YT YouTube2Text

YouTube2Text → Transcripts → ECON20110 Fall 2013 - Lecture 15: Competition Policy

ECON20110 Fall 2013 - Lecture 15: Competition Policy — Transcript

by Glen Weyl · 10,687 words · 2,041 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00um
0:02so we left off yesterday by talking
0:04about the real versus pecuniary
0:08um
0:09effects that our firms can induce by
0:13changing their strategy and we said the
0:14real effects is that they can
0:17change the industry's quantity
0:19and that affects profits by the amount
0:21of the markup right if you increase your
0:23quantity for every unit of markup you
0:25make more money but you can also change
0:28the prices which affects
0:30uh industry profits by the change in
0:32prices multiplied by quantity
0:35okay now
0:37uh which of these effects would a social
0:41planner want to take into account um
0:45uh and andrew's not here
0:47uh anybody else wanna
0:50say yeah julia so plans you care about
0:52the real life reality especially because
0:54cuny externality is actually benefiting
0:57consumers yeah so they'll want to care
0:59only about the real so here it's not
1:01just externalities i'm taking all the
1:02effects into account but in general
1:04they're going to take into account all
1:05those real effects right and so what
1:08would they want to try to equate to xero
1:11trillion so
1:13they want
1:14to
1:15make
1:16um the markup zero
1:19well yeah the more general way of
1:20putting that is they want to make the
1:22real effect zero so they want to say
1:24that set that equal to zero right
1:27and how about a monopolist or a cartel
1:31uh austin's not here
1:33um
1:34anybody else want to say what
1:36a monopolist or a profit maximizing
1:38cartel would do yeah rishi all right
1:41what is that both the real answer
1:43the sum of the two of them right
1:46exactly
1:48um
1:52and so we can define
1:54a quantity
1:56called the conduct parameter theta
1:59which is equal to the ratio
2:02of the
2:03real
2:05to the pecuniary effects
2:08and under perfect competition or social
2:10optimality what would this be equal to
2:13uh
2:15anybody
2:16yeah
2:17well because the markup would be zero
2:21yeah theta would be zero that's right
2:24um and under monopoly what would this be
2:26said equal to uh
2:29somebody else
2:30uh yeah rishi oh from the previous
2:33section
2:43um
2:44and
2:45more generally
2:47this thing is going to be equal to the
2:50ratio of the markup to
2:53negative of quantity times dq dp
2:57which you can rearrange by just bringing
3:00this q over here and then dividing the
3:02top and the dividing the bottom and
3:04multiplying the top by p and you get the
3:06elasticity so you get the markup
3:10relative to price times the elasticity
3:13right which is the here is the learner
3:15index and under monopoly learner index
3:17is equal to one over the elasticity but
3:19more generally this is a measure of the
3:21degree of market power yeah rishi oh
3:23when you say lead conduct
3:25all right
3:27what exactly do you mean by
3:29conduct
3:30so this is just a a parameter
3:33which captures the degree so remember we
3:36had all these complicated things in fact
3:38i think you came to office hours to talk
3:40about this we had all these complicated
3:41things that could lead firms to have
3:43more market power effectively it could
3:45be that there are fewer firms or more
3:47differentiated products etcetera this is
3:49one number that summarizes all of that
3:53and it summarizes it through the extent
3:55to which the pecuniary versus real
3:58effects are taken into account
4:01and so in some sense you can think of
4:03everything we've studied as a way of
4:05getting a handle on what this number is
4:08okay
4:09so let me talk about how the specific
4:11models we've been talking or let's talk
4:13about how the specific models we've been
4:14going through
4:15fit into this
4:17um can someone tell me what the conduct
4:19parameter is equal to under cornell
4:23competition
4:28because remember
4:30that
4:31this thing is equal to theta
4:34and do you remember what the markup
4:36relative
4:37to price
4:38was equal to
4:40under corno
4:47well
4:49remember that under cornell price minus
4:51marginal cost is equal to negative p
4:54prime times the q of that individual
4:56firm which is 1 over n of the total
4:58quantity
4:59and therefore theta is equal to 1 over n
5:05um
5:11so if we add in conjectures which i
5:15talked to you guys about informally
5:17right where the
5:20amount of quantity of another for of the
5:22other firms might change when you change
5:23your quantity
5:25that's going to increase
5:27the amount
5:29that
5:30firms expect the total quantity to react
5:33to a change in their quantity
5:35and so the conduct parameter
5:38is going to change depending on the so
5:40if i expect that other firms are going
5:42to increase their quantity when i
5:43increase mine they'll retaliate against
5:45me in that sense then this conduct will
5:48get higher that's like a way of doing
5:50collusion right on the other hand if i
5:53anticipate that when i increase my
5:54quantity other people reduce their
5:56quantity
5:58this will actually become
6:01this will be a negative number and in
6:03fact if this is equal to negative one
6:06then we get to
6:08effectively bertrand competition or
6:10theta equals zero
6:12because if i expect
6:14that um other people hold their price
6:16fixed that means that if i increase my
6:18quantity people other people reduce
6:20their
6:21quantity to perfectly offset mine
6:25under differentiated bertrand
6:29this
6:30is equal uh price
6:33minus
6:34marginal cost over price is equal to the
6:37inverse
6:38of the
6:39residual elasticity right which is the
6:42derivative of which comes out of the
6:44derivative of your quantity with respect
6:47to your own price
6:49right
6:51which we can express in terms of the
6:52derivative of the total quantity with
6:55respect to the total price
6:58times
6:59one uh so we can multiply the bottom and
7:03the top of this by that and then we can
7:04have a ratio
7:06of the derivative of the total quantity
7:08with respect to everybody's price
7:11changing
7:12which will be divided by one over n
7:15because
7:18your quantity is only 1 over n of the
7:21total quantity
7:22and so what this turns into is theta
7:25being equal to
7:27how much the total quantity changes when
7:30the price changes divided by n times how
7:32much your price changes your quantity
7:35changes when your price changes you can
7:37then through a series of
7:39calculations that i i won't go through
7:41in too much detail
7:43i derived this is equal to one minus
7:45what's called the aggregate diversion
7:47ratio
7:48so the aggregate diversion ratio is when
7:51i sell one more unit of the good
7:55how many units
7:56does everybody else
7:59in the industry lose
8:01so if i lower my price that's going to
8:04give me some more sales but give
8:06everybody else less sales
8:07and the question is what fraction of the
8:09sales that i gained
8:11came from other people's sales versus
8:13coming from outside goods goods outside
8:15of the market that's the diversion ratio
8:18and this theta is going to be 1 minus
8:21that now this that's very natural
8:22because we said that if goods are very
8:24good substitutes for one another that is
8:26my marginal sale comes overwhelmingly
8:28from other people's goods right then
8:30we're going to have very close tight
8:32competition on the other hand if my
8:35sales mostly come from outside goods
8:36because the goods are not very
8:38competitive with one another then d will
8:40be low and theta will be high will be
8:42near monopoly
8:45okay
8:48and this
8:49theta which depends on the diversion
8:51ratio is not typically constant it
8:53actually increases with price
8:55rather than being constant as it is in
8:57these other cases
8:59okay
9:00so that's it for that lecture now we're
9:01caught up
9:06let's move on to
9:08uh
9:09the next lecture
9:11where we're going to
9:13um
9:17where we're gonna apply these ideas
9:20okay
9:21uh to a very active area of economic
9:24policy and one that i've been quite uh
9:26involved with
9:29which is the control of mergers
9:32so we're going to start by looking at
9:35stability in the cornea model and how
9:37that leads to the comparative statics
9:41with respect to the number of firms
9:43and the benefits of entry then we'll
