YT YouTube2Text

YouTube2Text → Transcripts → Lecture 8: Industrial Policy with Reka Juhasz

Lecture 8: Industrial Policy with Reka Juhasz — Transcript

by International Growth Centre · 16,254 words · 2,544 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:04foreign
1:00foreign
1:36um welcome everyone thank you very much
1:37for joining us on this the last uh
1:40lecture in the uh we're at igc uh firms
1:43and development course reminder that all
1:45the previous lectures uh as well as
1:47slides are posted on the website so
1:49please uh uh visit that uh if you want
1:53to catch up on anything or revisit them
1:55in the future
1:57um we're kind of thrilled everyone's
1:59managed to uh or so many of you have
2:01managed to stick around and and join us
2:03all the way through uh what has been a
2:05fabulous course and uh today uh we're
2:09going to end with uh a bang and have a
2:11record you have a talk about industrial
2:14policy uh Records very much a leader in
2:17this field has done a lot of the path
2:20breaking work uh uh in recent years on
2:23Industrial policy and has a kind of a
2:26whole new research agenda on on
2:28measuring industrial policy that I'm
2:30sure she'll talk about uh today and I
2:31mentioned briefly in my first lecture so
2:33we're very cited for her to talk today
2:37this fits very nicely as a bookend to
2:40the course a lot of what we've been
2:42talking about uh up to now are various
2:46distortions that lead to misallocation
2:48one of the most prominent being
2:50externalities of various sorts and today
2:53we'll be thinking about kind of optimal
2:55policies in order to rectify that how
2:57those have worked historically
2:59and uh where to go from there so I am
3:04extremely excited uh to hear this
3:06lecture today so take it away uh Wrecker
3:10okay so good morning everyone or good
3:12afternoon
3:13um thanks very much for logging on
3:17um this is going to be
3:18uh
3:20electron industrial policy but but sort
3:24of just a flag up front that um it's way
3:26too large of a topic to fit into 90
3:28minutes so I've decided to narrow it
3:30down to the topic that I think fits um
3:33best into a course on firms and
3:36development and that's going to be the
3:37empirics of the infant industry argument
3:40so okay
3:43so where I'd like to start today is to
3:47just give you
3:48a definition of
3:51um
3:52of industrial policy
3:54so the way I like to think about it IP
3:57is intentional State action directed at
4:01changing the structure of the domestic
4:03economy and there's sort of two aspects
4:06to this that I that I'd like to call out
4:08one is that I think it's really useful
4:11to think about industrial policy as
4:13something that is goal oriented and in
4:16particular it is it has something to do
4:18with the government's long-term vision
4:20for where they want the economy to be
4:23for what the structure of the domestic
4:25economy should be in the long run so you
4:28know the the classic case is that
4:30historically developing countries have
4:32used industrial uh industrial policy as
4:35kind of the name suggests to achieve
4:37industrialization and structural
4:39transformation there's a lot of Ip these
4:42days that has the long-term goal of um
4:45of of achieving the green energy
4:47transition
4:48okay the second aspect
4:52um the second aspect of this that I'd
4:53like to call out is that we typically
4:55think of industrial policies as being
4:57selective in some sense in the sense
4:59that they're not sort of targeting all
5:01firms in the economy in some way or or
5:04they're not building kind of
5:06infrastructure that has the potential to
5:07affect
5:08um all firms so these are sort of
5:10selective policies the literature often
5:12refers to these as
5:15um vertical policies
5:16um because because they're they're often
5:18targeted at specific sectors
5:20I tend to like the The Selective policy
5:24better because I think there are
5:25policies that very much fit industrial
5:27policies such as r d policy
5:30um export promotion policy
5:32um green energy transition uh green
5:34energy promotion policies that aren't
5:35necessarily sort of sectorally specific
5:38but are selective activities nonetheless
5:42okay so among this broad class of
5:45policies that fits industrial policy
5:48today I'd like to spend a bit of time
5:50thinking about one strand of the
5:53literature which is the infant industry
5:55argument
5:57so I think this is maybe a helpful time
6:01to try to make the link to how this fits
6:04into this course more generally
6:07so so Dave and David started their first
6:10lecture with with with kind of the
6:12unsurprising question for a development
6:13course
6:14um that asks why some countries are rich
6:17and some countries are poor and the the
6:20infant industry argument sort of gives a
6:22very sharp prediction for this so
6:24through the lens of the infant industry
6:27argument developing poor countries are
6:29stuck in a bad equilibrium
6:31that is under free trade they specialize
6:35in low growth sectors
6:37so
6:38um sort of in in in in in in Dave's kind
6:41of organization of how we think about
6:42about how we think about these things
6:44actually both things are going wrong the
6:46economy doesn't produce the right things
6:48resources are being allocated to the
6:50wrong activity
6:51and they also don't produce them very
6:53well so productivity is also um
6:55relatively low
6:57Now sort of the the good news on the
6:59optimistic news is that government
7:02intervention can move the country from
7:04this bad equilibrium to a good
7:06equilibrium basically by changing what
7:09is produced so by by by
7:11implementing policy that will um that
7:15will shift the composition of economic
7:16activity in some way and this in our
7:20models kind of magically is also going
7:23to change how well it is produced so
7:25it's also going to improve productivity
7:27okay and I'm going to give you sort of a
7:29toy model in the next couple of slides
7:31to to fix ideas and I hope that this
7:33intuition is going to become really
7:35clear
7:36the key prediction is that in this world
7:41certain industries that are not
7:43competitive today at free trade prices
7:45can become competitive in the long run
7:48if they're given temporary Trade
7:51protection
7:52we typically think of this as being
7:55um uh so historically we've thought of
7:57this the simplest case of this is is an
7:59import tariff that that that's going to
8:01produce a protected internal domestic
8:04Market but there's going to be a whole
8:06host of sort of other policies
8:09um that that we'll try to achieve and
8:11that can achieve uh the same thing
8:14okay so um this is probably not the
8:17first time you're hearing about infant
8:18industry protection this is an extremely
8:21controversial idea
8:24it's also not a new idea so infant
8:26industry promotion and protection is an
8:28idea that's been around
8:30um for at least 200 years uh 200 plus
8:33years now
8:35um and and and here I'm giving you one
8:36example of sort of why it is often
8:40deemed to be very controversial
8:43um what you see on this picture is sort
8:45of the metaphorical infant industry
8:47labeled as protected privilege they're
8:50clinging to their nurse
8:52um that is labeled High tariff
8:54um and they're clinging to the nurse
8:55because they're being sort of offered
8:57the bottle of open market competition
9:00and so this already sort of suggests one
9:03of the big kind of issues with infant
9:05industry promotion and protection which
9:08is that that many people often rightly
9:10see it as as basically
9:13um sort of code for for for the
9:16government conducting all sorts of
9:17special interest
9:18um politics
9:21so
9:23where I'd like to start today is is it's
9:26just to sort of paints you
9:29um at least my view of what kind of the
9:33debate where the debate on this stands
9:35both in economics and in the social
9:39sciences more generally
9:41so there's a very influential
9:44um and I think actually quite compelling
9:46argument mostly outside of Economics
9:50um that's that's that's being put
9:51forward by
9:53um disciplines in other social sciences
9:55and and historians that IP works that
9:58industrial policy works and not only
10:00does it work but that it is necessary to
10:03facilitate the type of structural
10:05transformation
10:07um that that many developing countries
10:09are are seeking
10:12so here I'm giving you sort of one
10:14flavor of this argument this is from a
10:16popular economics book called How Asia
10:18Works um it's it's I think a very
10:20influential book many people have read
10:22it and and they're sort of Joe studwell
10:24makes the the claim that I think is
10:26shared by by many that you know all of
10:30today's rich countries basically with
10:32with a handful of kind of um
10:34idiosyncratic exceptions
10:36have been guilty in their formative
10:39stages so during the process of
10:41industrialization of protectionism
10:44um and and he makes the claim that there
10:46are no economies in the world that have
10:48developed to the first strength through
10:50policies of free trade
10:52so
10:53I think that that
10:55um with that that statement is actually
10:57um broadly true you know it is it is it
11:01is well known that um in the 19th
11:03century both the US and Germany
11:05industrialized behind High tariff walls
11:08and other forms of Ip and we we we're
11:11also coming to understand to a a an
11:14increasing extent that many of the sort
11:16of canonical famous growth miracles of
11:18the 20th century Japan South Korea
11:21Taiwan and most recently China have
11:23featured
11:24a lot of industrial policy
11:26so of course of course this is a
11:28correlation right so we're saying that
11:30that when these when these countries
11:31were industrializing they were doing a
11:34lot of they were doing a lot of
11:35interventions in their economy that can
11:38be thought of as industrial policy
11:39that's not yet a causal statement
11:42I think using the tools of other social
11:44sciences if you read sort of the the
11:46literature
11:48in my reading they make a very
11:49compelling case with their methods that
11:52there's that there's a causal link here
11:54um although this is not something that I
11:56think we in the economics literature
11:58have have have have tested as rigorously
12:00as as we could
12:02um using our sort of different and
12:04complementary methods
12:06but even if you sort of Buy in so even
12:09if we sort of put problems of causality
12:11aside I think that the the more obvious
12:14issue with this literature and and and
12:16you know this is something that the
12:18literature is well aware of is that it's
12:20the the story isn't as easy as you know
12:23countries that did IP succeeded and the
12:25ones that didn't did not we know that
12:28many many other countries also
12:30experimented with industrial policy with
12:33often spectacular failures
12:36um and more often with with sort of
12:37seemingly modest results and so I think
12:41before we sort of even get into talking
12:43about any of any of any of the papers
12:45we're going to look at I think it's kind
12:47of clear that even if industrial policy
12:50is a tool that can work
12:52the name of the game is going to be that
12:54it's it's it's it's it's very very
12:57difficult to make it work well in
12:59practice and and and sort of my punch
13:02line today
13:03um is going to be that we need so much
13:04more work on trying to think about what
13:07makes for the good design of
13:09Institutions and policies that make
13:11um that could make IP work well because
13:14the other punchline is going to be that
13:16I think we have sort of now a growing
13:18body of evidence that suggests that
13:20those economic mechanisms that I alluded
13:22to those market failures that I'm about
13:24to talk about
13:25those look real so this is something
13:27that we should worry about
13:30okay but just to just to finish off sort
13:32of the the the the the ID here you know
13:35this this this this issue that you know
13:38IP is also associated with a lot of kind
13:41of disappointing growth performances
13:42this is not something that that this
13:44literature is not not aware of and in
13:47fact I think a lot of the most kind of
13:49interesting qualitative work in this
13:52strand of the literature sheds light on
13:54what makes for good industrial policy
13:56what are the institutional sort of
13:58setups that that that that sort of Stack
14:01the cards in favor of industrial policy
14:04succeeding and I've and I've just listed
14:05a couple of my favorite
14:07um my favorite books there but um
14:09there's there's many many more this is a
14:11big literature
14:15so I
14:18let me pause here and see if anyone has
14:21any questions if not I'm happy to keep
14:23going
14:25but hopefully I also need a sip of water
14:28okay
14:30so
14:31that's sort of kind of one side of the
14:33debate
14:34um
14:35if I want to if I want to sort of
14:38summarize the view from economics
14:41I think it would be fair to say that
14:43until a couple of years ago
