00:03 Well,
00:03 since the sandwiches are all gone outside,
00:05 it's probably a good time for us to begin.
00:07 Uh,
00:08 thank you for coming to this,
00:09 uh,
00:09 policy research talk.
00:11 It's a pleasure to welcome Erhan Artu,
00:13 who will be,
00:14 uh,
00:14 presenting to us today on the distributional effects of international trade,
00:18 facts and misconceptions.
00:19 So we'll learn a lot of the former and unlearn
00:22 a lot of the latter thanks to your presentation.
00:25 So,
00:25 thanks,
00:25 Erhan.
00:27 So,
00:31 No worries.
00:33 OK,
00:33 thank you very much.
00:34 So first,
00:35 thank you.
00:36 Uh,
00:37 for pronouncing my name
00:38 perfectly.
00:40 So I would like to,
00:41 I will have lots of anecdotes and,
00:43 uh,
00:43 little examples along the way,
00:44 so I would like to get
00:45 to the heart of the topic immediately.
00:48 So why do we care about the distributional effects of trade?
00:52 So first,
00:53 uh,
00:54 international trade is inherently good
00:56 because it
00:57 gives opportunities to countries
00:59 and workers to take advantage of their competitive advantage.
01:03 So it increases efficiency,
01:05 it creates knowledge spillovers,
01:07 and
01:08 it catalyzes growth
01:10 and increases
01:12 output.
01:13 So
01:14 international trade,
01:15 it is good,
01:16 but it also creates winners and losers.
01:20 So we need to identify the distributional effects because we care about,
01:23 first,
01:24 inequality,
01:25 second,
01:25 poverty,
01:26 and in order to get the policy recipes right,
01:29 we have to
01:31 go to the research papers and analytical tools to
01:34 figure out the exact distributional effects of international trade.
01:38 So my question here is,
01:40 do we have enough analytical work and research papers on this topic?
01:45 So I just
01:46 produced some graphs looking at the data,
01:48 so this is the
01:49 number of papers on trade and labor,
01:52 and these are the number of published papers,
01:54 so I picked 2003 as the starting point because that was the first year
01:58 I submitted my paper
02:00 with
02:01 uh quarters to a journal
02:03 with my,
02:03 and then.
02:05 But as you can see,
02:05 it increased,
02:06 increased,
02:07 increased,
02:07 and like increased 10 times
02:09 over the last 1015 years.
02:12 So
02:12 I would like to give two anecdotes here.
02:14 First,
02:15 around this time,
02:16 uh,
02:17 in 2005,
02:19 the number of papers were so low around uh Bill Maloney,
02:23 he produced 3 papers with his co-authors and it was 20% of the total
02:29 output in the world.
02:31 So,
02:32 and then around 2006 I graduated and I actually had to cite Milli's model
02:38 because I had nothing to do with the Militz model.
02:40 People would keep asking like,
02:41 why,
02:42 what is the relation between your paper and Millit's model,
02:44 and I would say there's no relation,
02:46 so that's why I had to cite it.
02:47 So nobody was,
02:49 there was
02:50 almost no interest
02:51 on the topic apart from maybe a few people,
02:54 and then over the time,
02:56 uh,
02:56 it became.
02:58 Very popular.
03:00 So right now I would argue that we have enough research papers
03:04 and the quality of them
03:05 is,
03:05 I would argue that it's very high,
03:08 but still
03:09 there are lots of
03:10 misunderstandings and misconceptions about this topic.
03:14 So what is the reason?
03:15 So I will first argue that there is one very obvious reason,
03:18 and I would like to give a specific example about that.
03:21 This is David Alter,
03:22 one of my favorite labor economists,
03:24 and this is an interview with him with
03:27 NPR.
03:28 And
03:29 the title of the interior is China killed 1 million US jobs,
03:32 and then don't blame trade deals.
03:36 So
03:36 what do I understand from this title?
03:39 First,
03:40 trade with China is bad,
03:42 the first part of the sentence.
03:44 Second,
03:45 Don't blame trade deals,
03:46 so it is about,
03:48 particularly about the
03:49 Trans-Pacific Partnership,
03:50 so it's not the culprit,
03:52 so we don't,
03:52 we should not blame the policymakers for that.
03:55 But if you read through the interview,
03:57 somewhere in the middle like buried,
04:00 so you'll see that he's actually saying that we
04:02 actually win
04:04 from trade with China.
04:06 So
04:07 the title is kind of misleading.
04:09 It,
04:09 it gives a negative impression about trade.
04:12 Right,
04:12 so
04:13 if you keep reading these type of articles,
04:16 then you will find
04:17 numbers that are much more worse,
04:19 although based on the same paper,
04:22 and you will see that as if the,
04:24 the journalists are competing with each other to give the most negative extreme
04:29 number.
04:30 Right,
04:30 so why is this the case?
04:32 Because they usually want to give
04:34 selective presentation of the results so that they,
04:37 like it's like sensational,
04:38 so it's like outrageous,
04:40 and then
04:41 also the complexity of research magnifies the problem.