9:46talk about how competition will drive
9:48prices down to cost
9:50as well as more general ways in which
9:52some notion of competition can drive
9:54prices down to cost and we'll
9:56investigate some empirical analysis
9:59of
10:00the rate at which
10:02more competition will drive
10:04prices down to cost
10:07we'll then
10:09we'll then talk about
10:11mergers in differentiated products
10:13industries
10:15and the concept of upward pricing
10:17pressure
10:19and then we'll use that to talk about
10:22how mergers are actually evaluated in
10:24practice
10:26in the u.s and around the world
10:29as well as
10:31some other common
10:33approaches to merger analysis
10:35and some of the challenges that merger
10:39policy and analysis face
10:41going forward
10:42okay
10:44so remember that we said that
10:46comparative statics go haywire if an
10:49equilibrium is not stable
10:52and now i want to talk about what's the
10:55right notion of stability in cornell
10:56equilibrium
10:58well remember our reaction functions
11:00remember how i did a little stability
11:02thing in those diagrams right and if
11:04those crossed in the right direction
11:06then we had stability
11:08um and let me just take you through that
11:10logic imagine one firm increases their
11:12quantity q j by a little bit
11:17uh how much
11:20sol how much will the quantity
11:24of firm i rise by or fall by when
11:29quantity of firm j increases um
11:32do you have the reaction function um
11:34well let's relate it to the pass-through
11:36stuff that we're talking about
11:38do you remember the notion of quantity
11:40pass-through
11:42not entirely
11:43so quantity passed through is if one
11:45firm
11:47that's like other than the firm in the
11:49market
11:50uh increases its quantity by some amount
11:53how much does the total quantity
11:55supplied to the industry increase and we
11:57showed that that was equal to the cost
11:59price passed through as well because
12:00much faster okay so if we just call that
12:02the pass through so how would we figure
12:05out how much
12:07if we have two firms if one firm
12:09increases its quantity how much will the
12:10remaining firm increase we have the pass
12:12through but we have the cross price
12:14faster yeah cool so um
12:17if
12:18so okay
12:19if uj increases a certain amount then
12:21you have
12:22the passer of qj yeah that's how much
12:24the q rises
12:26that's how much that's what's the total
12:28one so it would be um
12:32um that divided by like
12:35and
12:36no it's just that minus one because
12:38because the thing is q j is increasing
12:41right
12:42and then how much is q i gonna increase
12:44well the total of q j and q i is going
12:47to increase by pass through okay
12:50and then you have to subtract the one
12:52which is the amount that q i increased
12:54the q q j increased right
12:58um
13:01which is
13:02oh
13:03this is even easier sorry uh
13:07well i mean this is also the slope of
13:08the reaction function
13:10right
13:12and
13:13if
13:14qi rises
13:16as a result of qj rising how much is
13:20that in turn going to cause qi uh to
13:24change uh christian
13:26just in terms of the reaction functions
13:27how would you express that
13:29um
13:31would it be the slope of the reaction
13:33function of j
13:34yeah exactly
13:38so
13:39the total amount
13:41is going to be the slope of the reaction
13:43function of i to j's change
13:46times
13:47j's reaction to i
13:49so it's the product of those two things
13:51yeah mike um does this assume that uh
13:53originally qj increased by one because
13:55well yeah one times epsilon you know
13:57it's like a standard differentiation
14:00thing
14:01okay
14:02so um
14:04julian can you see under what
14:06circumstances this is going to converge
14:08and under what circumstances it's going
14:10to go nuts
14:11um
14:12so
14:13it should be um
14:17if
14:18okay assuming that r a prime r j prime
14:22are both
14:23like
14:24well they can't be the same
14:26um
14:27well nonetheless
14:29r i prime times our j prime
14:32is greater than q j or less than q j
14:34yeah less than one right
14:37yeah yeah exactly whatever exactly
14:40very small yeah so if these are greater
14:42than one then you're going to keep
14:44running through this process and it's
14:45going to keep getting bigger and bigger
14:47and bigger each time
14:48right
14:49and if the product is less than one then
14:52it will converge
14:53right
14:55um
14:57now it turns out that the monopolies
14:59second order condition that the marginal
15:01revenue declines relative to marginal
15:03cost is always enough to ensure this
15:06but in general you just want to check
15:07when you're solving problems that you
15:09have stability in this sense that the
15:14that uh things don't spiral uh out of
15:16control
15:19okay
15:20so um
15:22one firm's increasing their production
15:26should always lead the total production
15:28to rise
15:30um
15:31and
15:32the amount that it will cause it to rise
15:35is the derivative of firm i's reaction
15:38function plus 1 right because there's
15:40the direct effect
15:41plus the reaction function
15:44right
15:45and
15:47this is the same for both firms
15:51because it's symmetric
15:53and this number is the quantity passed
15:55through which we i think we talked about
15:57before maybe we didn't talk about before
15:59did we talk about it in the earlier
16:01lecture
16:02no
16:03no it didn't okay we didn't talk about
16:05it so this is the quantity passed
16:07through so i was trying to extract
16:09uh flood from sue hell about something i
16:10haven't even taught you guys yet
16:14so
16:14this says if one firm's quantity
16:16increases how much does the total
16:18quantity increase that's the quantity
16:20passed through um
16:23and the reason it's called that is
16:24because it turns out this is equal to
16:26the cost price pass-through
16:29which i uh
16:30uh
16:31won't go through but uh the paper just
16:34got
16:35one of the papers in the references
16:36discussed is this
16:39so
16:40it turns out that if you like run
16:42through all those chains of stuff
16:46stability ends up being that the
16:48quantity passed through is less than n
16:50over n minus 1
16:55and
16:57the way you can see that
17:00is
17:03that you can uh
17:07trace through the logic so you can say
17:09imagine that there are
17:11two firms in the industry and now we
17:14bring in a third one
17:15and that third one raises its quantity
17:18well how much is that going to increase
17:20the total quantity
17:22of the industry
17:24uh
17:25well
17:26each firm
17:28is going to increase its quantity
17:30by symmetry by one half of the total
17:33amount that it increases minus one
17:36because
17:37one is the amount that the third firm
17:39increased it and the other two firms are
17:41going to increase by half of that
17:42difference or decrease
17:44to be consistent
17:46this has to be
17:48the pass-through
17:51the total amount that it increases has
17:53to be the pass-through
17:57of
17:58the amount that the other two
18:01that one other firm changes it
18:08plus
18:10the amount that directly the third firm
18:12changes it okay why is that so
18:15i'm trying to get down to one firm being
18:17isolated right so
18:19there's the one firm externally that's
18:22increasing its quantity
18:25then that's leading
18:28the individual the the second firm to
18:32increase their quantity by this much and
18:35now from the perspective of the third
18:36firm i'm going to pass that through at
18:39the quantity pass-through
18:40rate okay so then you can use this to
18:43solve out for things
18:45uh
18:47bringing this to the other side of the
18:48equation
18:50and solving for rho q and you get that
18:53the change in the quantity from a change
18:56in an individual firm
19:01in an individual firm is the quantity
19:03passed through over two minus the
19:05quantity passed through and more
19:07generally it's going to be equal to the
19:08quantity passed through over n over n
19:11minus 1 times the quantity passed
19:12through
19:13and
19:16from that you can derive
19:18that
19:19things will not go crazy
19:22as long as this inequality is satisfied
19:26and most importantly
19:27this
19:29uh is going to be positive anytime you
19:32have stability
19:33so that says that one firm increasing
19:35its quantity always increases the total