14:45um you know this is this is kind of a
14:46cynical take but until a couple of years
14:48ago the mainstream kind of um dominant
14:51view in the profession was that
14:53industrial policy is not a good idea um
14:55you can take this to the extreme and
14:57kind of quote Gary Becker who who says
14:59that the best industrial policy is none
15:02at all
15:03okay now if you if you think about sort
15:06of why economists
15:09sort of strongly
15:11have have sort of strong issues with
15:13industrial policy I think that there are
15:16very sort of compelling arguments that
15:18they make about things that we need to
15:20worry about
15:21so you know
15:23as we've already alluded to the
15:24justification for doing industrial
15:26policy is going to be some form of
15:28market failure and what what a lot of
15:32um economists worry about is that you
15:36know it's all good and well for the
15:37social planner to come in and fix the
15:39market failure through the lens of your
15:41model but when you try to get the policy
15:43maker to do this in reality you also
15:46need to contend with sort of the very
15:48real issues of government failure okay
15:50so this could be something as simple as
15:52information failures you know we will
15:55write down a model in which this market
15:56failures our policy makers don't know
15:59where the market failures are okay
16:00that's not something that's like super
16:01easy to go out and measure in reality so
16:04first of all there's just simply an
16:05information problem of whether the
16:07whether the policy maker knows what it
16:08is where the market failure is and how
16:11they should try to correct that
16:13and then there's sort of the whole layer
16:15of rent seeking and political economy
16:17problems that also
16:19sort of compound this problem when you
16:21think about doing industrial policy in
16:22reality and so I think
16:25I I I you know as someone who sort of
16:27strongly believes in in the potential
16:29for industrial policy I don't think that
16:31these are things that we should be
16:32taking lightly I think that these things
16:34that these are things that we should be
16:35taking very seriously and and make kind
16:38of topics of of of Investigation so I
16:41think this is something that that we
16:42should really you know
16:44keep in mind the other thing that I
16:47think has
16:48um
16:49has led many economists to believe that
16:52that the industrial policies is is
16:53broadly speaking not a good idea is that
16:56there's a deeply held belief I think in
16:58the profession that free trade is
17:00conducive to growth
17:02um and I think that you know I broadly
17:05speaking ascribe to that view of the
17:07world
17:07um I think that that
17:09what where I would sort of shade this a
17:12little bit is that
17:13um thinking that there's a place for
17:15industrial policy and the policy makers
17:16toolkit is not necessarily kind of
17:19mutually exclusive with a belief that
17:21that economic integration economic
17:23openness
17:24um open market competition in in in in
17:27World Markets is is typically going to
17:29be too many
17:31please can we all move
17:34move out move out
17:38quickly
17:40don't say
17:46oh Caleb could you mute
17:54okay
17:55um I think I'm gonna keep going yeah
17:57okay okay so
17:59um
18:01the point I was trying to make is that I
18:03think it's not sort of mutually
18:04exclusive to book to believe that that
18:07economic integration that economic
18:09integration is typically
18:11um something that we should that is
18:13typically sort of conducive to growth
18:15and believe that there's potentially
18:16also a place for industrial policy in
18:18the in the policy makers toolkit no
18:21taking even even if sort of I take the
18:23most extreme views on all this and think
18:25that you know there's a really
18:26compelling case for free trade in all
18:28cases you know government failure is
18:30something that I really really worry
18:31about where I think
18:34um the profession has sort of fallen
18:36short in in until until sort of the last
18:39couple of years is that when you look at
18:41sort of the the body of empirical work
18:44on which we've sort of based our strong
18:46views that industrial policy doesn't
18:48work
18:49that is that is sort of extremely thin
18:52so I encourage you all to look at Ann
18:55Harrison and Andres Rodriguez Claire's
18:57excellent handbook chapter from 2010 on
19:00this topic and you can see that there's
19:02there's sort of a really nice large body
19:04of theoretical literature they talk
19:06about the qualitative work
19:09um and and and when they talk about sort
19:10of the empirical work that's out there
19:12you know every single section ends we
19:14need so much more work in this area
19:16okay so I think I think what what
19:20my other big message to you all as as
19:22PhD students is going to be is that this
19:25is an excellent area to be researching
19:27in the the the the the the the the sort
19:30of the the number of low-hanging fruits
19:32that we have in this field is enormous
19:34and we need sort of new creative ideas
19:37fresh fresh people to to sort of enter
19:40the field
19:42okay so here's a road map of the talk
19:45um I'm gonna spend just a couple of
19:48minutes
19:49giving you sort of the simplest toy
19:52model possible
19:53to give the theoretical justification
19:55for industrial policy
19:57and I think this is going to be helpful
19:58because then as we go to the empirics
20:03um I'm I'm I'm I'm constantly going to
20:05keep referring back to sort of all the
20:07indirect ways in which we're trying to
20:08test for what the the the theory
20:11predicts and and and and where sort of
20:13indirectly we're going to be trying to
20:14get at at assessing whether those market
20:17failures that we're going to talk about
20:18are really there I'm then going to show
20:21you sort of two kind of recent papers
20:25um that I think really
20:27put front and center identifying kind of
20:29the economic mechanism of infant
20:31industry and then I'm going to talk a
20:33little bit about where I think there
20:37really hasn't been
20:39um enough work which is to to to try to
20:41get it understanding this implementation
20:43to sort of really work towards
20:45understanding what policy making looks
20:47like in the real world when does it work
20:49when does it go wrong and how can we
20:51sort of design institutions and policies
20:54that would potentially make it um uh
20:56that would potentially stack the cards
20:58towards us doing better versions of IP
21:02okay let me pause that for a second and
21:03see if there are any questions
21:08reckon there's one in in a q a here
21:11there's a 2019 paper by mazukato and
21:14Carter where they argue for
21:16mission-driven industrial policy that is
21:19to use industrial policy to Target
21:20social and environmental goals rather
21:22than just correct market failures would
21:24be great to hear your thoughts on this
21:26yeah okay so I have many thoughts on
21:29that but let me give you the sort of one
21:31minute view on this
21:32um it's it's it's it's it's it's ironic
21:34that um this question came up because
21:37um I am based in Vancouver and the the
21:40government of British Columbia has
21:42worked with matsukato on designing one
21:45of these Mission oriented industrial
21:46policies for British Columbia
21:50um and and and we've been talking to
21:52policymakers recently here about sort of
21:54how they're thinking about these things
21:55so I think the two big things
21:58um okay three big things that I think
21:59about this are one I think mazukado has
22:03a lot of great ideas I think she's she's
22:06sort of particularly her earlier work on
22:11the entrepreneurial State I found really
22:13interesting
22:15um where I think I disagree with her
22:16strongly is that is that she sort of
22:19really I think um
22:22she really rejects this idea that we
22:24should be thinking about industrial
22:25policy through the lens of market
22:26failures and I and I and I and I
22:29strongly disagree with that I think that
22:30that to think of this coherently we have
22:33to think about a world in which we're
22:36trying to to understand what the
22:38rationale is for intervention and and
22:41and you know I think there are ways to
22:43come up you know if you move outside of
22:46um so the strict lens of Economics there
22:48are possibly sort of geopolitical
22:49reasons why you might want to do things
22:51differently but I think that
22:54I think that market failures just gives
22:56us a very sort of rigorous framework for
22:59thinking about these things the other
23:01thing that I think is kind of
23:03problematic with this Mission oriented
23:04approach is that it's very slim on the
23:07details you know if you look at sort of
23:09BC's um I think I think their economic
23:11plan is stronger BC you know all their
23:14missions are things that you know we all
23:16want the mission is to have good jobs
23:18the mission is to have
23:20um a green energy transition like who
23:22doesn't want that but but but I think
23:24where
23:25where really the trade-offs and the
23:27difficulties with industrial policy come
23:30in is like what is it that the
23:32government should be doing like should
23:33the government set up a plant to to
23:36manufacture batteries in Vancouver maybe
23:40I don't know I think that's probably not
23:43a great idea but but you know you need
23:45to sort of move down one level into the
23:48details and the specifics and think
23:50about the trade-offs because we're
23:51spending a dollar here that means we
23:53can't spend a dollar somewhere else and
23:54so I think that while
23:57well that there's this kind of Merit to
23:59thinking about missions I think that we
24:01also need to take seriously how the
24:04implementation works I think where a lot
24:07of industrial policy goes wrong is
24:09precisely
24:10um in in in in implementation and I
24:13think that sort of not focusing the
24:15light there and and and and and and
24:17focusing it sort of one layer up that
24:20leads that really sort of leads policy
24:23makers astray and I think we're sort of
24:26where we as economists are really sort
24:29of well tooled up to think about these
24:31things is in terms of how to make policy
24:35work and and and I'm going to sort of
24:37try and allude to this later but you
24:39know the the things that
24:41the things that are difficult about
24:42industrial policy are not necessarily
24:44unique to Industrial policy like a lot
24:47of the things that we worry about in
24:48monetary policy share analogies with
24:50industrial policy yet we don't think the
24:52government shouldn't be doing industrial
24:54policy we've spent a lot of time as a
24:56profession figuring out how to make
24:57institutions work in ways that are
25:00incentive compatible for getting us to
25:02to to to what to what we want the policy
25:05maker to be doing in terms of in terms
25:07of the long run I think many of these
25:09issues are similar for for industrial
25:12policy and the devil is in the details
25:14like I think figuring out missions is
25:17great but we need to we need to move one
25:20level down and then thinking about you
25:22know if you want to build a a a publicly
25:26owned
25:27um EV plant in Vancouver like what is
25:31the market failure that you're trying to
25:32address there like what thing are you
25:34going to do that the private sector
25:35couldn't do well I think that market
25:38failures just gives you a very
25:41um
25:41clean way of thinking about these
25:43questions
25:44so um that's yeah I I I guess we can we
25:48can sort of
25:49I'm happy to talk more about this
25:51offline but I think that
25:52while mazukado brings sort of a lot of
25:54fresh thinking and and great ideas I
25:56think what we need to worry about are
25:59the details and and and and how to get
26:01implementation right
26:05okay so
26:06um
26:07I'm gonna jump straight into the
26:09conceptual framework
26:11so I'm going to show you something that
26:13is super super simple this is how we
26:15teach it to undergrads but I find this
26:17actually the most useful way of thinking
26:19about
26:20um infant industry
26:22so as as we alluded to many times the
26:26justification for doing industrial
26:27policy relies on some existence relies
26:31on some market failures this is
26:33typically going to be some sort of
26:35externality
26:37um I'm gonna write it down here as an
26:39external economies of scale because
26:40that's going to be static and so it's
26:42going to make our life really simple but
26:43I'll I'll talk in a couple of slides
26:45about how in reality we think that these
26:48things are probably also
26:50um dynamic
26:51okay but let's fix ideas let's assume
26:54that there is a representative firm in
26:56the economy that has
26:58um this very simple linear production
27:00function
27:01um it's it's it's it's transforming
27:03units of Labor into units of output and
27:07the labor productivity is is denoted
27:09with this cap I
27:10now the twist with with
27:12um external economies of scale is that