04:44 So I would argue that
04:46 despite the Number of many papers,
04:49 there are so many important nuances that we miss,
04:52 so we usually misunderstand
04:54 the results of research.
04:56 So in this talk I will try to give you examples of
04:59 why we misunderstand the research and what what what are the misconceptions.
05:05 So
05:05 there are,
05:06 I picked out,
05:06 of course,
05:07 many of them,
05:08 but I picked three particular
05:10 uh misconceptions that uh
05:13 That I see which is,
05:14 which are very common.
05:16 So what first misconception is measuring
05:18 gains and losses from international trade
05:21 is
05:22 possible because we have lots of recent micro data sets that are available now,
05:26 so we have new statistical tools,
05:28 so we should be able to identify gains and losses accurately.
05:31 So this is not correct.
05:33 Second,
05:34 We can identify the losers empirically using statistical tools,
05:39 but
05:39 it is prohibitively costly,
05:41 so it's so much,
05:43 the losses are so big that we cannot compensate them.
05:46 So this is the second misconception,
05:49 and the third misconception,
05:50 it is a bit more
05:53 complex.
05:54 Is that recent empirical research such as the local
05:58 labor markets approach like the Alter Dorn Hansen paper
06:01 and many others
06:02 are,
06:03 they can give unbiased account of wage and income effect.
06:07 So
06:09 This is also not true,
06:10 so I'm going to argue that
06:12 the papers in the literature are usually biased because of a specific reason.
06:17 So I'm not going to argue that a specific paper is biased,
06:20 but the selection of papers are biased,
06:22 so I'm going to be clear later on.
06:25 So why do we care about these misconceptions?
06:27 Because we would like to give correct policy advice and
06:31 for that we have to understand the limitations of research.
06:35 So in order to do that,
06:37 we have to understand the sources of these misunderstandings.
06:41 So what are the what are the,
06:43 what are,
06:43 what are the main reasons for these misunderstandings?
06:46 The first one is the measurement challenge.
06:48 So I'm going to show you that
06:49 measuring gains and losses is very difficult.
06:52 And
06:53 also
06:53 the second
06:55 problem is ignoring the labor mobility.
06:57 So I'm going to talk a lot about labor mobility in this talk,
07:00 OK,
07:01 so.
07:02 Let's start with the first source.
07:04 So why measurement is
07:06 difficult?
07:08 Because
07:09 international trade essentially has two direct impacts in the short run.
07:14 The first one is the income effect,
07:16 so wages change.
07:17 So if I'm in the import competing industry,
07:19 my wages will go down.
07:21 If I'm in the exporting industry,
07:22 my wages will go up.
07:24 And that will change my income,
07:26 and it could be
07:27 the case that the crop prices like price of rice or wheat could change,
07:31 so it might have some effect for the
07:33 farmers.
07:34 And third,
07:35 there might be some consequences for the government revenue.
07:38 So this is the first channel.
07:41 And the 2nd channel is the consumption channel.
07:44 The price of imported goods usually declines,
07:47 so that means the price of my consumption,
07:50 the price of my shopping basket
07:52 actually goes down.
07:54 So that's a positive effect.
07:56 So the problem here is
07:58 measuring the first channel is much easier
08:01 than measuring the second channel,
08:03 so that will create a bias in the,
08:05 in the research.
08:07 The second point,
08:09 why labor mobility matters?
08:12 OK,
08:13 so
08:14 labor mobility is important because it determines the
08:17 uh
08:18 income effects,
08:19 so it determines the wages and also
08:22 it
08:23 gives identification strategies
08:25 uh.
08:27 If
08:28 there are mobility frictions,
08:29 we can identify gains and losses.
08:31 Without mobility friction,
08:32 identification is impossible.
08:34 So what does it mean?
08:36 So let's think about two scenarios.
08:38 In the first scenario,
08:39 let's think that,
08:40 assume that the mobility frictions are very high.
08:43 So if the mobility frictions are high,
08:46 let's say that I'm in the import competing industry.
08:50 In that case,
08:51 after a trade liberalization shock,
08:53 my wage will go down,
08:55 right?
08:55 If my wage will go,
08:56 goes down,
08:57 and if I'm in the exporting sector,
08:58 my wage goes up,
09:00 so that will be clear winners and losers based on.
09:04 The trade shock,
09:05 so we can identify
09:07 and we can link
09:08 the trade liberalization shock to wages
09:10 in exporting and import competing sectors.
09:13 So that gives us an identification.
09:16 And also if
09:18 the wages decline in the import competing sector,
09:21 workers will switch to the exporting sector
09:23 and then the wages would equalize in the second scenario
09:26 if there are no mobility frictions.
09:29 OK,
09:29 in the second scenario,
09:31 the wages in both sectors will equalize,
09:34 so there will be no variation,
09:36 no cross-sectional variation in the data,
09:38 so we cannot link
09:40 the trade liberalization to changes in wages.
09:43 So there is no way to run some empirical analysis
09:45 or statistical analysis because there will be no effect.
09:49 And
09:50 in that case,
09:52 we cannot really talk about any empirics or any statistics.