19:38quantity supplied to the market
19:42was what happened there that a firm
19:44enters industry and as a result of
19:47like those other firms increased by one
19:49half times dpd q minus one
19:52and then as a result of that increase
19:54like the other things respond again
19:57uh yeah so that's how you would trace
19:58through the stability argument yeah
20:03okay
20:04so
20:05um
20:07if a new firm enters the industry it's
20:09going to start with a quantity of zero
20:11and it's going to raise its production
20:12from that point
20:16and
20:18this is going to
20:20increase the total quantity of the
20:23industry by exactly the same argument we
20:25just gave right if a new firm comes in
20:28and they raise their production from
20:29zero when they're you know initially
20:31there to uh to some higher quantity this
20:35has to raise total production and that
20:37increase in total production is going to
20:39drive down
20:41the um price in the industry
20:44and thus
20:45adding this new firm makes the industry
20:48effectively more competitive by driving
20:51prices closer to cost right
20:54and this is a more general feature
20:56as the number of firms increase the
20:58industry is going to become more
20:59competitive and in fact it's eventually
21:01going to get to perfect competition and
21:03i want to take you through a simple
21:05mathematical example of that like the
21:08ones that we've been doing so again
21:10imagine we have constant elasticity
21:11symmetric
21:13constant marginal cost
21:15we know each firm is going to set price
21:17plus quantity times derivative price is
21:20equal to cost
21:22so by symmetry all the firms are going
21:24to have 1 over n of the total quantity
21:28and um
21:30rishi can you convert this into an
21:32expression in terms of elasticities sure
21:36um so
21:38do you should i go for the board or
21:40should i just
21:41well i think yeah it should be pretty
21:44good p prime q over n yeah you can find
21:46it as uh dpdq yeah and you've got dpdq
21:50over
21:52uh
21:53oh actually so if you factor out the key
21:56yeah right you've got one plus
21:58p
22:00uh
22:01e p d q
22:02q over n yeah so that comes out to the
22:04because the p is factor out of your one
22:06plus
22:08elasticity over no one over yep this is
22:11exactly what we had before right one
22:13minus the but now you've got a one over
22:15n right because it's negative yeah
22:19exactly
22:20and so that gives us that the markup
22:23relative to price is equal to one over n
22:25times the elasticity if the elasticity
22:27is constant that means that the markup
22:29goes to zero as n gets large
22:31right
22:33um
22:34oh
22:35i was going to ask this let me get this
22:36one
22:37uh
22:38yeah so as n gets large the right hand
22:40side goes down to zero
22:42right
22:43and so competition drives prices down to
22:45costs and as a result quantities become
22:48efficient yeah what does that say
22:51the number of firms
22:55okay just to hammer this point home
22:57let's consider another example imagine
22:59that we have linear demand of this form
23:04then
23:05sue l
23:06what's the equilibrium equation in that
23:09case
23:10sure um remember it's
23:12given by this okay so just marginal cost
23:15equal to um
23:16marginalized
23:17exactly okay sure so
23:19um if you have q p over there
23:22then um
23:24you would take the derivative
23:26with
23:27quantity so i guess you'd have to find
23:29um
23:30p of q so you invert that yeah
23:33yeah sure right you
23:48and there's an eraser too okay
23:51so um
23:52and
23:53okay
23:55so
23:56demand is equal to sorry
24:01q e is equal to sigma times 1
24:06minus
24:07p over 2m
24:09um yeah so just inverting that i'm just
24:12going to call this quantity so
24:14just the cube to make it easier than
24:15isolate for this so um
24:18i'm isolating for price yeah
24:20q times or q over sigma
24:23yeah
24:24um
24:26plus p over 2m
24:31price is p of q
24:33so um
24:35now i multiply that by plus p of
24:37sorry p
24:38um
24:41right
24:45p you brought p over to m on to the
24:47other side right but then you've got one
24:50times sigma but i just okay um so i had
24:53q i um divided by sigma
24:56and i
24:57um yeah
24:58you're very close yeah yeah i just screw
25:01that up all right
25:02just remember okay
25:03i'm going on the non-skip step all right
25:06so
25:07um
25:10yeah
25:11um and i always so minus one
25:16yeah
25:18and i just multiplied everything through
25:19by and then we multiply it by two m yeah
25:22is that a negative two
25:28yeah so it's going to be 1 minus q over
25:30sigma right
25:34times 2 m times t
25:38that's equal to p
25:40good okay okay so now p what's p prime
25:44okay so p prime is well this is just
25:46sorry
25:47um
25:49okay
25:50am i finding p prime or p uh
25:53well you need p that's important right
25:56but you need p prime or oh you could
25:58yeah you could just multiply that by q
25:59and find the marginal revenue yeah
26:00that'd be fine right
26:02um all right so but the thing is that we
26:04we're doing it for cournot so it's
26:06actually better if you find p prime
26:08because in corneau right remember the
26:10equilibrium condition
26:12is not marginal revenue equals marginal
26:15cost it's p plus p prime times q divided
26:17by n is equal to cost sure okay
26:20uh oh right i mean well yeah that's
26:22easier so well in that case p prime is
26:24just um a negative two item over sigma
26:27great in which case um well i think
26:30what is equal to so it's p plus um p
26:33times q over f prime squared yeah
26:37so
26:38that's just so p is um
26:41so two m minus
26:44u over sigma two m yeah that's p
26:48um
26:50plus so e prime is
26:53negative two for sigma
27:00times quantity
27:02yeah
27:06yeah
27:08and that's equal to cost
27:10yeah
27:12okay can you try simplifying that down
27:14and solving for q alright
27:17um
27:19all right
27:21okay
27:23so
27:25what am i isolating for how am i
27:27for q we're trying to find the
27:29equilibrium quantity
27:33brutal yeah
27:35okay so
27:36i'm gonna multiply everything through my
27:38hands i feel like that's gonna make it
27:39slightly easier
27:41just
27:41i'm going in the wrong direction on the
27:43next
27:51if you just group all the q terms
27:52together it'll actually be pretty simple
27:56because you see that q stuff that's like
27:58all
27:592m over 2m over sigma
28:02it's like got a negative in front it's
28:04got a 2m over sigma and then it's like 1
28:06plus 1 over n right yeah so that's not
28:08so bad that's not that well all right
28:10so um
28:15q
28:17and
28:18i'm negative sigma 2m so if i just take
28:21out a negative q
28:22um i have a
28:242l over sigma
28:29plus times one over one plus one over n
28:34sorry i just i took out this term plus
28:37um this term right without a q just
28:39that's
28:41two of them over
28:42and seconds yeah the thing is you've got
28:43two m over sigma already in there you
28:45see what i mean so you can then just
28:46multiply that by one plus one over n
28:49oh okay so you i just factor that out as
28:51well right sure
28:52all right
28:54and then on the other side um
28:56because there's a 2m over here plus 2m
28:58is equal to cost yeah so um
29:02q is equal to
29:03minus 2m
29:05over um
29:07well you've got a negative so you can
29:08just make it 2m over minus cost right
29:12yeah
29:19that over
29:26nicely done
29:32um
29:34exactly
29:36okay so you solved that all
29:38out uh you get exactly what 0l had
29:42and in particular
29:44as n goes to infinity here notice that
29:46this becomes 2m minus c over 2m times
29:49sigma
29:51which is exactly what happens when you
29:52set this equal to cost if you put cost
29:55there right
29:56you're going to get
29:58it's okay 2m minus c over 2m times sigma
30:01we can bring the sigma to the top and
30:03then we get 1 minus c over 2m which is
30:06exactly what would happen if we said
30:08plugged in cost for price here so that
30:10means that price must be going to cost
30:12because quantity is going to what would
30:16equate a quantity to that level
30:18okay
30:20so again as the number of firms increase
30:22price is driven to cost
30:24and this suggests a much more general
30:26result so as the number of competing
30:28firms increases
30:32grows arbitrarily large price converges
30:34to marginal cost this is one of the most
30:36fundamental results in economics it's
