.920 --> 01:17:02.610 align:middle line:90% leave me before the five years. 01:17:02.610 --> 01:17:04.670 align:middle line:84% Can I use lockup long-term contracts 01:17:04.670 --> 01:17:08.270 align:middle line:84% like that to deter entry by rivals? 01:17:08.270 --> 01:17:12.320 align:middle line:84% And, again, the Chicago school argument was always, well, no. 01:17:12.320 --> 01:17:14.210 align:middle line:84% You can't use long-term contracts 01:17:14.210 --> 01:17:17.000 align:middle line:84% to deter entry because the customer wouldn't 01:17:17.000 --> 01:17:18.560 align:middle line:84% sign the long-term contract unless it 01:17:18.560 --> 01:17:20.190 align:middle line:90% was in their best interest. 01:17:20.190 --> 01:17:24.110 align:middle line:84% And therefore, we shouldn't think 01:17:24.110 --> 01:17:27.900 align:middle line:84% that firms can use long-term contracts to deter entry 01:17:27.900 --> 01:17:30.570 align:middle line:84% because the customers wouldn't sign the contracts unless it 01:17:30.570 --> 01:17:33.750 align:middle line:84% was in their interest to sign the contracts. 01:17:33.750 --> 01:17:36.450 align:middle line:90% Two problems with that story. 01:17:36.450 --> 01:17:42.930 align:middle line:84% One is the Chicago story may make sense for large buyers 01:17:42.930 --> 01:17:44.910 align:middle line:90% or monopoly buyers. 01:17:44.910 --> 01:17:46.650 align:middle line:84% But it certainly doesn't make sense 01:17:46.650 --> 01:17:48.510 align:middle line:84% in markets like the cell phone market, 01:17:48.510 --> 01:17:52.800 align:middle line:84% where you have a large number of infinitesimal consumers. 01:17:52.800 --> 01:17:59.040 align:middle line:84% Because you-- you graduate student at MIT, 01:17:59.040 --> 01:18:00.160 align:middle line:90% you walk into Verizon. 01:18:00.160 --> 01:18:02.130 align:middle line:84% They say, we only offer five-year contracts. 01:18:02.130 --> 01:18:02.850 align:middle line:84% You walk to T-Mobile, and they say, 01:18:02.850 --> 01:18:04.080 align:middle line:84% we only offer five-year contracts. 01:18:04.080 --> 01:18:05.190 align:middle line:90% So you're like, I'm not-- 01:18:05.190 --> 01:18:07.440 align:middle line:84% I don't even know I'm going to be here for five years. 01:18:07.440 --> 01:18:09.540 align:middle line:84% I'm not going to sign a five-year contract. 01:18:09.540 --> 01:18:11.040 align:middle line:84% Give me a one-year contract instead. 01:18:11.040 --> 01:18:12.720 align:middle line:90% They're going to say, fine. 01:18:12.720 --> 01:18:13.570 align:middle line:90% Don't buy a phone. 01:18:13.570 --> 01:18:14.280 align:middle line:90% And you're going to say. 01:18:14.280 --> 01:18:15.738 align:middle line:84% I want to buy the one-year contract 01:18:15.738 --> 01:18:19.050 align:middle line:84% because I know that if all of us buy one-year contracts, 01:18:19.050 --> 01:18:21.520 align:middle line:84% some third firm is going to enter the Boston market. 01:18:21.520 --> 01:18:23.610 align:middle line:84% Well, the answer is you, as a graduate student, 01:18:23.610 --> 01:18:24.880 align:middle line:90% that's not happening. 01:18:24.880 --> 01:18:26.610 align:middle line:84% Like, if there was a monopoly buyer 01:18:26.610 --> 01:18:29.460 align:middle line:84% who could commit to I'm only signing a one-year contract, 01:18:29.460 --> 01:18:32.080 align:middle line:84% then someone's going to enter, that would be better. 01:18:32.080 --> 01:18:36.720 align:middle line:84% But you-- if you know that everybody else is signing 01:18:36.720 --> 01:18:38.982 align:middle line:84% the five-year contract, you holding out and not 01:18:38.982 --> 01:18:41.440 align:middle line:84% signing the five-year contract is not going to do anything. 01:18:41.440 --> 01:18:42.780 align:middle line:90% It's not going to affect entry. 01:18:42.780 --> 01:18:44.790 align:middle line:90% So there's a free rider problem. 01:18:44.790 --> 01:18:47.640 align:middle line:84% Even if it's bad for all the consumers in aggregate, 01:18:47.640 --> 01:18:49.830 align:middle line:84% you could sign it because of the free rider problem. 01:18:49.830 --> 01:18:52.980 align:middle line:90% 01:18:52.980 --> 01:18:59.310 align:middle line:84% Aghion and Bolton note that even with a monopoly buyer, 01:18:59.310 --> 01:19:02.910 align:middle line:84% we can still tell stories where the long-term contracts get 01:19:02.910 --> 01:19:06.970 align:middle line:84% signed even though they're not optimal or not socially optimal. 01:19:06.970 --> 01:19:10.110 align:middle line:84% And the reason is that the long-term contracts 01:19:10.110 --> 01:19:13.080 align:middle line:84% are in some sense stealing money from firm 01:19:13.080 --> 01:19:14.710 align:middle line:90% two, the potential entrant. 01:19:14.710 --> 01:19:17.790 align:middle line:84% And you, in some sense, take that business stealing profit 01:19:17.790 --> 01:19:20.850 align:middle line:84% and divide it among the customer signing the contract 01:19:20.850 --> 01:19:22.350 align:middle line:90% and the initial monopolist. 01:19:22.350 --> 01:19:27.490 align:middle line:84% And by stealing the profits away from the future firm two, 01:19:27.490 --> 01:19:29.970 align:middle line:84% you can divide those up among and get a situation 01:19:29.970 --> 01:19:33.400 align:middle line:84% where it's bad socially to have these long-term tie-ups. 01:19:33.400 --> 01:19:36.300 align:middle line:84% But the monopolist wants to offer them, 01:19:36.300 --> 01:19:38.160 align:middle line:84% and the customers want to sign them. 01:19:38.160 --> 01:19:41.740 align:middle line:84% So anyway, I will not go through that argument. 01:19:41.740 --> 01:19:43.340 align:middle line:90% It's written there. 01:19:43.340 --> 01:19:44.170 align:middle line:90% I had one more. 01:19:44.170 --> 01:19:44.670 align:middle line:90% Anyway. 01:19:44.670 --> 01:19:46.410 align:middle line:84% So Wednesday, I'm going to go back 01:19:46.410 --> 01:19:49.230 align:middle line:84% to empirical papers on strategic investment. 01:19:49.230 --> 01:19:52.890 align:middle line:84% I think these three are the ones that I'm mostly 01:19:52.890 --> 01:19:55.240 align:middle line:90% going to talk about. 01:19:55.240 --> 01:19:55.930 align:middle line:90% OK. 01:19:55.930 --> 01:19:57.780 align:middle line:90% See you then. 01:19:57.780 --> 01:20:04.000 align:middle line:90%