09:56 So that's why
09:57 in order to understand gains and losses,
09:59 we have to think about labor mobility
10:01 and to measure it accurately.
10:05 OK,
10:06 so it matters for both transmission of the
10:09 shocks and also identification of gains and losses.
10:13 OK,
10:13 so
10:15 what do we need to do
10:17 to inform
10:18 the policy makers?
10:19 So
10:19 like I would like to write a research paper to inform the policy makers.
10:23 So what are the things that I should be careful about?
10:26 So first of all,
10:28 I have to measure
10:30 the gains and losses carefully.
10:33 So,
10:33 and then as I said before,
10:34 if the mobility
10:36 is high,
10:37 it will be very difficult to do it empirically using statistics.
10:41 So in that case,
10:42 we have to rely on a model or some simulations.
10:45 So in order to inform the policy makers,
10:47 we have to
10:48 have some model or simulation or
10:51 more recently we call this quantitative analysis,
10:53 so we have to have quantitative tools.
10:56 Second,
10:58 Of course,
10:58 the tools
10:59 based on models are not super reliable because
11:01 they are not based on direct evidence,
11:03 so we cannot really have show statistical
11:06 relation or causal link between two
11:09 impacts.
11:09 So
11:10 to
11:11 make sure that we have direct evidence,
11:13 we need statistics and econometric analysis.
11:15 So we have to complement the simulation models
11:18 with statistical models like empirical models so that
11:21 we can have both sides of the
11:24 story.
11:25 And then,
11:25 as I said before,
11:26 we have to also think about labor mobility.
11:29 So ideally
11:30 you should have
11:32 3 different types of research
11:34 to inform the policymakers,
11:35 so they will be complementing with each other.
11:38 So what I'm going to do today,
11:39 this is also the outline of this talk,
11:42 I'm going.
11:42 Talk about 3 example projects.
11:45 One of them is household impact of tariffs,
11:48 exports to jobs,
11:49 trade jobs,
11:49 and worker welfare.
11:50 So these 3 projects will be examples of each of these cases,
11:55 and I'm going to show in the last project how we can combine all of them.
12:00 So this is the
12:01 outline.
12:03 OK,
12:04 so
12:04 I'm starting with the first project.
12:07 Uh,
12:07 it is mostly about the channels,
12:10 the income channels and the consumption channels,
12:12 and it relies
12:14 on data
12:15 on farmers.
12:16 So that's why I picked the rice here as an example.
12:19 Uh,
12:20 so,
12:20 OK,
12:21 so what are the direct effects?
12:23 So
12:23 the direct effects of
12:25 a trade shock such as a decrease in tariffs,
12:29 I picked rice as an example.
12:33 First,
12:33 let's say that the tariffs go down,
12:36 then the rice price declines.
12:39 And
12:39 if the rice price declines,
12:43 As consumers.
12:45 Price is cheaper for us,
12:47 so we are better off,
12:49 so there's a positive impact on the consumption.
12:52 But how about the producers?
12:54 Since the rice prices are lower,
12:56 now the producers are worse off.
12:59 So depending on how much I produce rice and how much I consume rice,
13:04 depending on whether I'm a
13:05 net consumer or producer.
13:08 My welfare could go up or down.
13:11 OK.
13:11 So this is a very simple
13:13 example.
13:15 So in this project
13:16 with Guido Porter and Bob Rikers.
13:20 We have a simple framework where production
13:23 and consumption decisions are fixed,
13:26 the labor
13:27 is fixed,
13:27 all factors are fixed,
13:29 land allocation is fixed,
13:31 and this is a very good assumption for the short run.
13:34 And the
13:37 The power of this assumption is that
13:40 In the short run,
13:41 if all decisions are fixed,
13:43 the only thing I need is
13:45 income
13:46 and
13:46 consumption shares by industry.
13:49 So what we did is we collected household surveys,
13:53 the harmonized household surveys for 54 countries
13:57 with focused on agriculture.
13:59 We combined them with tariff data.
14:03 And then
14:05 we
14:06 design some web page,
14:07 web-based simulation tools.
14:09 So essentially what we can do using this data set,
14:12 you can use the data set online
14:14 and you can just run your own analysis,
14:16 but what else you can do,
14:17 you can
14:18 uh
14:19 pick a country,
14:20 this is the website of the tool.
14:23 And the paper and the,
14:25 the toolkit is here,
14:27 and data is here as well.
14:29 And,
14:29 OK,
14:30 so what you can do,
14:30 for example,
14:31 you can pick a product.
14:33 So here I picked
14:35 rice as an example
14:37 and then I picked Ghana,
14:38 so you can basically pick any country,
14:40 any country combination,
14:41 any product
14:42 among the 50 different product categories that we have
14:46 in any combination and you can just feed in any shock essentially
14:49 and each of these dots are households,
14:51 so you can calculate the impact of a
14:54 trade shock.
14:56 On
14:57 particular households,
14:58 so
15:00 This is just an example
15:01 and directly taken from the website
15:03 that
15:05 That I just ran before the presentation,
15:07 so.