30:38sort of the foundation of the idea of
30:41perfect competition
30:44and it's the basis of most antitrust and
30:46competition policy which aims to try to
30:49keep the number of firms competing
30:52high enough
30:54now i only did some very specific
30:56examples but this result holds even when
30:58firms have nonlinear costs are
31:00asymmetric
31:01as long as the demand is stable and it
31:04can satisfy weaker regularity conditions
31:06the key is just that everyone gets
31:07sufficiently small
31:09so that their markup is small uh and it
31:13fails only in silly or perverse uh cases
31:15yeah i don't know by like marginal cost
31:17is like marginal cost of like the
31:19average firmware like like the lowest
31:21marginal cost in the industry
31:24uh
31:25the
31:26as okay so you have to in the case that
31:28firms have different costs
31:30you have to think about when you're
31:32adding more firms what types of firms
31:34are you adding
31:36right
31:37so if you add a bunch of firms all of
31:39which have high cost
31:41then you're not going to get
31:43perfect competition
31:45if you add firms that have sort of like
31:48a representative or average cost of all
31:50the firms in the industry or are
31:52randomly drawn well then you'll
31:53eventually get enough firms that have
31:55the lowest cost and you'll all and those
31:57will be the only ones that end up
31:58producing
32:01okay
32:02so let me actually go through an example
32:04that's a little bit like what adam just
32:06said
32:08so suppose we have bertrand competition
32:10but there's a distribution of costs some
32:12firms have high costs in terms of lower
32:14cost
32:16turns out that this represents um
32:19something very similar to an auction so
32:21what firms are going to do is they're
32:23going to set their price equal to their
32:25best guess of the cost of the second
32:28this or the whatever every firm is going
32:30to set their price equal to just a
32:32little bit below what they think the
32:35cost of the next most costly firm is
32:37going to have
32:38assuming that they're the
32:40uh least costly firm because that's the
32:42only time they're gonna get to produce
32:43is if they're the least costly firm
32:45right and they're gonna try to just beat
32:47the next guy the same way you would in
32:48an auction
32:49right
32:50um
32:52and so
32:54um what happens as the number of firms
32:57grow large is that there's a lot of
33:00firms
33:01and so there's go there's very likely to
33:03be some
33:04someone who's very very close to whoever
33:07is the least costly and that drives
33:09prices down to the cost of the least
33:11costly guy because he feels people
33:13breathing down his neck really hard and
33:15he feels the need to bid very close to
33:17his cost
33:18right
33:19so this is another example whereas the
33:21number of firms grow large for totally
33:23different reasons
33:25prices converge to cost
33:27so this is a very general result so
33:31if
33:36if there were not constant marginal
33:38costs you'd need to take into account
33:41more effects but again you would get
33:42something that was very similar yeah all
33:44right um but
33:46if the if the price converge to like the
33:48lowest marginal cost
33:50wouldn't that drive out all the other
33:51firms yes so like but it won't drive out
33:54one firm but
33:56well we know in cornell competition i
33:57mean in portraying competition there
33:59might just be one firm
34:01like if one firm has looked a little bit
34:03lower cost than the other one they'll
34:04settle price just below the other guy to
34:07make sure they get all the business
34:09right and it'll be just constrained by
34:11that other guy's cost so in general
34:13there will just be one or maybe two
34:16firms competing
34:18but
34:19that's fine as long as he's held down to
34:22his uh cost yeah he's held down because
34:24if you raise the higher the other guy
34:26comes in
34:27exactly yeah maureen um what happens in
34:30this case if average cost is above
34:32marginal cost
34:35uh
34:36in other words if you have declining
34:38marginal costs yeah
34:40so
34:41that's a natural monopoly case yeah
34:45and that can be a little bit tricky
34:47because the problem is like other firms
34:50but still another firm if they had a
34:53lower cost function than yours they
34:55could still come in offer a price that's
34:57lower
34:58than your
35:01current average
35:02costs okay so firms in that case get
35:05forced to price at average cost not at
35:07marginal cost
35:09because if they were to price at
35:10marginal cost they take a loss
35:13but they can't price
35:15above average cost because if they do
35:17price above average cost someone else
35:19will have
35:20and undercut their current price take
35:23all of the market and still earn profits
35:26so in that case firms are driven to
35:28average cost pricing not to marginal
35:30cost pricing and that's one reason why
35:32like when we were studying regulation i
35:33was saying that often competition can
35:35create average cost pricing
35:37so that's an example
35:39yeah
35:40that doesn't that's not like
35:42a reasonable application of the theorem
35:44right because it says constant marginal
35:46cost yeah
35:47yeah
35:48or increasing it works as well
35:51okay
35:52so this is a general rule about
35:54competition eventually but not
35:55immediately
35:57competition will drive prices to cost it
35:59doesn't immediately do it in either of
36:01these cases because it takes a while
36:03until there's like enough firms to fill
36:04things in
36:05and a natural question is how much
36:08competition is enough
36:10so
36:10an interesting paper by bresnahan and
36:14race tried to get at this question
36:16so um
36:19they
36:20noticed that you know
36:23as markets grow larger they're able to
36:25support more firms if you go to a big
36:27city there will be a lot more
36:28restaurants than a small town right
36:31so
36:32they looked at a bunch of different
36:33professions and they said suppose we're
36:35going to have n firms
36:37how many people do we need to have in a
36:39town before it's going to have n firms
36:41let's call the number of people needed
36:43to support n firms capital s sub n
36:47and then we can construct little s of n
36:50as the number of people per firm that's
36:52needed to support end firms now you
36:55could think that number should be
36:56constant right
36:58but that's not going to be the case if
36:59competition drives down the price
37:01right because if competition is driving
37:03down the price then people are going to
37:05make less profits when there are more
37:08firms and therefore you'll need more
37:09people to support a particular firm
37:12right
37:14and
37:17as a result you should expect that the
37:18number of people per firm needed to
37:20support three firms is greater than the
37:22number of people per firm needed to
37:24support two firms is greater than the
37:25number of people per firm needed to
37:27support one firm and so forth right
37:30now there's other reasons why you could
37:32get patterns that are like this or that
37:34even violate this
37:35because it depends a lot on you know it
37:38could be that once there's already one
37:40person in town it's easier to set up
37:42shop because you can share supplies or
37:44it could be
37:46that uh
37:47you know the people who most love living
37:49in that town are willing to work there
37:50for less
37:52than etc etc so it's not completely
37:55obvious
37:56that this is all driven only by what's
37:58happening to prices but maybe that's a
38:00reasonable place to start
38:02and um
38:04so what they did is they found five
38:06professions that show up in many rural
38:08towns and they looked at how many people
38:10there were in each of these towns and
38:11how many people there were in each of
38:13the professions
38:14and they tried to look for the point at
38:16which s starts to level out as an
38:19indication of how many firms it takes
38:21before prices stop dropping anymore and
38:24therefore we've sort of got into perfect
38:25competition
38:28okay
38:29so here's the picture
38:32so the five professions are druggists
38:35tire dealers doctors dentists and
38:37plumbers
38:39and
38:40you'll you see that um
38:42so this is s sub 5 divided by s sub n
38:45with n running from 0 to 5. so you see s
38:49sub 1
38:50is way way lower than s sub 5.