27:15at the aggregate level so the individual
27:17funds in the economy take this a to be
27:20given uh take this a is given okay so so
27:23for them this is a parameter and so this
27:26this production function is constant
27:27returns to scale at the firm level
27:29however at the aggregate economy level
27:32this a depends on how much
27:36um on on what the domestic economic
27:39activity is in this sector right so
27:42basically what they say what this is
27:44saying is that labor productivity
27:45depends positively on the scale of the
27:48domestic industry
27:49and and and if if you're taking nothing
27:52else away from the conceptual framework
27:54what I want you to sort of
27:56take a mental picture of is this is this
27:59is this downward sloping industry-wide
28:02average cost curve
28:03the idea is that with external economies
28:06of scale the more the industry produces
28:08so Kappa is industry production the more
28:11the industry produces the lower is our
28:13average cost the higher is our labor
28:15productivity
28:16okay now why does this lead to an infant
28:20industry type prediction well imagine
28:22now that this is a small open economy
28:25um they take prices as a given and the
28:28the free trade price is this P star
28:31okay under free trade prices if this
28:34economy starts at zero levels of
28:36production then no one in the economy no
28:39firm in the economy wants to enter right
28:41I as a single individual firm that is
28:44taking that a as given I look at sort of
28:47what the price is p star I look at what
28:49my costs are they're too high and I say
28:51I don't want to enter the market
28:53okay we all think this way and no one
28:55enters the market
28:56precisely because no one is
28:58internalizing the fact that if we all
29:00enter together and start producing then
29:03very very quickly we're going to move
29:05down this industry-wide average cost to
29:08go
29:08and so this is where it becomes possible
29:11for the policy maker to come in
29:14and by providing a sufficiently large
29:16tariff or subsidy they make it
29:18profitable for funds to enter right so
29:20when you when you put the Tariff in
29:21place for example that's going to raise
29:23the domestic price it's going to make it
29:25profitable for fans to enter and here's
29:27the catch the Tariff only needs to be
29:29temporary right because once you've got
29:31your firms producing somewhere here you
29:34can remove that tariff and it becomes
29:37um it remains profitable for these firms
29:41to continue to produce at P star
29:44okay so this is kind of the magic of
29:47um infant industry
29:50we're not there yet at the intuition
29:53that I showed you I need to put sort of
29:54a little bit more structure on the
29:56problem but but the reason I say just
29:58remember remember this is that because
30:00in all our empirics what we're really
30:02going to try to be getting at is does
30:05some short run
30:06um intervention
30:09show us evidence that looks consistent
30:11with an industry moving down its average
30:14cost curve put differently does some
30:18sort of temporary intervention change
30:20the long-term profitability of the
30:22industry okay so this is going to be the
30:24name of the game
30:26this doesn't the reason I need to put
30:28more structure on this is that this
30:29doesn't tell me that doing this is a
30:30good idea right there could be some
30:32other product in the economy that that
30:34we would be much better off at producing
30:36and just importing why
30:39okay so I want to spend sort of
30:42really just a a a couple of minutes
30:46walking you through kind of what the
30:48structure of of of of of of of a more
30:51fully fledged infant industry model
30:53looks like
30:55this is still going to be as simple as
30:58possible
31:00um I'm taking this from Harrison and
31:02Rodriguez Claire's handbook chapter
31:04it's one reason is that it's extremely
31:06simple and it's sort of really nice to
31:09it's easy to present the other reason is
31:11if you're interested in sort of how to
31:13make this richer and more complicated
31:15and add sort of more things to it than
31:18than the first section of that handbook
31:20chapter sort of walks you through all
31:22the additional layers you can add onto
31:23it so this is I think a great entry
31:25point
31:26basically the the model is going to be a
31:28simple two-sector ricardian model with
31:31the twist that we're going to add these
31:33external economies of scale to one
31:34sector
31:35okay so there are going to be two goods
31:37one and two and there's going to be one
31:40factor of production cap l in fixed
31:42Supply
31:44we're going to continue with the small
31:45open economy assumption
31:47um we're going to label this economy
31:48South they take prices
31:50um P star is given P star is the
31:52relative price of good two to good one
31:55okay good one is really boring this is
31:57sort of the the constant returns to
31:59scale The Firm level no aggregate
32:01externalities
32:02okay so this is the production function
32:05um for firms producing good one
32:08good too looks like what I just showed
32:11you good two is also produced with
32:13consonant returns to scale at the farm
32:15level just like before
32:17but again here we're going to add the
32:19aggregate externality so this is
32:21it's it's that there's nothing sort of
32:23conceptually difficult here it's just a
32:25more complicated term for for reasons
32:28that I won't get into but what you
32:29should see here is that this this term
32:32in curly braces this is labor
32:34productivity it's taken as given by the
32:37firm and again it depends on how much
32:40labor you allocate
32:42um to the sector which is which is where
32:45the externality is coming from again the
32:47more labor you allocate to this sector
32:49the higher is your labor productivity
32:51and no form is internalizing this
32:55um basically the the the reason this has
32:57to be a little bit more complicated is
32:58that we make the assumption that you
33:00can't get these productivity gains
33:02forever so if at some point once you add
33:05more and more labor to this sector you
33:06sort of max out your productivity gain
33:08at this L Bar
33:09and you don't need to worry about this
33:11but but in what follows we're going to
33:13assume that the economy has enough labor
33:15that it can reach this this this L Bar
33:17when when when when when it allocates
33:19labor to this production of good two
33:22okay so now we're ready to talk about
33:24some things and and this is where we can
33:26really link back to that intuition that
33:28I started with at the beginning about
33:30multiplicity of equilibria and good and
33:33bad equilibria
33:34okay so for this restriction on
33:37parameters so when when when the labor
33:39productivity's
33:41um satisfy the following inequalities
33:42there's going to be multiple equilibria
33:45what we mean by that is that for this
33:48for this country south complete
33:50specialization in either good one or
33:53good two is in equilibrium okay
33:55basically the intuition is that when
33:58we're fully specialized in good one
34:01um this is going to be our opportunity
34:03cost of production and so it's gonna
34:04we're gonna have a comparative advantage
34:06in producing good one and when we
34:09allocate all our labor to good two what
34:12our our opportunity cost of production
34:15is going to look different
34:16precisely because we're getting these
34:20external economies of scaling good too
34:21and then we're going to have a
34:22comparative advantage in good too
34:25Okay so
34:26there's there there are multiple
34:29equilibria fine the second thing is the
34:33equilibrium with specialization and
34:35good2 is superior okay so yes there are
34:37good and bad equilibrium now
34:40good too the one with external economy
34:42is a scale that one is better it yields
34:44a higher wage okay so in in this world
34:47we think of the South as having a latent
34:50comparative advantage into it's latent
34:53because when you're specialized in good
34:54one it doesn't look like you're good at
34:57doing good too in a comparative
34:58advantage sense
35:01okay this is exactly the intuition I
35:03started with that that through the lens
35:05of infant industry
35:08in developing countries some Industries
35:11are stuck or the economy is stuck in a
35:14bad equilibrium
35:16and a temporary tariff can possibly move
35:19the economy from the bad to the good
35:21equilibrium and this may be welfare
35:23enhancing
35:24okay this is this is sort of the the
35:26simplest way of stating the the the
35:29infant industry model
35:31and this is really
35:33um
35:34this is really just to show you that the
35:37intuition Works what we're really going
35:39to be testing for
35:41what we're really going to be testing
35:43for when we when we look at the empirics
35:45is this okay we're going to be seeing
35:48does temporary intervention change the
35:50long run profitability of the of the of
35:53the industry
35:57so let me give you a couple just a
36:00couple of notes one thing that I want to
36:02point out is that you know we've done
36:04the static model just because it's it's
36:06it's so much easier to do but more
36:08generally we think of this process as
36:10being Dynamic that is that it's not that
36:13we all enter the sector together today
36:16and and magically our product our
36:18productivities are higher we think of
36:20productivity today as depending on
36:23cumulative output today
36:24and this is going to be important
36:26because this is going to basically
36:27justify why in empirical work we really
36:29want to focus on sort of long-run
36:31outcomes we think that if these we think
36:34of these as learning by doing
36:35externalities we think that if these
36:36learning by doing externalities take
36:38time to realize we need to be sort of
36:40observing this industry
36:41over over a fairly long time period to
36:45allow time
36:47for the industry to to move down its um
36:51average cost curve
36:55the the other question that I think I
36:58just wanted to sort of address
37:00immediately is that you know I can write
37:02down lots of models of the economy
37:04um and and and and and and and
37:06uh and and derive all kinds of crazy
37:08predictions like is writing down an
37:10externality in this way is this
37:12something that seems just ludicrous or
37:15do we think that it's reasonable
37:16reasonable to believe that there are
37:18these types of localization benefits
37:21that that Farms have these sorts of
37:23spillovers from production
37:26and I think that that you know just
37:29before we get into any of the empirics
37:30it's important to point out that this is
37:32not something that that we believe isn't
37:35sort of a potentially a strong feature
37:38of the real world so there's a nice
37:40handbook chapter by Rosenthal and
37:42strange
37:42which talks about this
37:45um and you know just
37:46the the the the
37:48sort of simplest stylized fact that
37:50speaks to this is that if you think
37:51about the the distribution of economic
37:53activity across space it is extremely
37:56extremely concentrated there are many
37:58reasons for that but one sort of
38:00prominent reason in the literature is
38:02that Farms benefit from being close to
38:04each other in this type of spillover way
38:08okay now the other sort of thing that I
38:10want to flag at the outset is that
38:13this really does look magical right like
38:15you don't you know the policymaker and
38:16this this comes back to this point that
38:18I made at the beginning that the policy
38:20maker changes through their policy the
38:24allocation of resources you know you put
38:26labor from good one into good two and
38:28all of a sudden we also become more
38:30productive
38:31okay and and and one thing that I think
38:35um is is probably not true in reality is
38:38that this process is unlikely to be
38:40completely passive in this way and I'm
38:42going to try and show you some evidence
38:44for that um a little bit more later I
38:46think that there's actually sort of this
38:48is more costly and potentially more
38:49risky for firms and and this is also a
38:53reason why getting industrial policy
38:55right may be difficult
38:56the last sort of
38:58technical point is that um tariffs are
39:01actually not the most efficient way of
39:02correcting this market failure um
39:04mallets has a nice paper on this
39:06um but but particularly in the case of
39:09developing countries it's often the only
39:10one that is feasible
39:12okay so how are we doing on time okay
39:19so let me start
39:21building the Bridge from from the
39:24theoretical predictions to empirics
39:29when we think about evaluating
39:31industrial policy in the real world I
39:33think something that is very important
39:35to bear in mind is that there are really
39:37two layers of treatment
39:39and it's typically very difficult to
39:41isolate them usually we get a context
39:43and we we have to isolate the two
39:45together
39:46so one the first layer of treatment is