15:09 As you can see,
15:10 the
15:11 poor households,
15:12 this is the per capita income,
15:14 and this is how much they are better off or worse off,
15:16 the Y axis,
15:17 so each of these dots,
15:19 they are
15:20 uh households.
15:21 So as you can see,
15:23 the ones around here.
15:25 The households that are poor
15:27 are more likely to be worse off.
15:29 The richer
15:30 households and middle income households are more likely to be better off.
15:34 So the reason is simple.
15:35 So these households are more likely to be producers of rice,
15:40 and
15:40 they are net exporters in Ghana.
15:44 They are net sellers of rice in Ghana.
15:45 About 50% of the rice
15:48 consumption
15:49 is
15:51 supplied by the local
15:53 producers,
15:54 and about 50% is imported.
15:57 So here you can see.
16:00 Some households are worse off and some households are
16:02 better off and it changes the income distribution.
16:06 OK,
16:07 so.
16:08 The two channels that we have,
16:10 production channel
16:11 and the consumption channel
16:13 will tell us
16:15 some.
16:16 Uh,
16:17 story,
16:17 so
16:18 the first misconception.
16:21 So measuring the gains and losses from international trade is
16:25 easy
16:25 or
16:27 possible
16:28 because of the recent micro data sets.
16:31 And this is not true.
16:32 Why this is not true?
16:34 Because it is much more difficult to
16:36 measure the consumption channel channels compared to
16:39 the income channels.
16:40 The income effects are concentrated and the consumption effects are diffused.
16:45 So
16:46 how can I show this using this
16:48 household impacts of tariff database?
16:50 So I went to the database
16:52 and just
16:53 plot impact of overall.
16:57 Tariff cuts for all 54 countries in the data sets for all products that we have,
17:02 and I put some
17:04 uh polynomial fit
17:06 for
17:07 income effect and consumption effects.
17:09 The income effects are red
17:11 and the consumption effects are green
17:13 and the total effect is the sum of 2 is the blue line.
17:17 So as you can see,
17:18 the income effect,
17:19 the red line
17:21 varies a lot.
17:22 Across households,
17:24 but the consumption effect is more flat.
17:27 So what happens if I try to run some statistical analysis using this data set,
17:31 uh,
17:32 if I don't have the model or simulations,
17:35 I would not be able to identify
17:38 the consumption effects,
17:40 the green effect,
17:41 because there is less variation
17:42 and it will be much easier for me
17:45 to
17:46 take the income effects into account.
17:48 So
17:49 what will happen?
17:51 When I run regressions,
17:52 I would find
17:53 large impacts
17:55 on
17:57 income,
17:57 but small impacts on consumption,
17:59 so that will create a bias and
18:01 all the empirical work,
18:03 as
18:03 David Alter said in the interview,
18:06 the consumption effects are diffused,
18:08 so they are difficult to identify.
18:11 OK,
18:11 so this is misconception number one.
18:17 OK,
18:17 the
18:18 2nd project.
18:21 Uh,
18:21 now we are going to
18:23 South Asia.
18:24 This is,
18:24 uh,
18:26 a textile worker.
18:29 So
18:30 the
18:31 This is about the local labor markets,
18:33 and it is probable,
18:34 I will argue that
18:36 the,
18:36 in the recent years,
18:38 the most popular empirical
18:40 method in the
18:42 In the trade literature.
18:44 So what is the idea here?
18:45 The idea is
18:47 there are some regions
18:49 uh which are heavily concentrated in export,
18:52 in import competing industries,
18:54 so like the Rust Belt in
18:56 uh US
18:58 or it could be
19:00 uh
19:01 other regions which you have
19:03 heavy concentration of imports.
19:06 So
19:06 trade liberalization will affect
19:08 different regions differentially.
19:11 It could be a commuting zone
19:13 in the US
19:14 or it could be a Zilla
19:16 in India or it could be a micro regionion
19:19 uh in Brazil.
19:21 So most of the papers in this literature about the growth of China
19:25 and the impact of
19:27 US employment and wages.
19:28 So this is the famous author Dawn Hansen paper.
19:31 And there are also other papers about developing countries.
19:35 So one example is India by Topolowa and another example is COA
19:40 for Brazil,
19:41 and these papers usually look at a
19:44 Trade liberalization episode like a tariff cut and
19:47 try to figure out what happens to employment,
19:49 wages and poverty.
19:51 OK,
19:51 so
19:52 all these papers are mostly about the negative effects.
19:56 Why they are about negative effects?
19:57 Just because of the,
19:59 the problem I just mentioned,
20:00 because it's much easier
20:02 to identify the negative effects.
20:05 So what we did in this project,
20:07 we flipped the question.
20:09 So
20:09 what happens
20:11 if exports to OECD countries affects wages and employment in South Asia
20:15 and
20:17 the
20:18 The,
20:19 the better data was available for India,
20:21 so a significant part of the
20:23 book focuses on India.
20:24 So it was a joint work with
20:26 uh Gladys Lopez Acevedo,
20:28 Ray Robertson,
20:29 and Daniel Saman.
20:31 OK.