38:54and it starts to get closer and most of
38:57the gap has been closed by the time you
38:59get to three firms
39:01essentially all has been closed by the
39:03time you get to four firms and there's
39:04like really no difference
39:06uh from five firms so what they
39:08concluded from this
39:10is that about three to five firms is
39:13like enough to get perfect competition
39:19okay
39:20so um
39:22this is uh
39:24a model
39:26that we've been talking about where we
39:27thought that there's just one product
39:29there's one industry and there's like a
39:31bunch of people you know supplying this
39:34uh product
39:37but um as we talked about in the last
39:39class it's much more realistic in many
39:41cases that
39:43firms produce imperfect substitutes
39:46for one another not perfect substitutes
39:49and so a natural question is how does
39:51the merging
39:53of two firms
39:56or the removal of one of the products
39:58affect prices
40:04so
40:05when we have differentiated prices and
40:06there's multiple different products
40:09what competition means is a little
40:11subtle because when a firm comes in to
40:13compete it not only like
40:16increases or decreases competition but
40:18actually makes new products available to
40:19the customers right
40:21and so the simplest way to think about
40:23this is holding fix the set of prices
40:25just allowing two of the firms to merge
40:27with one another
40:29and uh i want to consider the simplest
40:31model of a situation like this imagine
40:34we have firms each of whom has a linear
40:36demand that is their price their
40:39quantity decreases when they increase
40:41their price
40:42and the other
40:44and when the other firm increases their
40:45price uh that firm gets more demand
40:50right okay
40:52um
40:55so let's consider what the optimal price
40:58for each firm is
41:00when they both have a cost of c
41:03and beta is greater than gamma uh joan
41:06can you see how to solve out for that
41:26okay so the this top equation is telling
41:29you how much demand you get when your
41:31price is this and the other guy's price
41:33is that
41:35and this is linear demand remember
41:38so
41:39do you remember the rule for how firms
41:42price under linear demand
41:44to maximize profits
41:56well because you don't know what your
41:58quantity is you have to choose your
41:59quantity and your price optimally to
42:01maximize profits right
42:09do you remember what i'm what price in
42:11monopolist with no cost charges
42:19a monopoly a profit maximizing monopoly
42:24okay
42:26yeah halfway up the demand curve right
42:29so halfway to the point where there
42:30would be no demand so if this guy had no
42:32cost
42:34what price would he charge
42:35what what price does it take before he
42:37has no demand
42:56yeah that's right so that would be his
42:58price if his costs were equal to zero
43:01and then what's the pass-through of
43:03someone who has
43:05um
43:07who has a linear demand
43:19one half
43:21so it's one half so you would you would
43:23say
43:24okay
43:25so he's gonna set
43:27his price
43:29equal to alpha
43:31plus
43:33uh gamma times pj
43:35divided by beta
43:38plus one half of cost
43:42and by symmetry
43:45uh
43:46the other firm is going to do the same
43:48thing
43:52um joan do you want to come up and try
43:55to write that down and solve it
43:57okay so his price
44:00is equal to
44:02alpha
44:03plus gamma pj
44:08so pi equals
44:19i'm sort of trying to teach you guys
44:20this through these tricks rather than
44:22just doing all the math out so you can
44:23use those on the exam and it'll speed
44:25things up
44:28so pi
44:31equals alpha plus gamma pj
44:37over beta
44:40plus
44:41c over 2.
44:44now
44:44by symmetry p and i and pj have to be
44:47the same so you can just erase the
44:49subscripts underneath them
44:52and now you can solve out
44:54for what p has to be equal to
45:06so
45:23okay
45:27um if there are imperfect substitutes
45:30why is it necessarily have to be the
45:32prices or the progress
45:35because i assume that they're imperfect
45:38substitutes but symmetric
45:44this is called symmetrically
45:46differentiated bertrand competition
46:08nice job
46:09perfect
46:18okay
46:20um
46:30okay so these are the prices before the
46:33firms merge
46:36if we take what joan did here
46:40and
46:41we
46:47um
46:48beta c
46:50plus alpha
46:52over 2 beta times
46:55beta minus
46:57theta cancels
47:00hmm
47:00i think i might have made an example
47:02here
47:03mistake i think this should have been
47:04two times beta minus
47:06gamma unless someone thinks
47:10oh
47:11wait i had her go all the way up the
47:13demand curve
47:14uh it really should have been halfway up
47:17up that right yeah yeah yeah yeah okay
47:21sorry so you just divide them yeah so
47:23anyways it becomes beta c plus alpha
47:26over two beta minus gamma
47:29now after
47:30the firms merge they're going to treat
47:32the two products as one product
47:34and therefore
47:36because by symmetry they're the same
47:38right
47:39and so they're just going to have now a
47:41quantity of 2 times alpha minus
47:44beta minus gamma times p
47:49and
47:50mori
47:51can you again use the pricing halfway up
47:53the demand curve idea
47:56to figure out what the optimal price is
47:58uh now
48:00sure
48:01so you just want to
48:03invert it
48:05or get the price point
48:08uh yeah you want price where the
48:10quantity equals zero right and you want
48:12to go halfway to that
48:14right so what's that going to be
48:17here
48:18um
48:20sure
48:57the reason it's beta minus gamma there
48:58guys right is that there's when you
49:00change the price there's two effects you
49:01both get the fall in your demand by beta
49:04and you get the increase from the other
49:05guys
49:06by
49:07gamma because we're increasing the two
49:09prices by the same amount at the same
49:11time
49:26good let's not make the mistake of going
49:28all the way up now so we only have half
49:30of that right
49:33and then we have to add half of cost
49:35right so you
49:36because uh the pass-through rate under
49:39linear demand is one-half right so plus
49:41plus c over two
49:45and now you can make a common
49:47denominator of those things right by
49:49multiplying the c over 2 on top bottom
49:52by beta minus gamma
49:56could you explain why we hadn't seen it
50:00because uh this
50:03the other one
50:04is what a monopolist with zero cost
50:08would do
50:10and then we know that there is a
50:11constant pass-through rate of one-half
50:13so we just in have however much the cost
50:16is he's going to charge that times
50:18one-half more than what someone with
50:20zero cost would charge
50:22this is like the trick for solving
50:24monopoly problems
50:26with uh linear cost
50:29nicely done maury
50:31um
50:33okay
50:34so that gives us
50:36uh this
50:39okay so if you take the difference of
50:40these two things by map which i did on
50:42mathematica it becomes one half
50:45times gamma over beta minus gamma
50:49times
50:50alpha minus beta minus gamma c over 2
50:53beta minus gamma
50:54okay so why did i break this out in this
50:57particular way
50:58well there's basically
51:00three terms here
51:02so the first one is alpha over beta
51:05minus alpha
51:07which is a measure of how intensely
51:09these two firms are competing
51:12uh can somebody describe to me why
51:14that's the case yeah phillip it
51:16approaches
51:18data approaches
51:20and what happens is beta approaches
51:22gamma
51:23uh as beta approaches gamma beta minus
51:25gamma approaches zero so yeah but why
51:28does that measure inten the intensity of
51:30computation as beta
51:32approaches gamma the products become
51:35close to perfect substitutes as one's
51:38sales are being pretty much diapered on
51:40there yeah so this is actually exactly
51:42that diversion ratio we were talking
51:44about before right because this is how
51:47many
51:48sales when i lower my price do i get
51:49from the other good
51:51and this is how many sales do i get in
51:53total
51:54right
51:56um
51:57and
51:58uh so that's exactly our diversion ratio
52:01which we said was a natural measure of
52:03competition
52:05in this differentiated products case
52:07right
52:08okay
52:10this
52:10thing here
52:12is the markup that the firm had before
52:14the merger
52:16can someone explain to me why that's the
52:18case uh mike alvaren
52:21um
52:24so here's the price that you had before
52:25the merger what's going to be your
52:27markup over talks
52:34what is the markup
52:36it's uh
52:38the
52:46price minus fast right
52:48yeah so what is price minus cost it'd be
52:52um
52:54p
52:54times two beta minus gamma minus beta c
52:59plus alpha over two beta minus gamma
53:03or like yeah so the price is this right
53:06sorry and then if we subtract cost from
53:09that
53:10what do we get
53:12um
53:14you would multiply cost by two beta
53:16minus gamma yeah and then put all that
53:18over two minutes and that's what gives
53:20us this right yeah good
53:22um
53:25uh yeah subtract
53:27c or c over two
53:29minus c because we're interested in the