39:48the economic mechanism you know this is
39:50a question about here's this industry is
39:53the infant industry mechanism here
39:55empirically relevant that is
39:57our market failures in this sector
39:58really there and and and do they look
40:01sort of like that they're sizable
40:05typically we don't get to analyze this
40:07very cleanly okay because typically
40:10we also have an entire layer of the
40:14policy maker doing the industrial policy
40:16on top of this
40:18and so when you're when you're
40:19evaluating a particular case of of
40:22infant industry protection you're also
40:25evaluating did the policy maker identify
40:27the right sector
40:28is the policy measure effective you know
40:31in our models a tariff is going to set
40:33the right incentives for firms in
40:35reality a tariff just might create a
40:37cushy domestic market and it might not
40:40sort of get firms to move down that
40:42long-run average cost curve in the model
40:44that happens passively as I said in
40:46reality it might be a lot more
40:48complicated and then there's the whole
40:49political economy
40:51the reason I say this is that
40:55most empirical studies
40:57are going to have to get a lot of both
40:59in there
41:01and so we shouldn't be surprised if the
41:04literature is going to find a lot of
41:06mixed findings we shouldn't we shouldn't
41:09think of it as being somehow
41:10contradictory that sometimes we find
41:12that industrial policy works and
41:14sometimes we find that industrial policy
41:15doesn't work in fact I think the name of
41:17the game is if we have enough evidence
41:19to believe that this is this is an
41:21important force in reality then this
41:24becomes kind of the object of Interest
41:26the object of Interest becomes in what
41:28cases can the can the policy maker do
41:30this well and how do we design
41:32institutions how do we get the policy
41:33maker to do this well
41:36I'm going to show you today I'm going to
41:38show you two papers which get as close
41:40as possible I think to isolating the
41:42economic mechanism
41:44and and and it's going to be for that
41:46reason that that at the end I'm going to
41:48sort of argue that
41:50it seems that in some cases you know it
41:53seems likely that the economic mechanism
41:54is relevant and then that that gives us
41:57all the more cause to worry about
41:58implementation
42:01okay there are three empirical
42:03challenges that I want to talk about
42:04related to evaluating IP
42:08foreign
42:12is hard to observe you know different to
42:14a minimum wage different to tax policy
42:17industrial policy is typically not
42:19something you can go out into the real
42:21world and easily observe why well think
42:24of our sort of textbook example of
42:27tariffs an import tariff
42:29we know we've talked about this an
42:31import tariff could be implemented to to
42:34to protect an infant industry
42:37but we know that a very important
42:39justification or a very important
42:40motivation for doing
42:42um for putting in place tariffs is to
42:44raise government revenue that's even
42:47also the case in developing countries
42:50we also have empirical evidence that
42:52tariffs are used for terms of trade
42:54reasons for special interest politics
42:56and so you know just observing the
42:57Tariff schedule of a country is not
43:00going to tell you what the industrial
43:02policy stance of that country is
43:03typically because there's all these
43:05other things that tariffs are being used
43:07to do
43:08this this this problem is then
43:10compounded by the fact that
43:13a tariff is only one policy tool that
43:15you can use there's many many others a
43:17production subsidy a land grant
43:19technology policy these are again all
43:22things that could be used for IP
43:24rationales or for other rationales and
43:26so again we have this observability
43:28issue
43:29I think for this reason
43:31much of the best work I think in the
43:33industrial is policy literature in in
43:35this type of infant industry strand uses
43:37industry-specific studies because in
43:40this case we have sort of the contextual
43:42information we have the government
43:43policy documents we have the historical
43:45accounts we have the newspaper reports
43:47that all talk about what the goal of the
43:50policy maker is when they put this
43:52policy in place and here again I want to
43:54link back to that definition I started
43:56with industrial policy is goal-oriented
43:59action
44:00we really need to understand the goals
44:02of these policies before we start
44:04classifying them as industrial policy
44:07the second the second sort of challenge
44:10um is that we want to evaluate long-run
44:12outcomes for the reason that I told you
44:14typically we think that these external
44:16economies of scale are not static that
44:18they're going to have a dynamic
44:19component and it's going to take time
44:20for the industry to move down its
44:22average cost curve
44:24all the recent work that I'm aware of
44:26does this so they all evaluate medium to
44:29long-term outcomes and we'll actually
44:31see in in one of the papers that I'm
44:33going to show you today that you really
44:34see this sort of nice gradual kind of
44:36takeoff of the industry and and it does
44:38seem at least sort of consistent with
44:40this idea that these these these Dynamic
44:43learning by doing the these Dynamic
44:45learning by doing gains and taking time
44:47to materialize
44:49okay the final thing that I want to sort
44:52of flag is that with industrial policy
44:55credible research design is really
44:57crucial and this is where I think a lot
44:59of the research um the the earlier
45:02research in the 90s
45:04um sort of fell short a little bit
45:07more than sort of your typical
45:10um regression here IP is inherently
45:13endogenous right the policy maker is not
45:15targeting Industries at random
45:18um
45:18and so and and and and and and making
45:21that even sort of worse is that it's not
45:23obvious which way the bias will go so in
45:25the case of infant industry the typical
45:28worry is that the policy maker is
45:29targeting an industry that would have
45:31taken off anyway so you know think of
45:34the US and Germany in the 19th century
45:36with the high tariffs the worry is we
45:38can't interpret that as causal because
45:40maybe the uh maybe the US and and
45:42Germany would have industrialized with a
45:44much lower rate of tariff too
45:47but with with other cases of industrial
45:49policy the bias might be downwards in
45:52OLS and so so so so so really we need to
45:55think about credible research design
46:00the literature has focused on three
46:02broad questions
46:03one and this is the one I'm going to
46:06talk about today
46:08one is whether the infant industry
46:10mechanism is empirically relevant and
46:13and this is really trying to get at that
46:15first layer of treatment like can we
46:17sort of
46:18either design a credible research design
46:20that gets rid of
46:21um a lot of of Layer Two or can we
46:24somehow find some natural experiment
46:26that switches off most of Layer Two
46:28we want to assess is this something we
46:31should worry about in the real world you
46:32know if the market failures are not
46:33there then Gary Beck is right we should
46:36do no industrial policy at all if the
46:37market failures are there I think at
46:39least there's a reason to kind of worry
46:40that we're that we're missing something
46:42really big
46:44the second question that that I think
46:46the literature has made some recent
46:48progress on but I'm not going to have
46:50time to talk about it in a lot of detail
46:51is whether industrial policy targets the
46:54right sectors right this is getting at
46:56the question and this is already getting
46:57a little bit into the implementation
46:59layer into Layer Two does it look like
47:01the industrial does it look like the
47:03policy maker can figure out which
47:05sectors of the economy are the ones
47:07where the market failures are and where
47:09where it should be um directing its
47:11interventions
47:13and where there's been really no
47:14convincing empirical work I think is
47:17this is this third question which I
47:19think is crucial and I think there's
47:20there's ample scope to do so much more
47:23here what is the effective industry of
47:25industrial policy on a country's
47:27economic performance this is where sort
47:30of we need
47:32some sort of empirical evidence to speak
47:35to this large literature outside of the
47:37social sciences which says oh industrial
47:39policy that was sort of the main driver
47:41of the East Asia growth Miracle you know
47:44this is bread and butter Economics work
47:46we have
47:47great new tools developed over the past
47:50two three years in econometrics that let
47:52us think about assessing these maybe not
47:55in perfect ways but at least in better
47:58ways than we've done before and and you
48:00know a level of ignorance at the moment
48:01is pretty high and so I think the bar
48:04here to do something to do something
48:06that is good is is is pretty low
48:10okay
48:12um
48:13any questions
48:20there are a couple from that there are a
48:22couple from the chat
48:25um one quick one uh from uh Royal Castro
48:29Corona is um
48:31do you have anything to say about
48:33industrial policy induced institutional
48:36change uh is this a relevant area of
48:38research another question here is
48:41um
48:43are there any papers out there that make
48:45a horse race between economies of scale
48:46and learning by doing and finally
48:49besides IP helping firms produce under
48:52lower marginal costs is it unambiguous
48:54that it should increase competition uh
48:57perhaps it will lead to higher markups
49:01once you take industrial policy away so
49:03feel free to answer some or any of those
49:06okay let me let me answer let me answer
49:09one that I think
49:10um I have a good answer to I off the top
49:14of my head so the one that asked about
49:15evaluating external economies of scale
49:17or learning by doing there's a nice
49:19paper by a PhD student out of Stanford a
49:21couple of years ago Ariel Ponce Venegas
49:24I think it's it's I mean I'm gonna it's
49:25in the slides and it's in the references
49:27and it'll be in the reading list that I
49:28upload he evaluates industrial policy in
49:31Japan and he looks at evaluating whether
49:36Japan was successful at targeting
49:37industrial policy towards sectors with
49:40these types of static um external
49:43economies of scale or towards Dynamic
49:45um economies of scale these kind of more
49:47learning by doing type things and he
49:49finds that the policy was more oriented
49:52towards the the the the the sectors with
49:54external economies of scale though I
49:56think he finds evidence on on on both so
49:58so that's one thing I think if I
50:01understood the question on tariffs right
50:02I think this is the name of the game so
50:04so
50:05I think in our model the Tariff is going
50:08to do the right thing the firms are
50:09going to the firms are gonna move down
50:10their their average cost curve in
50:13reality we know from a great
50:16literature
50:17um in in in trade that when you
50:19liberalize trade in developing countries
50:22there's a substantial productivity
50:24effect for Farms part of that is going
50:27to be access to intermediate inputs but
50:30another part of that is the the Escape
50:32competition effect that um that tariff
50:35protection creates this is sort of this
50:38is precisely kind of the problem that I
50:40think we need to worry about in reality
50:43and this is why sort of Designing and
50:46thinking about how we design better
50:47policy that can get around
50:50um these types of these types of issues
50:52would be would be would be much more
50:54important at the end of the day I think
50:56the trade-off is going to be and I'm
50:57going to show you some evidence for this
50:59if I have time a tariff is probably a
51:01really really blunt tool to do
51:02industrial policy at the same time
51:06many other industrial policies are
51:07extremely expensive they cost the
51:10government money think about what a
51:12tariff does it's raising government
51:13revenue for it from a developing
51:15country's point of view that is very
51:16fiscally constrained
51:18a tariff you can kind of kill two birds
51:20with one stone right a tariff is going
51:22to provide some infant industry
51:23protection and it's also not costing the
51:25government anything and so very often I
51:28think that the difficult trade-off is
51:30that a better form of industrial policy
51:32might be something
51:35um more technocratic more finely
51:37targeted
51:39um something that is expensive
51:41but that might not be feasible either
51:43from a physical point of view or from an
51:44administrative capacity point of view
51:46I'm going to show you that most of what
51:47rich countries do is they target their
51:49industrial policy towards individual
51:51firms using sort of very technocratic
51:54tools like trade financing or or or or