20:32 So I would like to just like explain what we are trying to do in this
20:36 project and
20:37 this chart,
20:37 although it's for this particular project,
20:39 it
20:40 basically explains
20:42 the intuition behind all these local labor market papers.
20:46 So what we have is
20:49 some trade shock in South Asia,
20:51 let's say that exports
20:53 and you would like to map these changes to
20:57 Local labor market employment outcomes such as wages,
21:00 informality and employment.
21:02 OK,
21:02 so what you would do,
21:04 you would just run some statistical analysis,
21:06 but the problem is
21:08 the
21:09 local labor market conditions might be affecting.
21:13 The exports,
21:14 so you cannot really use exports of India as an explanatory variable.
21:18 What you need to do
21:20 is you need to find an instrument
21:22 that is exogenous to India
21:24 or South Asia.
21:26 So what we have here is the import demand
21:28 of OECD countries from all countries apart from South
21:32 Asian countries and
21:34 use that
21:35 to discipline
21:36 the direction of causality in the data.
21:39 And then what we find is that we find that the export.
21:43 of
21:44 South Asia.
21:46 Caused by the import demand of OECD countries
21:50 actually have significant effects on informality and employment.
21:54 So
21:56 how do we present the results?
21:57 So this,
21:58 actually these two
22:00 Tables
22:01 or charts show
22:03 precisely what is great about this type of research,
22:06 local labor markets approach,
22:08 and also what is not so great about this research.
22:11 What is great
22:12 is that I have lots of different worker types here.
22:15 So
22:17 we have male,
22:17 female,
22:18 low skilled,
22:19 higher skill,
22:19 young,
22:20 old,
22:20 so there's a lot of heterogeneity,
22:22 and,
22:23 but I have to present the results in relative terms.
22:26 So this is
22:28 The 75th percentile exposure relative to 25th percentile exposure,
22:33 so what is the difference
22:35 of azilla in the 75th percentile relative
22:38 to 25th percentile of exposure to exports?
22:44 And we find that
22:46 the effect is especially large for
22:49 female.
22:50 So this is how much the informality decreases thanks to the exports.
22:56 So this is important because
22:57 that means workers switch from
23:00 informal jobs to formal jobs,
23:02 and they are better jobs.
23:04 OK,
23:04 so
23:05 we were not able to find any effect on the total number of employment,
23:08 but we show.
23:10 We found effects about the quality of jobs,
23:14 so the exports increases the quality of jobs in India.
23:19 And
23:20 how about wages?
23:22 So we also find that
23:23 there is a heterogeneous impact on the wages.
23:26 So
23:28 for example,
23:28 we found that
23:29 the wages in rural regions declined relative to the
23:36 25th percentile and the wages in the urban regions actually increased,
23:40 so it changes the wage composition
23:42 and female workers,
23:44 women,
23:45 and
23:46 higher skilled workers are especially better off.
23:51 So this is the,
23:52 the snapshot of the results from this approach.
23:57 But
24:02 Let's see if we can interpret the results,
24:04 so.
24:07 Here,
24:08 the common misconception about this type of research is that we can identify losers
24:13 using empirical research,
24:16 empirical analysis,
24:17 but we cannot really compensate them because the,
24:19 the losses are very large.
24:23 So this is not true.
24:24 Why it's not true?
24:25 Because we don't really have a control group.
24:28 International
24:29 trade affects all regions in India or in the US,
24:32 and because of that,
24:34 we can only tell the relative difference across regions,
24:37 but we cannot really know what happens in the aggregate.
24:42 There are also other issues like migration is not
24:45 accounted carefully and the new technologies or ideas can be
24:49 creating dynamic effects and creating new opportunities,
24:52 which is also difficult to account for.
24:54 So
24:55 what do I mean by that?
24:57 You see that this is rural workers are worse off,
24:59 urban workers are better off.
25:01 It could be the case that both are better off,
25:04 but rural workers are less better off than urban workers,
25:07 so we don't really have
25:09 an aggregate number.
25:10 We don't really know the mean.
25:11 We just know
25:12 the relative effect
25:14 to the to the compared to the mean.
25:16 OK,
25:16 so this is problem number one.
25:20 Problem number 2,
25:22 can we
25:23 really compensate them?
25:24 So you will think that compensation is very costly.
25:27 So I picked uh
25:29 like a more popular paper,
25:31 the most popular paper in this literature,
25:33 the China syndrome paper.
25:35 It shows that
25:37 $1 additional imports from China
25:40 accounts roughly for 4% decrease in
25:43 income of workers,
25:44 and this includes
25:47 not only wages,
25:48 but also employment loss.
25:51 Uh,
25:51 so it is the,
25:52 it is the expected impact on
25:56 US workers and how the wage bill declines.
25:59 And if you calculate with this,
26:01 with the imports from China,
26:03 it will be $21.5 billion US dollars per year.
26:07 So,
26:07 OK,
26:08 this number is huge,
26:09 so,
26:10 but
26:11 would the US be able to compensate these workers?
26:14 The GDP is about $20 trillion
26:17 so 0.1% tax
26:20 will be enough
26:21 to
26:22 compensate all losers from
26:25 Trade with China.