53:31markup
53:32okay not
53:34not how much they're going to price yeah
53:36yeah
53:37okay so this term was the diversion
53:39ratio
53:41this term was the markup and then the
53:43final part is the pass-through rate
53:45which is one-half so that's exactly what
53:48this change in price is equal to
53:52now it turns out that those three
53:54factors that i just broke out for you
53:55are what matter in general
53:58so
54:00let's focus on
54:03the
54:05diversion ratio and the markup
54:08first and let me try to explain why
54:10those should be what matters yeah evan
54:12why is pass-through one happen
54:14under linear demand well we went through
54:16that in the monopoly lecture i mean it's
54:18just a consequence of linear demand and
54:20the fact that the marginal revenue
54:22diverges at a constant rate under linear
54:24demand from the so that's only getting
54:26one half of the demand it's only one
54:27half with linear demand you're not
54:29saying that it's generally no it's just
54:30this happens to be because of linear
54:32demand the pass-through rate but let me
54:35try to go through the logic of why in
54:36general the this this type of thing is
54:39what's going on okay so these three
54:42factors show up more generally okay so
54:44imagine that burger king buys mcdonald's
54:46right
54:48now
54:49uh why would this lead
54:51burger king to want to increase its
54:53prices evan just intuitively
54:55less competition
54:57okay but uh
54:59let's go a little bit more into the
55:01mechanics so like what
55:04um
55:05you know
55:06what
55:08gain does burger king get out of raising
55:10the prices on burger king burgers
55:13well
55:13they get more money and people who
55:16otherwise would have gone to mcdonald's
55:18if they were deciding between the two
55:20now only have burger king to go to
55:23no but imagine that mcdonald's stays
55:25open but it's just owned by the people
55:26who who own burger king yeah because
55:28then people will go to mcdonald's and
55:30you'll profit from the burgers that they
55:32buy at mcdonald's which you weren't
55:33profiting from before yes right so now
55:36every time you sell a burger at burger
55:39king there's an opportunity cost
55:40associated with that
55:42a partial burger or a probability of a
55:45burger bought at mcdonald's which you
55:46could have made a profit on
55:48right
55:49um
55:50and so
55:52what is the
55:54um
55:56uh what is that
55:58well the number of burgers that i would
56:00have sold at mcdonald's had i not sold
56:02the burger at burger king that's exactly
56:04the diversion ratio we're talking about
56:06right
56:08and
56:09then what's the marginal profit that i
56:11lose on that well it's whatever my
56:12markup was
56:14right so that markup multiplied by the
56:16diversion ratio is exactly what i gain
56:21by failing to make a sale at burger king
56:23or what i lose every time i make a sale
56:26at burger king the cost which i wasn't
56:28taking into account before we merged but
56:30i'm now taking into account now that we
56:31have merged right
56:34so this new opportunity cost that i face
56:37is
56:38the product of the diversion from burger
56:40king to mcdonald's
56:42times the markup at mcdonald's which is
56:44also called the upward pricing pressure
56:47on burger king and that's exactly what
56:49we derived in these two factors here
56:51right this was the markup that i was
56:53earning
56:54and this is the diversion ratio
56:58okay
57:01um
57:03okay so the upward pricing pressure is
57:05the diversion times the markup
57:08and this is the opportunity cost that's
57:10created by the merger yeah maury uh i
57:12had a question about the merger yeah if
57:14you think of like the merger like 13
57:17mcdonald's one firm afterwards is it
57:20kind of a form of price discrimination
57:22to raise the price of the burger at
57:24burger king
57:27like if you think people at burger king
57:28well we could have done the same
57:30exercise for mcdonald's for the moment i
57:32was thinking of them as being sort of
57:34like coke and pepsi symmetric one is not
57:36like better than the other it's just
57:39that
57:40you know people like
57:42uh one better than the other okay that's
57:44sort of how i was thinking of that but i
57:46guess you were thinking working slightly
57:47higher quality or something like yeah
57:49yeah
57:50yeah
57:51yeah evan so
57:53now there's an opportunity cost if uh
57:55burger king owns mcdonald's not selling
57:57uh for every bird or sold
58:00so why would they not increase the price
58:01uh
58:02they will as well they'll increase both
58:04of the prices so this this logic will
58:06apply symmetrically at both in both
58:08places
58:10yeah also takes into account the fact
58:11that people may
58:13not buy it either
58:16yeah because the diversion ratio is not
58:17one
58:19diversion ratio is not one means that if
58:21someone stops buying burger king they
58:23may go and buy some some third thing
58:25that's outside
58:26now
58:28we had that gives us two of these terms
58:30that gives us the diversion ratio and
58:31the markup but why is the pass-through
58:34rate relevant well
58:37um we know that increased cost we just
58:39said this is an opportunity cost of a
58:41sale right
58:43and
58:44what is the rate at which an increased
58:46cost translates into an increase in
58:48price well that's exactly the
58:50pass-through rate right that's the
58:52definition of the pass-through rate
58:54okay so this is the reason i keep
58:55reiterating the pass-through rate is
58:57because it comes up in all these
58:58contacts and you think a merger what
59:00does that have to do with the
59:01pass-through rate but of course when we
59:03think about it really all the merger
59:05does is create this opportunity cost
59:06which then gets passed through at the
59:08pass-through rate
59:09uh yes so
59:10opportunity cost depends on the previous
59:12price
59:14depends on the previous price and it
59:15also depends on the diversion ratio
59:18but you wouldn't take an account like
59:20like the post post merger changing the
59:22markup
59:24uh it turns out that's not what you want
59:25to do yeah
59:28the markup's constant
59:30no you don't assume that the markup is
59:32constant
59:33it's just that um
59:35it's very hard to measure where the
59:37markup is going to go to
59:39and so the right way to think about
59:41approximating it which is exactly right
59:43in a linear case is you look at the
59:45pre-merger markup times the diversion
59:47ratio and that's the opportunity cost
59:49that you sort of get hit with
59:51and then you multiply that by the
59:52pass-through rate and the pass-through
59:54rate already takes into account the fact
59:56that the markup's changing
59:58because the pass-through rate i mean
59:59think about what the pass-through rate
1:00:01is like if the markup was constant
1:00:03pass-through rate would always be equal
1:00:04to one if the markup decreases as the
1:00:07price increases then the pass-through is
1:00:08going to be less than one so the
1:00:10pass-through rate is already taking into
1:00:11account the fact that the markup changes
1:00:14as the
1:00:15um
1:00:16as the price changes
1:00:20yeah
1:00:21so is that only stable if the path is
1:00:23less than one no
1:00:26it can be stable it's only stable
1:00:28capacitors less than infinity
1:00:31so
1:00:32yeah
1:00:33okay so if you multiply by upward
1:00:36pricing pressure by the pass-through
1:00:37rate this gives you the change in prices
1:00:40and that's exactly why we got this thing
1:00:42up there okay so now imagine we want to
1:00:45get the effect of this on consumer
1:00:47welfare
1:00:48uh
1:00:49can anyone tell me what we'd have to do
1:00:51with it
1:01:02by multiplying the price change by the
1:01:03quantities exactly
1:01:05great tenor
1:01:07exactly right
1:01:08because
1:01:10why
1:01:12because the uh that measures the loss of
1:01:13consumers
1:01:15it's by the envelope theorem right
1:01:16because
1:01:17uh like the
1:01:20if you think of the consumer surplus
1:01:22it's that
1:01:23little
1:01:24rectangle
1:01:25which is the change in price multiplied
1:01:28by how much they're consuming yeah so um
1:01:31is this snickerleading neglecting the
1:01:33consumers old part of the deadweight
1:01:35loss triangle that's now opened up yeah
1:01:38so that that's going to be second order
1:01:39so as long as these price changes are
1:01:41small
1:01:42okay
1:01:43okay
1:01:45so this gives a relatively simple
1:01:46methodology for predicting the effects
1:01:48of mergers so to make a prediction you
1:01:50need to get internal documents or
1:01:53like some
1:01:54studies or surveys or internet data
1:01:57which indicate to you something about
1:01:58the diversion ratio
1:02:00you have to measure their markup
1:02:02and then you need to know something
1:02:03about what the pass-through rates are
1:02:05and basically most of what the
1:02:06anti-trust agencies do i would say 15 of
1:02:10it is they have economists trying to get
1:02:12documents like this and try to estimate
1:02:13these different numbers