51:57or or or or a variety of other things
51:59that
52:00at scale are probably not going to be
52:02feasible for for developing countries so
52:05I think this is kind of the challenge
52:09okay
52:10um
52:12let me
52:13talk about the two papers that I want to
52:16do in some amount of detail
52:19these are both answering the question of
52:21is the infant industry mechanism
52:23relevant
52:24and they fit into this nice sort of
52:27history of these single industry studies
52:29which I think are very useful because
52:31they have a lot of information on the
52:32context of the policies
52:34the way in which I think this modern
52:36work has built on the past work is that
52:40is that much of past work actually all
52:42of the past work that I'm aware of uses
52:45a model to simulate the counterfacture
52:46of no protection I think it's been very
52:49important in in terms of sort of getting
52:51a grasp on how much we should worry
52:53about infant industry it's been really
52:55important for the literature to sort of
52:57get more credible reduced form evidence
52:59where you have sort of a research design
53:01that thinks carefully about the counter
53:03factual
53:05and and for reasons that I'm going to
53:07talk about most of these papers leverage
53:10natural experiments the reason being
53:12that you're really trying to get rid of
53:14a lot of the stuff in level two of of of
53:16of of treatment
53:19okay so the first paper I'm going to
53:20talk about is
53:22um a recent paper temporary protection
53:24and Technology adoption
53:26I think I've already introduced actually
53:28some of the key ideas the key idea in
53:30this paper is that we really want to
53:32find a way to narrow down on the
53:34economic mechanism and what this paper
53:36does is it actually doesn't analyze a
53:39case of industrial policy it takes a
53:42natural experiment that mimics tariff
53:44protection of an infant industry
53:46and for that reason it's gonna sort of
53:48it's not going to completely do away
53:49with Layer Two but it's going to do away
53:51with a lot of implementation there's no
53:53policymaker picking this industry
53:55there's no obvious political economy
53:58rent seeking that's happening around
53:59this
54:00the other thing the paper does in terms
54:03of identification is that it's going to
54:05move what I think is sort of a a more
54:08difficult problem which is to find a
54:10credible counter factual for the country
54:13as a whole it's going to move this
54:14problem to within country
54:17um and it's gonna I I think it's it's
54:19intuitive to see why sort of thinking
54:21about counterfactuals and thinking about
54:23regions within a particular country is
54:25being more comparable to each other is
54:28is is sort of an easier sell than than
54:30comparing countries
54:31the particular context this paper looks
54:34at is the development of mechanized
54:35cotton spinning um in the 19th century
54:37and the and the sort of exogenous shock
54:40that it's going to use is going to be
54:41the Napoleonic blockade against Britain
54:45okay so
54:47um we've had a lot of words let me show
54:49you some pictures
54:50um one discussion to this paper once
54:52referred to mechanized cotton spinning
54:54is the iPhone of the 19th century and
54:56I've used it ever since because I think
54:57this is actually a really nice analogy
54:59cotton textiles was sort of the engine
55:03of the the first Industrial Revolution
55:04it was the flagship industry and
55:07importantly for us it was invented in
55:09Britain I we're going to be looking at
55:11France France is going to be the
55:12technological follower country it's
55:14going to be the country that gets
55:16they're late to moving down its average
55:19cost curve Britain's going to have the
55:21first mover Advantage when we pick up
55:22the story Britain has already moved down
55:24it's it's its industry-wide average
55:26Costco and France is not competitive
55:30it's going to be important for what I
55:32show you later that you take a mental
55:35sort of picture of this um of this slide
55:38here is the old technology it's a hand
55:40Spinning Wheel here's the new technology
55:42it is mechanized cotton spinning
55:45um the productivity improvements are
55:47huge that's one thing the other thing
55:49that I want you to see is that there is
55:51a massive reorganization of production
55:53that is happening the old technology is
55:55organized as rural cottage industry it's
55:57women spinning in their homes in
55:59agricultural downtime
56:01the new technology is large-scale
56:04factory production and within kind of a
56:06remarkable space of time the industry
56:09goes to having 60 employees on average
56:13and and one of the things that I'm going
56:15to be sort of highlighting at the end is
56:17that you know through the lens of our
56:19toy model learning how to do this well
56:21that's gonna seem easy and and Costless
56:24I'm going to show you some evidence that
56:26here in fact it looks like this was
56:29pretty costly to do for the farms and
56:33then everyone learned how to do it the
56:34technology diffused that's the
56:35externality
56:37but that this idea that we will
56:39necessarily move down that average cost
56:41curve that's not always going to be true
56:43and again a tariff is probably going to
56:45be a pretty blunt tool if we're trying
56:47to get our firms to go to this from
56:49doing this
56:53so
56:55mechanized cotton spinning in France in
56:57the early 19th century is a good
56:58candidate for infant industry
57:01um pre-mechanization France is Britain's
57:03closest European competitor this is
57:06important because I'm trying to build a
57:07narrative that these two countries are
57:09in all other respects the same obviously
57:11that's not true but actually at the at a
57:14high level for what we're trying to do
57:16it's it's pretty accurate
57:19um historians have made a case for large
57:21learning by doing externalities here
57:24and what I what I also think is kind of
57:27important is that French firms actually
57:29acquire this technology before the
57:30French Revolution but they're not
57:31competitive
57:33and so so one possible take on this on
57:36this setting that you could give is that
57:38there's a historical accident it happens
57:40to be the case for for some
57:42idiosyncratic reason Britain figures out
57:44how to mechanize Cotton spinning they
57:46get the first mover advantage and
57:47they're moving down their their
57:48industry-wide average cost curve France
57:51is trying to get in the game but they
57:53can't do it because because they're
57:54further up their average cost curve
57:57and in the midst of all this the
57:59Napoleonic blockade happens
58:02um
58:02leading to some areas of France to be
58:05better protected from trade with Britain
58:07than others and this is going to be sort
58:09of the quasi-exogenous variation that
58:11we're going to be trying to leverage to
58:15ask what the temporary what the effect
58:17of temporary Trade protection is on the
58:18French cotton spaying industry
58:21okay France was not competitive these
58:24are prices before the blockade
58:28um
58:30the key to the blockade so in the
58:33background you have war between Britain
58:35and France Napoleon has this idea that
58:38he's gonna try to crush Britain
58:41economically by denying them access to
58:44European export markets and so the name
58:46of the game for the next couple of years
58:48is going to be to try to stop British
58:50goods from entering all of Continental
58:52Europe that's a that's a continental
58:53Europe is Big so this is like a big
58:54undertaking and it doesn't it doesn't
58:57work basically instead of stopping Trade
58:59It diverts trade to more costly and more
59:02circuitous routes which is great for
59:04this paper because basically this is
59:06what's going to drive the variation if
59:07the blockade was successful everywhere
59:09would be shut off now instead places are
59:12going to get sort of differential trade
59:14shocks depending on where they are
59:16and this is going to allow us to sort of
59:18exploit this this this quasi-random
59:21variation in in trade protection
59:24okay it was clear even at the time that
59:28the blockade was not successful here is
59:30the mighty British Commerce
59:32um throwing all its goods at Napoleon
59:35across the channel
59:36printed calicos here where you see the
59:38cotton yarn
59:39um that we're talking about
59:42I'm gonna speed up a little bit because
59:44I want to get through
59:46um to the next paper and I think
59:50you can you can sort of access the
59:52slides and of course the paper as well
59:53so I'm going to sort of speed up the
59:55next slides because I think they're just
59:56going to make the point in sort of try
59:59to make the convincing point that there
1:00:01is this sort of interesting variation in
1:00:04trade that we can exploit
1:00:06this graph shows that trade did not stop
1:00:08rather the direction changed and and so
1:00:11let me explain the blockade with this
1:00:14slide and then I'm going to skip a
1:00:15couple of the next slides
1:00:17basically if you if you think about the
1:00:19map of Europe
1:00:20um in front of you Britain is north of
1:00:22France and so prior to the blockade most
1:00:27of British exports were going into
1:00:28northern Europe this this is this is
1:00:30really where
1:00:31um things start to go things start to go
1:00:33south so to speak for the British
1:00:37um and and and trade with southern
1:00:38Europe was a little bit lower basically
1:00:40the the the high level
1:00:42um story is that the blockade was
1:00:46successful to some extent in the north
1:00:47it was unsuccessful in the South and so
1:00:49what the British did was they diverted
1:00:51their trade through the South through
1:00:53the Mediterranean and into France and so
1:00:56basically what ends up happening is that
1:00:58before trade was coming into France
1:01:01these these these cut these pieces of
1:01:03cotton yarn that were competing with the
1:01:05French cotton yarn they were coming into
1:01:07to France from a Northerly Direction now
1:01:10that you have to reroute trade through
1:01:12the South they're coming in from a
1:01:13southerly direction and so this is going
1:01:15to give a different size shock depending
1:01:19on where you're located
1:01:21given what I just told you what we what
1:01:24we do in the paper is we calculate the
1:01:26change
1:01:28the effect the change in effective
1:01:29distance to to Britain before and after
1:01:32the blockade and it gives us this heat
1:01:34map for the French Empire our units of
1:01:36analysis are going to be
1:01:39um French departments and given what I
1:01:42told you it it it it it makes sense that
1:01:45the the blockade had the biggest sort of
1:01:47effect on protecting trade along the
1:01:49channel and it sort of decreased
1:01:51intensity as we as we move down south
1:01:55there are two questions that we're going
1:01:57to ask in this paper did regions which
1:02:00became better protected from trade
1:02:01increase capacity in mechanized cotton
1:02:04spinning
1:02:04okay this is this is just the question
1:02:06of whether the blockade was effective in
1:02:08getting entrepreneurs to get into
1:02:09mechanized cotton spinning
1:02:11and because there's this old technology
1:02:13that's available it's actually not
1:02:15obvious that the producers are going to
1:02:17scale up in the new technology there's
1:02:19this interesting model
1:02:21um by sorry that actually argues that in
1:02:23developing countries this is a very
1:02:24relevant context in the sense that what
1:02:27infant industry protection actually ends
1:02:29up doing very often is just promoting
1:02:31the the old technology the rural cottage
1:02:33industry instead of instead of getting
1:02:35the new technology off the ground
1:02:37and then in the long run we're going to
1:02:39ask
1:02:40whether the profitability whether we see
1:02:43any evidence that the profitability of
1:02:45production changed okay so did the
1:02:47effects persist after pre-blockade
1:02:49variation and protection was restored
1:02:51so in the short run we look at uh a
1:02:56different diff strategy with varying
1:02:58treatment intensity and the question is
1:03:00did protection render mechanized cotton
1:03:02spinning profitable in the short run
1:03:04the Baseline specification that we're
1:03:06going to look at looks at the number of
1:03:08spindles that's that's sort of the
1:03:10relevant the relevant measure of
1:03:12spinning capacity
1:03:13the number of spindles in Department I
1:03:15at time T we normalize this by
1:03:17population to make them comparable and
1:03:20then
1:03:20um the the coefficient that we're
1:03:22interested in is gamma
1:03:25um that is we want to see how changing
1:03:27protection changing effective distance
1:03:30to Britain changed capacity
1:03:33um in mechanized cotton spinning the
1:03:35identifying assumption is that more and
1:03:37less exposed regions of France would
1:03:38have trended similarly in the absence of
1:03:40the trade cost shock