26:26 So and this is an anecdote
26:28 from Bill Gates.
26:29 He said that he paid $1 billion in taxes and he's happy to pay 20 billion.
26:34 So
26:35 Bill Gates himself
26:37 with his taxes could compensate all workers,
26:40 but not only that,
26:41 he could compensate all workers because
26:44 when he compensates,
26:45 the losers will not be.
26:48 Indifferent,
26:49 they will be just as better off as the average guy.
26:53 So he could actually make everybody unanimously better off.
26:57 By just the tax amount that he wants to pay.
27:02 OK,
27:02 so.
27:04 OK,
27:05 so let's move on to the
27:07 project.
27:08 Number 3,
27:10 and this is uh Favela in Brazil.
27:12 Uh,
27:13 so we start with agriculture
27:15 for many countries and move to South Asia,
27:17 then we are going to,
27:18 I'm going to finish with,
27:19 uh,
27:20 Brazil.
27:21 Uh,
27:22 so,
27:23 The facts and misconception,
27:25 the first one was,
27:26 I think like
27:26 kind of obvious.
27:27 The second one was a bit more nuanced,
27:29 and
27:29 this is a much more
27:32 sophisticated misconception that like it's,
27:35 it's much more difficult to,
27:37 to see.
27:39 So
27:41 In the recent empirical literature,
27:42 like
27:43 local labor markets papers and
27:46 Uh,
27:47 we
27:48 The results in this,
27:49 in,
27:50 in the literature is unbiased so it can give unbiased
27:53 effects of,
27:54 it can give unbiased effects of income and wage effects.
27:58 So is that true?
28:01 So
28:02 if you look at
28:03 the local labor market effects,
28:06 You will see that they are only identified when there are very high frictions.
28:11 So what does it mean?
28:13 So
28:13 if the frictions are high,
28:16 let's say that I'm in the import competing sector
28:18 and I can't move to the exporting sector.
28:21 Then
28:22 when the trade shock hits me,
28:23 my wages will go down,
28:25 so I cannot really change my sector.
28:27 So
28:28 there will be a difference between my wage
28:31 and the wage of the person who is in the exporting sector.
28:33 So there will be a wage differential.
28:35 There will be differentials in the
28:37 unemployment rates in those two
28:39 industries or regions.
28:41 So what does it mean?
28:43 That means first,
28:44 as an econometrician,
28:46 I can identify this impact and link it to trade.
28:50 But at the same time it means that
28:53 the gains from trade is low
28:55 because ideally
28:56 you want exporting industries to grow.
28:59 So if they cannot attract workers from the import competing industries,
29:03 they cannot grow
29:04 and the gains from trade will be limited.
29:07 So here's the thing,
29:08 when the gains from trade is low,
29:11 the
29:13 Economists can identify the effects of trade.
29:17 And when the
29:19 Gains from trade is high,
29:21 there is no effect,
29:22 statistically identifiable because there's no variation.
29:26 So that means
29:27 the research papers in the literature
29:29 will be selected,
29:31 not randomly
29:33 for the cases where you have
29:36 larger labor market frictions and low gains from trade.
29:39 So that means the entire literature will be biased if you just count
29:43 the look at the effects,
29:44 so you can only identify them
29:46 when the things are bad.
29:48 OK,
29:49 so what do we do?
29:50 So I just want to mention this
29:52 because I find it interesting,
29:53 although I'm not going to mention any equation or I'm not going
29:56 to show you any math or algebra or anything like that.
29:59 So in order to account for these
30:01 dynamic effects,
30:02 we have to
30:03 calculate
30:04 the labor flow equation.
30:06 So we have to understand how workers change industries,
30:09 regions,
30:09 occupations,
30:10 and it's a very complicated process.
30:12 So as economists,
30:13 what we do is
30:15 we just borrow.
30:17 Equations from physics.
30:19 So the,
30:20 in the physics literature,
30:21 they found
30:22 equations that
30:24 identifies the flow of liquids or the flow of heat,
30:28 which is called Euler equations and we just take
30:30 them like we just steal the equations from them
30:32 and apply it to economics.
30:35 So
30:36 what
30:37 new insights this brings us?
30:39 So,
30:40 OK,
30:41 let's think about a region.
30:43 This is
30:44 a region that is exposed
30:46 like the Rust Belt in the US
30:49 and then there's a trade shock.
30:50 Let's say that
30:52 region A.
30:54 Is competing with
30:56 uh
30:57 with foreign producers
30:58 and there's a trade liberalization episode.
31:01 And the trade shock
31:03 decreases wages and number of jobs in this region.
31:07 So then what will happen?
31:09 So normally,
31:11 if you just use empirical tools,
31:14 you will look at this region
31:16 and you will look at this region,
31:17 region B,
31:18 which is not exposed.
31:20 And then you will find variations across these
31:22 two regions and try to link it with trade
31:24 because
31:25 this region is exposed and this is not exposed.
31:28 But what happens in reality?
31:30 In reality,
31:31 there is some mobility across these
31:34 two regions,
31:34 reallocation of resources.