1:02:16okay so what's the institutional
1:02:17framework that lets them do that well in
1:02:20the u.s
1:02:21we talked about the sherman antitrust
1:02:23act
1:02:24last class
1:02:25which bans firms from making illegal
1:02:29combinations in the restraint of trade
1:02:31but
1:02:32um pretty soon after the sherman act
1:02:35people realized well if you do that then
1:02:37they'll just merge with each other and
1:02:38then it'll be perfectly legal for them
1:02:40to collude and so they created something
1:02:42called the clayton act which says that
1:02:44firms before they're allowed to merge
1:02:45have to submit it as a proposal to the
1:02:47government and the government can
1:02:48approve or reject the firm's right to
1:02:50merge
1:02:51now why wouldn't the government just
1:02:53block all mergers if mergers tend to
1:02:55increase prices mike fosco
1:02:57you have recounted as a scale yeah
1:02:59because there can be economies of scale
1:03:01the government doesn't want to stop
1:03:02firms from realizing those economies of
1:03:04scale which might even benefit consumers
1:03:06right but if the two firms are more
1:03:08efficient together
1:03:09even a monopoly might charge a lower
1:03:10price than the separated firms would
1:03:12charge so like a classic example of this
1:03:15is um a t
1:03:18claimed that if it merged with t-mobile
1:03:20then you'd be able to get reception at
1:03:22more places in the country and this
1:03:23would make consumers better off right
1:03:26so therefore the agencies have to weigh
1:03:28the benefits through those economies of
1:03:31scale against the anti-competitive
1:03:34effects that tend to raise prices
1:03:36okay so there are many agencies that
1:03:38have a responsibility for looking into
1:03:41these issues but there are two primary
1:03:42agencies that share responsibility for
1:03:44this
1:03:45um
1:03:46in rishi what are those two agencies
1:03:50from the
1:03:51okay
1:03:58yes ftc is one of them and
1:04:02uh
1:04:04it's not part of justice it is
1:04:06anti-trust division of the department of
1:04:07justice
1:04:09so um
1:04:11and any company in the us that wants to
1:04:13merge that's above a certain size has to
1:04:16pre-notify these agencies of their
1:04:19intentions
1:04:20and then the agency's subpoena data in
1:04:23several stages so first they like ask
1:04:25for some basic information if it looks
1:04:27fishy then if it looks fine then they
1:04:30just say go ahead with your merger if it
1:04:31looks fishy then they ask for more data
1:04:34and eventually
1:04:37they
1:04:38might take it to court or they might ask
1:04:40the companies not to merge and say that
1:04:43they will take it to court if they
1:04:44insist
1:04:45or ask them to do some things like
1:04:48spinning off parts of the companies that
1:04:49are competing
1:04:51etc
1:04:52um they have a team of about 150 phd
1:04:54economists
1:04:56who are in charge of analyzing these
1:04:58issues okay
1:05:00this is in some ways perhaps the most
1:05:02sophisticated
1:05:04economic regulation infrastructure in
1:05:05the world and it's been copied by most
1:05:07developing developed countries so the
1:05:10european union is now a major leader in
1:05:12this area the uk australia new zealand
1:05:15all these countries have infrastructures
1:05:17for doing this type of merger review
1:05:19it's much more unstructured and
1:05:20primitive in the developing world in
1:05:23some countries there's no law against
1:05:25mergers in fact so
1:05:27peru
1:05:28improve this was true but we helped
1:05:31pass merger regulation there a few years
1:05:33ago
1:05:35in others
1:05:36there's a law but there's no
1:05:37infrastructure to really enforce
1:05:40that law so it's technically illegal but
1:05:43nobody actually is there to review
1:05:45others have some review process but they
1:05:48have very little economic guidance
1:05:50and a few countries have gone and really
1:05:53in the developing world really
1:05:54established some real infrastructures
1:05:56like china brazil chile etc
1:05:59now this is very heterogeneous across
1:06:01the world and in my view like you know
1:06:03there's a lot of work in the united
1:06:05states on anti-trust policies um
1:06:09for like
1:06:11much more complicated things than
1:06:12mergers like are you allowed to make
1:06:15some sort of a contract
1:06:17that might force someone to buy from you
1:06:20and so forth and those are
1:06:23relevant issues but i think even more
1:06:25important is thinking about what are the
1:06:28impact on these countries that don't
1:06:29have any antitrust uh you know
1:06:32infrastructure of adding an anti-trust
1:06:34infrastructure yeah this is applied to
1:06:36horizontal
1:06:38as well or is it just this is all
1:06:39horizontal oh sorry both of those
1:06:42so vertical mergers are a very different
1:06:43issue we'll talk about those next class
1:06:45actually those might actually be
1:06:47beneficial in some ways
1:06:49because of complementarities and and so
1:06:52forth so we'll talk about that uh those
1:06:54are quite different and they're governed
1:06:55under very different infrastructure
1:06:57okay
1:06:59so um traditional evaluation
1:07:03uh
1:07:03what
1:07:04of these types of mergers was based on
1:07:07industry concentration
1:07:09uh and that was a sort of strange system
1:07:12where you had to treat the industry as
1:07:15if it was just one coherent product
1:07:18right
1:07:19so
1:07:19um
1:07:21this was quite an awkward process
1:07:24because you had to define what the
1:07:26market was that you were thinking about
1:07:29so there's a famous um
1:07:32ex okay so what they did
1:07:34would be that they would say okay these
1:07:36two companies are merging let's say
1:07:39blockbuster and redbox are going to
1:07:41merge
1:07:42right
1:07:43and the question is what is the market
1:07:45that they're in
1:07:47well you might say they're in the market
1:07:49of
1:07:50movie rentals in physical locations
1:07:54but then people might say well that's
1:07:55not big enough of a market because you
1:07:57know really they're also competing with
1:07:58netflix and so then you could say all
1:08:00movie rentals and so there would be a
1:08:02giant debate over is it physical movie
1:08:04rentals or all movie rentals that's the
1:08:06market
1:08:07and the way that they would try to
1:08:08settle these debates is they would start
1:08:10with the
1:08:12two products that were merging they
1:08:14would then add products that were the
1:08:15closest substitutes for those products
1:08:18until
1:08:20and that was both geographic
1:08:22and
1:08:23in terms of the type of product so for
1:08:24example if you had two hospitals merging
1:08:27uh they would add hospitals that were a
1:08:30little further away from those hospitals
1:08:32right
1:08:33until they reached a market where the um
1:08:37person anyone who were to control that
1:08:39whole market
1:08:40would uh as a monopolist would impose
1:08:44what they called a small but significant
1:08:47and non-transitory increase in prices or
1:08:50a snip
1:08:51which is usually defined as five percent
1:08:54price increase for at least one year
1:08:57then within that market they would
1:08:59measure industry concentration
1:09:02um
1:09:03and um can anyone tell me how they
1:09:06measured industry concentration
1:09:08uh
1:09:09yeah mike fosco
1:09:12what's that um measure how many firms
1:09:14are in there
1:09:15um and i believe that the max is ten
1:09:17thousand and then they have a four for
1:09:19if it increases by specific amount there
1:09:21they're getting four yeah
1:09:23do you remember what the formula for
1:09:24calculating the hyphen doll index is i
1:09:26do not know
1:09:28yeah so it's but that that's right so
1:09:29it's um
1:09:31it's the
1:09:32you define the share of each firm which
1:09:35is their quantity as a func a fraction
1:09:38of the total quantity and you do the sum
1:09:40of the square of this multiplied by ten
1:09:43thousand
1:09:44and then basically ten thousand divided
1:09:46by h is sort of an approximation of the
1:09:48number of firms now from the
1:09:51breslin and race exercise we should sort
1:09:53of start getting worried if there are
1:09:56less than three to five firms
1:09:58this is also related to
1:10:02stigler's notion of
1:10:03collusion and under the cornell model
1:10:06because the markup is equal to your
1:10:09price times your quantity and your price
1:10:11is equal
1:10:13and this is the same as price over the
1:10:15elasticity times your share
1:10:18you can get that this
1:10:20the share weighted markup of firms is
1:10:23equal to the perfect all index over the
1:10:25elasticity
1:10:28now the thresholds that the government
1:10:30set up from this we would say oh we
1:10:32don't really need to worry until this is
1:10:34about equal to 2000 right because
1:10:36there's approximately five firms in the
1:10:38industry
1:10:39they set out as threshold for worrying
1:10:42there was a little bit less than that it
1:10:43was about 1800 or when uh mergers
1:10:47increased the hyphen doll index by 300.