1:03:42um there are two sort of things you
1:03:43worry about here
1:03:45I'm going to skip these but you can you
1:03:46can see the details in in in the slides
1:03:49in terms of what the paper does to try
1:03:51to address these
1:03:53okay
1:03:55I'm gonna do the eyeball econometrics
1:03:57thing and show you the the visual of how
1:04:00spinning capacity changed across France
1:04:03and then you can look at the regressions
1:04:05um if you like
1:04:07um afterwards what you should see here
1:04:10is that pre-block Aid
1:04:13um spinning capacity is of course
1:04:14concentrated more in some Departments of
1:04:16France than others but sort of
1:04:19reassuringly for us there seems to be
1:04:21sort of
1:04:22a fair amount of activity happening in
1:04:25the South and this is good because we're
1:04:27trying to build a case that the the
1:04:28South would have trended similarly to
1:04:30the north in the absence of protection
1:04:33what you see is that after the blockade
1:04:36there was this is not drawn to scale
1:04:38because because there's a huge increase
1:04:40in in capacity so we wouldn't be able to
1:04:42see
1:04:43um this if I drew this to scale so what
1:04:45you see is that spinning capacity
1:04:48quadrupled in the space of these 10
1:04:50years during the blockade but it was
1:04:52very much concentrated here in the north
1:04:54and particularly along the channel which
1:04:56got the largest amount of protection
1:04:59a spinning capacity stagnated
1:05:01um or even declined somewhat in the
1:05:03South particularly here along the along
1:05:05the Pyrenees
1:05:08I I'm not going to show you the
1:05:09regressions in the interest of time but
1:05:11basically what the regressions show is
1:05:13that this short in the short run
1:05:15protection
1:05:19increased mechanized cotton spinning
1:05:22more in the regions that were more
1:05:23protected from trade
1:05:25okay here are the regressions I'm just
1:05:26going to skip through them in the
1:05:27interest of time
1:05:31I think those results are somewhat
1:05:32interesting because as I said there are
1:05:34sort of two margins that that the French
1:05:36Spinners could have gone they could have
1:05:37stayed in the old technology or they
1:05:39could have scaled up in the new
1:05:40technology
1:05:41but I think from the point of view of
1:05:43the infant industry argument what we
1:05:45also really want to know is what happens
1:05:46in the long run okay did the short run
1:05:48protection affect the long-term
1:05:50profitability of mechanized cotton
1:05:52spinning and for this we sort of want to
1:05:54evaluate whether we see any evidence
1:05:56consistent with these Northern French
1:05:59departments moving further down this
1:06:01average cost curve
1:06:03the Baseline specification that we're
1:06:05gonna use to analyze this is going to be
1:06:08to look at sort of outcomes at the
1:06:10department level mostly whether this
1:06:13mechanized cotton spinning capacity also
1:06:15in the long run regressed on how much
1:06:17spending capacity there was in 1812. now
1:06:20this is not a regression that we can
1:06:22sort of seriously interpret when run as
1:06:24OLS because because
1:06:26um spinning capacity in 1812 is
1:06:28endogenous but of course the trade cost
1:06:30shock gives us an IV strategy we're
1:06:33going to try to use
1:06:35the spinning capacity that is only there
1:06:39in 1812 because of this exogenous event
1:06:41of increased trade protection
1:06:45okay and and
1:06:46what you see and you we see this also in
1:06:49the two SLS regressions that I won't
1:06:51show you but visually you can see that
1:06:54that the industry sort of not only
1:06:56persists in the north but this is where
1:06:59sort of the the the the the
1:07:02the dominant location of the industry is
1:07:05by by the the late 19th century you
1:07:08don't really see any spinning capacity
1:07:10um in the South and this is very
1:07:12consistent with the idea that Northern
1:07:15departments got sort of a head start
1:07:17moving down their average cost curve
1:07:19when pre-blockade variation in trade
1:07:21protection was restored these Northern
1:07:23departments had an advantage and now in
1:07:26the within France competition these
1:07:29Southern departments weren't really able
1:07:30to to catch up
1:07:34okay
1:07:36um
1:07:37I want to spend just a couple of minutes
1:07:39thinking about why temporary protection
1:07:42worked in this case in the sense of
1:07:44helping the industry scale up
1:07:46and and and there are three things I
1:07:49want to say
1:07:50one is that no this case fits nicely I
1:07:54think as as close as possible to the
1:07:56textbook case of one country gets a
1:07:58first mover advantage in reality I think
1:08:01when we're looking at this from the
1:08:02point of view of developing countries
1:08:04today you've just seen a bunch of
1:08:06lectures for other distortions that are
1:08:09happening in the economy and so we can't
1:08:11really think of all else being the same
1:08:13okay so I think I think when we think
1:08:15about so how to apply the logic here
1:08:18into into the real world we need to
1:08:20think about what other distortions
1:08:23um any type of infant industry
1:08:25protection would would interact with
1:08:27the other thing that's really important
1:08:28here is that the the entrepreneurs in
1:08:30France they actually wanted to mechanize
1:08:32okay they only needed a little bit of
1:08:34help and I think this is what is
1:08:36potentially not the same in in other
1:08:39contexts where the state is really
1:08:41trying to strong-arm entrepreneurs into
1:08:43entering a youth sector you saw that
1:08:44even in 1803 even at the beginning of
1:08:46the blockade there was already some
1:08:48mechanized cotton spinning capacity
1:08:50entrepreneurs want to do this
1:08:53and that that I think is important
1:08:55because they already have sort of their
1:08:56own incentives for trying to move down
1:08:58that average cost curve
1:09:01and then the final thing I want to say
1:09:02and and this is something we've already
1:09:04talked about a little bit is
1:09:06here of course protection happened
1:09:07because of war and and even though this
1:09:10protection sort of mimics infant
1:09:11industry
1:09:13um sorry it mimics the effects of a
1:09:15tariff
1:09:16I think there are there there's sort of
1:09:18certain important differences the most
1:09:20important being that because this is a
1:09:22situation of War I think it's it's it's
1:09:25reasonable to expect that these
1:09:26entrepreneurs had a lot of uncertainty
1:09:28as to whether protection would actually
1:09:30last okay and and this came up in
1:09:33response to a question that we talked
1:09:34about
1:09:36um in reality I think this Escape
1:09:38competition of effect of a tariff is is
1:09:41really important particularly if the if
1:09:43the policy maker is not very credible in
1:09:46signaling that this temp this protection
1:09:47is truly temporary
1:09:50okay now the final thing that I want to
1:09:52talk about here
1:09:53is what learning did the entrepreneurs
1:09:55do like wouldn't it be great if we could
1:09:57kind of try to understand what the
1:09:59magical thing is that happens that
1:10:01allows them to to to increase their
1:10:03productivity
1:10:04and so I just want to show you something
1:10:06real quick from a related paper
1:10:10um that we have with monastic and Nicole
1:10:12voigtlander in which we look at these
1:10:14same we look at this same industry
1:10:16except now we're going to look at the
1:10:17plant level we're going to look at what
1:10:19is happening to to individual plants
1:10:22at this early stage of Technology
1:10:24adoption and at the latest stage of
1:10:26Technology adoption
1:10:27and what you see here this is the
1:10:30productivity this is the labor
1:10:32productivity distribution this gray
1:10:33dashed line for mechanized cotton
1:10:36spinning
1:10:37at the beginning of the blockade this is
1:10:39the the productivity distribution
1:10:42um at the in in 1840 when the technology
1:10:45is mature
1:10:46and and the point that I want to impress
1:10:48on you here is that there is a
1:10:50really
1:10:51a lot of heterogeneity in farm
1:10:54performance at the early stage of
1:10:55Technology adoption relative to what
1:10:57comes later and this is not something so
1:11:00so don't worry too much about what these
1:11:02two sectors are basically the point here
1:11:04is to show you that there's something
1:11:05sort of unique happening at this at this
1:11:08early stage of Technology adoption in
1:11:11Cotton spinning there's a lot of poor
1:11:14performing Farms what happens in the
1:11:16long run is the productivity grows sort
1:11:18of
1:11:19strongly in this sector over this this
1:11:22three decade period and it's because we
1:11:24get rid of this this left tail of the
1:11:25distribution
1:11:27what we've been able to do in this paper
1:11:28is to kind of figure out what it is that
1:11:31firms are doing here and what seems to
1:11:33be the case is that Farms are figuring
1:11:36out how to do factory-based production
1:11:38right think about those those two
1:11:40pictures I showed you funds of figuring
1:11:42out how to efficiently run a factory and
1:11:45that is not easy and that is not
1:11:47happening as passively as that simple
1:11:48toy model of infant industry suggested
1:11:51so we've been able to find the building
1:11:54layout of 60 of these plants we know
1:11:56what the shape of the buildings were and
1:11:59this seems to be one important dimension
1:12:01for figuring out how to run a factory
1:12:03efficiently
1:12:04okay and what we see we see the number
1:12:06of flaws that a building has and we see
1:12:08sort of what the what the what the shape
1:12:10of the building is how close it is to a
1:12:12perfect square
1:12:14and what you can see is that again gray
1:12:16is the early stage of Technology
1:12:17adoption at the early stage of
1:12:19Technology adoption there is a great
1:12:21deal of experimentation about how to run
1:12:24a factory firms are doing all sorts of
1:12:26things later down the line
1:12:29as the technology matures the building
1:12:31layout becomes far more
1:12:35um the distribution for these becomes uh
1:12:38far less dispersed they Converge on some
1:12:40sort of best practice
1:12:42and we show that getting this right is
1:12:43related is is associated with a higher
1:12:47probability of surviving so that getting
1:12:48this right confers some sort of an
1:12:50advantage and we also show that the
1:12:54firms are learning from each other
1:12:55through spatial knowledge diffusion
1:12:58okay what I want you to take away from
1:13:00this paper is that what I've learned
1:13:03here that links to infant industry is
1:13:06that this type of Technology adoption
1:13:09that we're talking about doesn't happen
1:13:12sort of as magically as as as as as our
1:13:15simple models predict there seems to be
1:13:17sort of a lot of
1:13:19costly Investments That firms are making
1:13:22in experimentation
1:13:24and again this could be a reason once
1:13:27once this is not taking place passively
1:13:29this could be another reason why
1:13:32um
1:13:33a very blunt tool like a tariff is not
1:13:36really shedding is is is not really
1:13:38going to get the industry off the ground
1:13:40typically and here I sort of want to
1:13:42Echo Eric verhugan's point from lecture
1:13:45two which is that I think it's it's it's
1:13:47it's an excellent point that we need so
1:13:49many more detailed case studies as
1:13:52opposed to just focusing on tfp we need
1:13:54to sort of try to find creative ways of
1:13:56opening up what is happening
1:13:59um inside the firm when they adopt a
1:14:01technology for example I think this is
1:14:03really relevant for thinking about
1:14:04infant industry or in many cases you are
1:14:07TR the policy maker is trying to get the
1:14:09industry not just to enter a new sector
1:14:11but to to adopt Technologies new ways of
1:14:14doing things and I think that's that's
1:14:17actually pretty hard and that might have
1:14:19its own sort of market failures as well
1:14:26okay so in my last sort of 15 minutes or
1:14:29so
1:14:30I'm gonna walk you through
1:14:33um the other empirical paper
1:14:36I think
1:14:37to conclude with this paper
1:14:40um I think there's like a reasonable
1:14:41question one can ask which is whether we
1:14:44threw the baby out with the bath water
1:14:45so what does this episode really teach
1:14:47us about industrial policy you know this
1:14:49is such a sort of synthetic laboratory
1:14:52like setting we don't even have real
1:14:53industrial policy taking place and I
1:14:56think that's actually for learning about
1:14:58industrial policy in the real world I
1:15:00think this study is like it falls really
1:15:02short people have often asked me like
1:15:05what the policy conclusions of this
1:15:06paper are and and my honest answer is
1:15:09nothing like you know don't take this
1:15:11paper and and and go out into the real
1:15:13world and do anything because
1:15:15because we're missing all of layer 2.