31:36 The workers would move from region A to Region B,
31:40 which would reduce wages in region B and increase wages in region A,
31:44 so the wages
31:45 will be somewhat equalized,
31:47 not perfect at all.
31:49 And then the number of jobs will change.
31:51 So if I can find a way to take
31:53 these flows
31:55 into account
31:56 like the flow of liquid
31:57 to
31:58 these two buckets,
31:59 I can really calculate the actual impact of trade.
32:02 So this is what
32:03 we are trying to do in this
32:05 research.
32:07 So I have several papers
32:09 uh based on this research agenda.
32:10 This is with
32:11 uh Paulo Bastos
32:13 and Yung Lee.
32:15 So what we do
32:16 is using Brazilian data,
32:17 we develop a model
32:19 that combines
32:20 empirics
32:21 uh theory and transition dynamics.
32:24 So it has some backward-looking analysis,
32:27 so we have to show
32:29 how
32:30 the past
32:31 experience of Brazil.
32:33 And link it to trade so that we can give some
32:35 statistical evidence and we can make sure that the model is
32:39 not completely crazy
32:40 and then we can also do simulations to
32:43 account for different channels correctly.
32:45 So these are the two things that we do
32:47 in,
32:47 in this paper.
32:49 So
32:50 since I don't want to go into any of the
32:53 uh details,
32:54 I just want to show you a few results and what we can do with this type of analysis.
32:59 So
33:01 This is.
33:04 Exposure to import competition of different micro region
33:08 and this is a relative rate change.
33:10 So as you can see,
33:12 as you are more exposed to import competition,
33:15 the lower the wages are.
33:17 Right,
33:17 so,
33:18 but
33:19 I cannot really put
33:21 numbers there
33:22 because everything is relative.
33:24 So even if I put some number that wouldn't mean anything,
33:27 right?
33:28 So this is using empirics.
33:31 What about
33:32 the distribution of income after the trade shock?
33:36 So we have the number of microegions
33:39 and the relative weight change,
33:40 and after the shock I plot the
33:43 weight change histogram,
33:44 and again on this time on the x-axis,
33:46 I cannot put any numbers
33:48 because I don't know on average if the workers
33:50 are better off or worse off using statistical tools.
33:53 So what I need is I need to combine this
33:56 with some sort of model,
33:57 so I have to combine empirics and model together.
34:00 And now I can tell you that
34:03 this is 0,
34:05 so I can tell that more workers are better off.
34:08 And
34:09 a lot of workers are worse off based on the rate change.
34:13 OK,
34:14 so,
34:15 but here's the thing,
34:16 so this is
34:17 a wage change effect in the short run,
34:20 but the workers are not stuck.
34:23 So
34:23 even if I'm here,
34:25 my wage decreases,
34:27 I could go and find a job in another industry
34:30 or in another region possibly,
34:32 but I have to take the frictions into account to calculate
34:35 the present
34:37 discounted value of wages or the welfare of a worker.
34:40 So I have to take the workers' mobility and the dynamics into account.
34:45 So what happens,
34:45 I'm not going to show you uh any of the equations or
34:48 uh the dynamic graphs,
34:49 but
34:50 I could calculate
34:52 the impact,
34:53 the welfare impact of this trade shock,
34:55 and I can show that
34:57 the wages
34:58 in micro regions on average increase everywhere
35:02 in Brazil after
35:03 uh import,
35:06 uh after uh.
35:07 Tariff liberalization shock,
35:09 but
35:10 this is
35:12 Not
35:13 true for all workers.
35:14 So if you just slice the data more carefully and look at
35:18 workers in different industries,
35:20 the workers who are
35:21 in the
35:22 import competing industries are actually
35:26 Many of them are worse off,
35:27 so in order to compensate or in order to find a policy,
35:30 we have to
35:31 have this type of graphs and understand
35:33 how each worker,
35:35 how each worker is affected from trade shock.
35:38 OK,
35:39 so,
35:40 uh,
35:41 OK,
35:42 so now the important part.
35:45 So I'm glad that I was able to go fast so that I have enough time for this uh
35:49 uh part,
35:50 which is like,
35:50 I think more interesting and
35:53 I would like to ask like a few questions.
35:55 Uh,
35:56 and I'm not going to pretend that I know all the answers,
35:59 but I,
35:59 I would like to share how I think about these policy questions with you.
36:04 So
36:05 The first question.
36:09 Should we really care why the workers are worse off?
36:13 So
36:14 it could be international trade.
36:17 Or it could be some other reason and a worker
36:20 is worse off.
36:21 So do we really care if it is technology or trade or macroeconomic
36:26 shocks?
36:26 So why do we care what is so special about international trade?
36:30 So
36:34 International trade
36:35 can be attributed to policy
36:38 and then you can actually,
36:39 the workers can blame policymakers,
36:42 so,
36:42 but it's more difficult for them to blame.
36:45 Uh,
36:47 technology,
36:48 so
36:49 for that reason,
36:50 maybe we should care more about international trade checks because
36:53 that is
36:54 something that you can remedy with policy.
36:57 Or that's created by policy,
36:59 so that's.