1:10:51so that was that used to be the standard
1:10:53now the problem is this was sort of a
1:10:55mess
1:10:56in practice
1:10:57why was that
1:10:59well market definition was the zero one
1:11:01thing you know is the market blockbuster
1:11:05and redbox or just the physical delivery
1:11:08ones maybe including hollywood videos or
1:11:10does it include netflix the
1:11:13blockbuster and red box probably have
1:11:14like an 85 share in physical movie
1:11:16rentals and so that would be a huge
1:11:19problem but if you put netflix they
1:11:21probably only have a 10 share and then
1:11:23it wouldn't be a problem at all
1:11:25right um
1:11:27and that just seems like the wrong
1:11:30question in some sense right this is not
1:11:32like
1:11:33you know there's one market there are
1:11:35many different markets the real question
1:11:37is are these two products really good
1:11:39substitutes for one another are they the
1:11:41best competitors for each other right
1:11:44and that's the sort of question that
1:11:45diversion leads you to ask right if mark
1:11:49if uh
1:11:50blockbuster increases its prices by a
1:11:52bit probably the overwhelming majority
1:11:54of people who stop using blockbuster are
1:11:56going to start using redbox rather than
1:11:58using netflix as the next one right
1:12:00because if there's people who want to
1:12:02get it physically right
1:12:04and so
1:12:05um merger guy i guess blockbuster is
1:12:08bankrupt now so it's not even relevant
1:12:11but
1:12:12the
1:12:13uh merger guidelines
1:12:16uh
1:12:17issued the most recent ones issued by
1:12:18the agencies moved away from this
1:12:21so basically the guidelines of the issue
1:12:23about how they evaluate mergers to give
1:12:25firms like a sense of whether their
1:12:27emergent is going to get blocked or not
1:12:29they've had these since the early 1980s
1:12:33and until
1:12:34the most recent revision they did
1:12:36revisions about every 10 years
1:12:38it was all focused on market definition
1:12:41but they rewrote it in 2010 and i was
1:12:43pretty closely involved with that
1:12:47and
1:12:48the new ones say that don't even worry
1:12:51about defining markets just calculate
1:12:53this diversion ratio
1:12:55maybe calculate the markups and the
1:12:58um
1:12:59and the pass-through rates and try to
1:13:01predict how much the prices will go up
1:13:02as a result
1:13:04the 1992 or the 2010 2010 one
1:13:08it talks a little bit about the market
1:13:09definition stuff but you saw there's
1:13:11more emphasis on the diversion
1:13:14and yeah
1:13:15so these were designed by carl shapiro
1:13:17and joe farrell who were obama's uh
1:13:22campaign contributors and advisors
1:13:25during the 2008 election and when he got
1:13:27came into office they were made the
1:13:29heads of the economics groups of the ftc
1:13:31and the doj respectively and they were
1:13:34friends of mine and and so we sort of
1:13:36worked on this a little bit together now
1:13:38this stuff
1:13:40is very much
1:13:42in progress
1:13:44and
1:13:45there will be new guidelines again in
1:13:48you know maybe
1:13:49uh five or six more years and these will
1:13:52reflect i think uh some of the
1:13:55additional
1:13:56uh concerns that will come up in
1:13:58economics uh in the time in between then
1:14:01so in particular there are many
1:14:03important concerns that are left out of
1:14:05the guidelines canner can you think of
1:14:08what some of those might be
1:14:10whether or not the merchant is
1:14:12more efficient well so those are
1:14:14included they talked about efficiencies
1:14:16can you think of some others
1:14:18uh whether it's fair
1:14:20fair
1:14:21equity what is fair mean whether it be
1:14:24consumers well but i mean that's what
1:14:26we're trying to figure out right whether
1:14:27it's going to like raise prices
1:14:29another one is uh
1:14:31oh free entry
1:14:32yeah so entry by new firms that's not as
1:14:37much of a focus
1:14:38um
1:14:40effects on non-price things other than
1:14:42prices like the quality of the products
1:14:44is another issue right
1:14:47effects on innovation
1:14:50like we talked about in the intellectual
1:14:51property class
1:14:55uh these are all things that are
1:14:56mentioned in the guidelines
1:14:58but um
1:15:01uh but are not a focus and there's many
1:15:03issues that don't show up in the
1:15:04guidelines at all so for example we
1:15:06talked about how pass-through plays an
1:15:07important role that that's not mentioned
1:15:09anywhere in the guidelines
1:15:11so we talked about selection in the
1:15:13regulation class we didn't uh really
1:15:15talk about that with competition but we
1:15:17know that monopoly could actually be
1:15:18better than uh perfect competition under
1:15:22advantageous selection and so it seems
1:15:24logical that a merger could actually be
1:15:26beneficial even if it raised prices in
1:15:28that context right
1:15:29um
1:15:30some of these long-term effects
1:15:33uh like how
1:15:35people gain profits
1:15:37uh
1:15:38okay so we said that you know
1:15:40companies that merge those they're going
1:15:42to be able to increase their profits
1:15:44that'll encourage more people to be like
1:15:46the types of companies that are going to
1:15:48end up merging right so that might
1:15:50encourage some people to enter the
1:15:52industry and discourage other people
1:15:53from entering the industry so that's
1:15:55probably an important thing to take into
1:15:56account that the guidelines don't think
1:15:58about at all
1:15:59merging could make firms too big to fail
1:16:02right and that might be a problematic
1:16:05issue
1:16:08there's some issues about informational
1:16:09coordination which you guys can look at
1:16:13on your problem set
1:16:14in an old problem set we talked about
1:16:16political power that firms might gain
1:16:17from merging because they might be able
1:16:19to lobby the government more effectively
1:16:21if they're bigger right
1:16:23and many of these things aren't included
1:16:25in the merger guidelines currently and i
1:16:28think that they're very important issues
1:16:29maybe as important
1:16:31as the
1:16:32issues that are considered in the
1:16:34guidelines and so i think these are
1:16:35always interesting things to think about
1:16:37for theses or research so

About this transcript

This page contains the full transcript of ECON20110 Fall 2013 - Lecture 15: Competition Policy by Glen Weyl, generated from the public captions YouTube serves with the video. The transcript has 10,687 words across 2,041 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.