1:15:18Okay so
1:15:20I think the strength and the weakness of
1:15:22this paper are the same the strength is
1:15:24that it gets us to focus as cleanly I
1:15:27think as possible on the economic
1:15:28mechanism and show that this is indeed
1:15:30potentially something to worry about but
1:15:32it but it also doesn't really speak to
1:15:36what industrial policy looks like in the
1:15:38real world and so for that I think the
1:15:41the next paper I want to talk about Lane
1:15:432022
1:15:45is is is is is is a much better example
1:15:47because it is evaluating industrial
1:15:50policy in as as it is practiced by the
1:15:52policy maker in the real world
1:15:55Okay so
1:15:57this paper belongs and I think it's
1:15:59actually the first paper in a now
1:16:01growing list of papers that is
1:16:03re-evaluating the East Asia growth
1:16:05miracle I think there's still plenty of
1:16:07work to be done here the paper that I
1:16:10referenced um in response to one of your
1:16:12questions is this paper by Ponce Venegas
1:16:15Liu 2019 is also a paper you should look
1:16:17at that is also trying to evaluate
1:16:19through the lens of a model whether
1:16:21South Korea and China were targeting the
1:16:24sectors
1:16:25um with with with sort of the largest
1:16:28um externalities and the largest market
1:16:29failures and and he he finds
1:16:33um also that this is the case
1:16:34but the paper that I think
1:16:37um is is most relevant for us here
1:16:40thinking about the empirics just just
1:16:41the economic mechanism of infant
1:16:43industry is Lane 2022.
1:16:46I think here the trade-off is going to
1:16:49be that because we're we're adding
1:16:51layers we're adding elements of Layer
1:16:53Two the the policy maker actually doing
1:16:55the treatment the identification can't
1:16:59be as clean almost almost by definition
1:17:01but we're getting a lot more of of of of
1:17:03reality here and I think that that is
1:17:05that is a trade-off well worth paying
1:17:09um so the title of this paper is
1:17:10manufacturing revolutions industrial
1:17:12policy and industrialization in South
1:17:14Korea and what Nathan does in this paper
1:17:16is he evaluates South Korea's um heavy
1:17:19chemical Industries push in the 1970s
1:17:24I'm a huge fan of this paper and I
1:17:26wanted to show you this today because it
1:17:28really is kind of a textbook case of a
1:17:31country attempting to shift its
1:17:33comparative advantage in the way that
1:17:35our toy model was was was was was
1:17:37suggesting
1:17:39um in a pretty radical way okay so 1970
1:17:42South Korea had very very little
1:17:44experience in the type of heavy industry
1:17:46that it was trying to get into every
1:17:48single person that the Koreans went to
1:17:49in the International Community to get
1:17:51money kind of laugh them out the door
1:17:53okay there was no one thought that this
1:17:55plan was a good idea
1:17:58um
1:17:58so no one you know
1:18:01in in through the lens of vowel through
1:18:03the lens of our model no one but the
1:18:05government was willing to bet that they
1:18:06had a late and comparative advantage in
1:18:08heavy industry
1:18:10so how did they pull it off well
1:18:13what what what Nathan argues in the
1:18:15paper
1:18:15is that basically the political crisis
1:18:18of the U.S kind of withdrawing a lot of
1:18:21its overt
1:18:22um military support for the for the
1:18:24South that led to a crisis
1:18:27um Park Chung hee launches a
1:18:29dictatorship and one of the flagship
1:18:31policies
1:18:32um in this era is an industrial big push
1:18:34for military self-sufficiency right the
1:18:37the South Koreans are worried about an
1:18:39invasion from the north they don't have
1:18:41any capacity to build arms and Munitions
1:18:45themselves and so the idea is that they
1:18:48need to get better at doing they need to
1:18:49get better at doing these things really
1:18:51really fast because we've just learned
1:18:53that the US is not going to be as
1:18:54helpful as they said they would be
1:18:56okay and this comes to an end quite
1:18:57abruptly in 1979 when patreonghi is
1:19:00assassinated so the policy is really
1:19:01this kind of short six-year policy
1:19:06we have some information on on on on on
1:19:10on what the policy actually is so
1:19:12they're really using two policy measures
1:19:15a massive amount of money is being
1:19:17poured into these HCI Industries
1:19:20and they're also using trade policy to
1:19:22steer resources and economic activity to
1:19:24treated sectors basically the government
1:19:27is messing with prices in the economy to
1:19:29try to get you from good one to good to
1:19:31to try to get you from the bad to the
1:19:33quote good equilibrium
1:19:35okay and and and the rationale for this
1:19:37kind of the natural experiment flavor of
1:19:39this is that what the Koreans want to do
1:19:41is to build rockets and Small Arms those
1:19:44are just like not on the table there's
1:19:46no way that they're going to be able to
1:19:48do this and there's no way that they're
1:19:49going to get financing from anyone to
1:19:50help them build Rockets
1:19:52okay what what seems feasible is to do
1:19:55the Upstream sectors to do the steel and
1:19:57all the other heavy industry that's
1:19:59going to allow them to one day do this
1:20:01stuff too
1:20:02okay so the goal is to shift Light
1:20:04Industry into heavy industrial economy
1:20:08here are two major policies that are
1:20:11being used one is the tax policy and the
1:20:14other is just lending by the Korean
1:20:16Development Bank
1:20:18so so the red line here shows you HCI
1:20:21sectors and you can see that tax rates
1:20:24uh marginal tax rates become much much
1:20:26lower dramatically lower for HCI
1:20:28Industries than non-hci Industries and
1:20:30you see that the government is also
1:20:32through the Korean Development Bank
1:20:34lending a bunch of money to these
1:20:36sectors
1:20:37the other thing that they're doing is
1:20:39that and this is I mean this is so
1:20:41impressively modern they're cutting
1:20:43input tariffs differentially for HCI
1:20:46sectors and they're also easing quantity
1:20:48restrictions a little bit for hcis
1:20:50sectors there's no evidence that that
1:20:53they're using import tariffs right and I
1:20:56say that this is a very modern policy
1:20:58because there's there's recent papers
1:21:00that show that that one of the main ways
1:21:04to improve productivity
1:21:06um when you one of the main channels
1:21:08through which productivity improves when
1:21:09you liberalize trade is that firms get
1:21:11access to um higher quality imported
1:21:14inputs and the Koreans were doing this
1:21:16they weren't doing that type of tariff
1:21:18protection of infant Industries as far
1:21:20as we can tell that that we've been
1:21:22talking about until now
1:21:24the other thing and I asked Nathan about
1:21:26this this week lurking in the background
1:21:28there's this export promotion that
1:21:30you've heard of
1:21:33Nathan said he didn't find a huge paper
1:21:34trail for this but it's it there's no
1:21:37explicit policies that he found that was
1:21:38pushing firms to export but this was
1:21:40sort of the unwritten rule of the day
1:21:42you were producing you had to export so
1:21:44there's this massive you know you should
1:21:46remember that in the background there's
1:21:48this massive sort of
1:21:49de facto push from the government that
1:21:52you have to be putting your stuff on on
1:21:55on on on on on International markets
1:21:58okay so Nathan's gonna what what the
1:22:01paper does is it estimates a difference
1:22:03in difference between targeted and
1:22:05non-targeted industries before and after
1:22:07the policies so again this is a diff and
1:22:09diff strategy looking at targeted and
1:22:11non-targeted industries and comparing
1:22:13them the identifying assumption being
1:22:15that the treated sectors would have
1:22:16evolved similarly to untreated sectors
1:22:18in the absence of treatment
1:22:21so to me what's compelling about this
1:22:24this setting
1:22:25is that usually what we worry about here
1:22:28is that the industry was just about to
1:22:30take off anyway right the policymaker
1:22:32targets something it takes off
1:22:34but we overestimate the effect of that
1:22:36industrial policy because because this
1:22:38industry is about to take off anyway you
1:22:40know the fact that no one was willing to
1:22:42bet on this you know no one put their
1:22:44money on the table to do this suggests
1:22:47to me kind of very strongly that these
1:22:49sectors were not sort of on the cusp of
1:22:51taking off
1:22:53Nathan evaluates two things the direct
1:22:55and the network effects of the policy so
1:22:57he's going to look at what happens to
1:22:59Downstream sectors so the users of these
1:23:01HCI Industries
1:23:03and he's going to find a positive effect
1:23:05on both the the ones that I've pulled
1:23:07out for you today he he looks at all
1:23:09sorts of
1:23:10um outcomes the ones that I've pulled
1:23:13out for you today are the ones that look
1:23:14at what happens to reveals comparative
1:23:16advantage this is
1:23:18um sort of a a proxy that that people
1:23:21use to look at whether a country is
1:23:24increasing its comparative advantage in
1:23:27in a certain sector and you see that
1:23:30um in lines sort of with with with the
1:23:33infant industry model
1:23:34once the policy starts the RCA of the
1:23:37treated sectors starts to differentially
1:23:39increase and it increases sort of Beyond
1:23:41when the policy is withdrawn again this
1:23:45is suggesting that there are these
1:23:48Dynamic learning by doing gains it's
1:23:50suggesting that this temporary treatment
1:23:53changed the long-term profitability of
1:23:55this sector
1:23:57and more importantly I think it suggests
1:24:00that Korea managed to sort of pull off
1:24:02something that is that is really kind of
1:24:04remarkable which is to shift comparative
1:24:06advantage in this sort of very very
1:24:07dramatic
1:24:09um way
1:24:12Nathan also the the paper also finds
1:24:15effects on Downstream sectors so so
1:24:18basically here the diff and diff that
1:24:20we're looking at is whether sectors
1:24:22that are more Downstream sectors that
1:24:25use more heavily the targeted sectors
1:24:27whether they see a differential increase
1:24:29in their in their revealed comparative
1:24:31advantage and he finds some evidence for
1:24:34that
1:24:35what's interesting about this you know
1:24:37before we all get sort of too optimistic
1:24:39about industrial policy is that this is
1:24:41kind of in sharp contrast to what we've
1:24:43seen in other papers so Bruce blerningan
1:24:46has a really
1:24:47um nice paper in The Economic Journal in
1:24:492014 where he evaluates sort of the
1:24:51exact same thing he looks at steel
1:24:53industrial policies
1:24:55in a variety of different countries and
1:24:57he evaluates how these industrial
1:24:59policies affect Downstream sectors
1:25:01export performance so he looks at
1:25:03whether whether sectors that use steal
1:25:05heavily
1:25:06um whether they see an increase in
1:25:09export after differentially after after
1:25:12these industrial policies are put in
1:25:13place and so the the the sort of
1:25:16sobering news is that particularly for
1:25:18the for the less developed countries in
1:25:20the sample these sectors do
1:25:21substantially worse okay so so South
1:25:24Korea is very very much an outlier here
1:25:28um
1:25:29again I'm going to sort of reiterate
1:25:31something that I've that I've that I've
1:25:32said many many times I think that the
1:25:34big question here is how and why is some
1:25:37countries able to do good IP and and and
1:25:40others not
1:25:45um
1:25:46I think I'm actually ready to mostly
1:25:49wrap up
1:25:51um
1:25:52what I would love for you to take away
1:25:54from today is that
1:25:57my reading of the literature
1:25:59is that
1:26:01these papers and some other papers
1:26:03really sort of make a pretty strong case
1:26:06I think that the infant industry
1:26:09mechanisms that we that we've seen in
1:26:12these in in the toy model at the
1:26:13beginning that they're empirically
1:26:15relevant now remarkably this HCI Drive
1:26:18in South Korea managed to engineer a
1:26:20pretty remarkable shift in comparative
1:26:22advantage
1:26:24but we've also talked about both
1:26:26episodes being somewhat unique and
1:26:29somewhat synthetic now there seems to be
1:26:31a lot of things in these episodes that
1:26:34suggest that the Stars really needed to
1:26:35align and come together
1:26:37um when I asked Nathan about this this
1:26:39week about the South Korea paper I know
1:26:41I think I think his stake was very much
1:26:43that this political crisis sort of
1:26:46solved a lot of the political economy
1:26:47problems and created a context in which
1:26:49this policy could be done and it's not
1:26:52clear that the policy would have been
1:26:54implemented
1:26:55um in in in in in in in normal times
1:26:58okay but
1:27:00while I think we need a lot more papers
1:27:03in in evaluating the economic mechanism
1:27:06as I think you've you've probably sort
1:27:09of
1:27:10um surmised from what I've said where I
1:27:12think the gaping hole in the literature
1:27:13is is that we have
1:27:15we have very little information and we
1:27:18have very little data on evaluating IP
1:27:20as it is implemented in the real world
1:27:23and and and that's where
1:27:25Nathan and I with some fantastic
1:27:27co-authors have been trying to sort of
1:27:30think about pushing the literature
1:27:31forward a little bit in particular in
1:27:34getting to these Layer Two questions
1:27:37about real world implementation
1:27:39and to do that I think
1:27:41we feel that there are sort of two
1:27:42things we need to do one is that we need
1:27:44to figure out a better way of measuring
1:27:46industrial policy right the the the the
1:27:48approach in the literature that we both
1:27:50have taken in previous papers has been
1:27:52to study individual contexts that we can
1:27:54understand really well but this is not
1:27:56going to work at scale like we have to
1:27:58find a way that isn't reading off the
1:28:01Tariff schedule that allows us to
1:28:03identify industrial policy and we'd
1:28:05really like to do this
1:28:06for many cases so that we can start
1:28:09comparing things so that we can start
1:28:10looking at you know what does the policy
1:28:13maker really Target when they do
1:28:16um when they do this type of industrial
1:28:17policy and so just to give you sort of
1:28:20the the one
1:28:23um the one sentence overview basically
1:28:25what what we've been trying to do is to
1:28:28build a text-based approach to measuring
1:28:30industrial policy using tools from
1:28:32supervised machine learning and the and
1:28:35the reason for this goes all the way
1:28:36back to slide one industrial policy is
1:28:39goal-oriented action the the the policy
1:28:42maker wants to achieve its long run
1:28:44vision for the economy
1:28:46by by
1:28:48changing what is produced by changing
1:28:51the composition of economic activity in
1:28:52some way and so because of this
1:28:55goal-oriented nature of industrial
1:28:57policy we think the paper trail to look
1:28:59at is not the policy measures themselves
1:29:01the paper trail to look at is to look at
1:29:03what policy makers say when they
1:29:05Implement policy
1:29:07we have a proof of concept that does
1:29:09this I'm not gonna I'm not gonna
1:29:11walk you through the details I've got
1:29:13the slides here and and I'll link the
1:29:14paper in the in the reading list
1:29:16um there are three sort of main
1:29:18takeaways that I think already speak a
1:29:20little bit
1:29:21um to some of the questions we've talked
1:29:22about here
1:29:24one thing is that at least in our data
1:29:26we see that IP is more prevalent in rich
1:29:29countries
1:29:30we see that there's very little in in
1:29:33the least developed economies that we
1:29:34can measure
1:29:36and we also see that these policy tools
1:29:38are typically very very narrowly
1:29:40targeted they're often targeted at
1:29:42individual farms and they're expensive
1:29:45tools the one exception here is that we
1:29:47do see import tariffs being used for
1:29:49industrial policy in lower and middle
1:29:52income countries it's one of the it's
1:29:53one of the top three instruments used
1:29:56and I think that makes a lot of sense
1:29:59um you know this is what is going to be
1:30:01feasible for many of these countries to
1:30:03do they don't have a ton of money to be
1:30:05giving out State loans and trade
1:30:07financing
1:30:08um interest payment subsidies etc etc
1:30:11and then the final thing which goes very
1:30:13much against what we just saw with with
1:30:15with the South Korea paper is that in
1:30:17rich countries we see that
1:30:20um we see that we see that industrial
1:30:22policy is correlated with revealed
1:30:25comparative advantage sectors rich
1:30:27countries seem to be targeting
1:30:28industrial policy towards sectors in
1:30:31which they have a revealed comparative
1:30:32advantage the the effect is not there um
1:30:35for for poor countries we there's
1:30:37basically no effect that we can discern
1:30:40in the lower part of the income
1:30:41distribution
1:30:43so let me wrap up my my conclusions one
1:30:49to twenty that we need more research on
1:30:51Industrial policy
1:30:53um I think we've made some recent
1:30:55progress in identifying the economic
1:30:57mechanisms market failures are
1:30:59definitely important in some some
1:31:01circumstances in ways consistent with
1:31:04infant industry but you know like like
1:31:07like the paper
1:31:08um with with scrutiny and voigtlander
1:31:11shows I think we need sort of plant data
1:31:14and micro studies to help us understand
1:31:16what other frictions might be might be
1:31:19taking place in these new industries
1:31:21that that that the policy makers might
1:31:24be trying to Launch
1:31:25and then I think that the second thing
1:31:27is that I think implementation is going
1:31:29to be the name of the game designing
1:31:31institutions that overcome a lot of the
1:31:33political economy issues that plague
1:31:34industrial policy is key and here I
1:31:37think here I think there's so much work
1:31:39for economists to do you know the
1:31:41analogy with monetary policy is not
1:31:42perfect but we figured out ways to
1:31:45design institutions so that so that
1:31:47policy makers can can can do what what
1:31:50seems in the best interest of the
1:31:51economy in in the long run as opposed to
1:31:53being captured by sort of political
1:31:56um problems
1:31:58two final conclusions I think we should
1:32:00be a lot more open to experimentation
1:32:02and have a lot higher tolerance for
1:32:04failure I think there's no way to do
1:32:06industrial policy in which the policy
1:32:07maker is always getting things right and
1:32:09I really worry that that particularly
1:32:12now that we're entering an era of of a
1:32:14lot of industrial policy people are
1:32:16going to jump on failures you know again
1:32:18the analogy is not perfect but a venture
1:32:20capitalist needs to strike lucky with
1:32:22one out of a hundred and I think you
1:32:25know I I hope the policy maker is going
1:32:26to be better than one out of a hundred
1:32:28but this this is this is hard to do and
1:32:32and failure is almost part of the part
1:32:34of the story
1:32:36um and we should figure out how to again
1:32:38design institutions that allow us to
1:32:40sort of step out of Industries in which
1:32:42in which in which policy is not working
1:32:45and then the final thought I want to
1:32:47leave you with is that as many of you
1:32:49will probably have seen in the news
1:32:51um um the the landscape of globalization
1:32:54is very much changing developed
1:32:56countries in particular
1:32:58um
1:32:59have moved very aggressively towards
1:33:02um a much more sort of protectionist
1:33:05stance
1:33:06and and and and and I think I'm not
1:33:08seeing sort of a lot of discourse on how
1:33:10this changing landscape is going to
1:33:12affect developing countries which I
1:33:14think is particularly important because
1:33:16of the fiscal administrative constraints
1:33:19in many of the developing countries that
1:33:20are going to mean that they're not going
1:33:22to be able to compete with the types of
1:33:24industrial subsidies and and and
1:33:25spending that we're seeing in developing
1:33:27countries you know maybe they don't need
1:33:29to because they're producing different
1:33:30things but we see a lot of talk about
1:33:33sort of the US and the EU and China it
1:33:36would be nice to to have people also
1:33:39thinking about what's going to be
1:33:40happening um at lower parts of the
1:33:42income distribution if the if the world
1:33:43is really moving to
1:33:46um a different type of globalization
1:33:50acknowledging that tariffs are often
1:33:51going to be the only viable tool for
1:33:53doing IP and that these are actually
1:33:55also the most easily observable which
1:33:58which which is important for sort of
1:33:59monitoring and multilateral institutions
1:34:01okay so I know I'm about a time thank
1:34:04you so much
1:34:05um for logging in today
1:34:07and um yeah
1:34:10do industrial policy research
1:34:14Rebecca thank you so much for for doing
1:34:16this for stitching so much together in
1:34:18only 90 minutes and um and for your very
1:34:22important very true in my opinion Call
1:34:25to Arms there
1:34:26um so I just wanted to say a huge thanks
1:34:30um to Brad and the igc for their support
1:34:32and putting together this entire uh you
1:34:34know eight part lecture series we've
1:34:36been running this this month uh to the
1:34:38lectures who um who uh you know donated
1:34:42their their time and their effort to put
1:34:44together those eight lectures uh to
1:34:46Caleb andreasen who's been behind the
1:34:48scenes organizing all the logistics
1:34:50along with uh his team at the igc and I
1:34:53want to remind you that all the
1:34:54materials from this class are on the igc
1:34:56website um as well as on as well as
1:34:58recordings of the lectures are also
1:35:00posted on YouTube
1:35:01um we hope those are useful resources
1:35:03for uh for the next generation of work
1:35:05in this area thanks again for joining us
1:35:07and we'll see you next time
1:35:10thanks again rekka thanks

About this transcript

This page contains the full transcript of Lecture 8: Industrial Policy with Reka Juhasz by International Growth Centre, generated from the public captions YouTube serves with the video. The transcript has 16,254 words across 2,544 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.