37:01 One question.
37:03 Second question.
37:04 Should we compensate workers if we choose to compensate
37:08 based on
37:10 Relative effects or absolute effects.
37:13 So
37:14 this question is actually a bit more,
37:16 sounds like a very trivial question,
37:17 but it's a bit more
37:19 uh sophisticated than that because.
37:22 If as a stati statis if as an econometrician or an economist,
37:26 I have difficult time identifying the absolute effects
37:30 and I can.
37:32 Much,
37:32 it's much easier for me to identify the relative effects
37:36 if I think from the worker's perspective,
37:38 they face the same challenge.
37:40 So a worker,
37:42 for a worker,
37:43 it will be more difficult to identify the relative effects,
37:46 uh,
37:46 identify the absolute effects and it will be much easier
37:49 uh to identify the relative effects.
37:52 So
37:53 for many cases,
37:55 workers might care more about
37:57 relative effects compared to absolute effects because
37:59 it's much easier for them to identify.
38:04 And second,
38:05 there might be some behavioral differences,
38:07 so there might be
38:08 maybe they don't care about absolute effects but they just care about
38:12 uh
38:13 relative effects.
38:13 So we don't really know the answer for this.
38:17 Third,
38:19 how should we finance compensation?
38:21 So I just showed that for US it
38:23 is possibly feasible because it's a developing country,
38:26 they can collect taxes easily,
38:28 but unfortunately,
38:30 many developing countries have to rely on
38:33 tariff revenue
38:35 for the government budget.
38:37 So if you liberalize,
38:39 then how are you going to compensate the workers because
38:41 the only thing you have is the tariff revenue,
38:44 and then you lose it
38:46 by liberalizing trade.
38:47 So this is a dilemma
38:49 that
38:50 developing countries would face
38:52 when they try to finance their compensation policies.
38:55 It will be much more difficult for them.
38:59 Uh,
39:00 here are some
39:01 like quick answers,
39:02 so it's not like
39:03 the
39:04 uh
39:05 definitive answers,
39:06 but
39:07 some thoughts.
39:08 So we need to identify ideally relative and absolute effects
39:12 and it looks like
39:13 one piece of research.
39:15 Would identify one side and another piece of research would identify another side.
39:20 So you really need to have different types of
39:22 research
39:23 and
39:24 dialogue between
39:25 researchers and policymakers to actually
39:28 have a good identification and quantification of the effects.
39:30 It's not an easy task.
39:33 Second.
39:36 Adjustment speed will matter.
39:37 So since workers
39:39 can and
39:40 can adjust to the shocks,
39:42 if you have long term plans
39:45 rather than having
39:46 lots of little changes
39:48 over time,
39:48 if you have a plan and if you announce it in advance,
39:51 it will give time to
39:53 workers and firms to adjust in advance,
39:55 so it could
39:56 increase the chances of unanimity
39:58 for the trade policy.
39:59 So
40:00 rather than shock therapy.
40:03 I think it will be much better to have planned policies.
40:07 And the
40:09 more difficult question,
40:10 how should we compensate?
40:12 So basically you can have two different types of policy.
40:15 One targeted,
40:17 you can target precisely
40:19 the worker.
40:22 Who is worse off because of trade,
40:24 and for that
40:25 you have to think about workers' industry
40:28 because import competing and exporting industries will be different,
40:31 but also location of the worker matters.
40:34 So in some cases,
40:36 some regions,
40:37 even though the worker is in let's say services sector,
40:40 he could be worse off.
40:41 So his region
40:43 and industry would matter also.
40:44 Worker characteristics.
40:46 Young workers could be much
40:48 more adaptable to new situations compared to older workers,
40:50 so you have to also think about
40:52 the worker characteristics
40:54 or
40:54 alternatively you could have
40:57 a system
40:58 like flexicurity in Denmark
41:00 which basically compensates based on
41:02 anything,
41:03 so they don't care
41:04 the reason for the
41:05 unemployment or job loss.
41:08 But of course
41:09 these policies could be feasible in Denmark,
41:11 but it will be much more difficult for
41:13 developing countries.
41:17 OK.
41:18 So,
41:19 conclusion.
41:20 So,
41:21 uh,
41:22 identifying losers
41:24 is much easier than winners,
41:26 and this is a problem in the literature,
41:28 so we see,
41:29 seem to see the losers more
41:31 frequently
41:32 and the results are relative and
41:34 context-specific,
41:35 so they are very difficult to interpret.
41:38 Compensation,
41:39 as I showed,
41:39 could be feasible,
41:40 but it will be very difficult
41:42 for
41:43 Developing countries and identification is always a challenge,
41:46 and the distributional effects of trade is complex,
41:50 and there might be workers who are severely affected from trade shocks,
41:54 but protectionism does not seem to be the answer.
41:57 OK.
41:58 Thank you very much.
42:05 Thanks very much,
42:06 Arahan.
42:06 Um,
42:07 it's a pleasure now to both introduce and welcome,
42:10 uh,
42:10 Bill Maloney,
42:11 who is the chief economist in EFI and
42:14 has a long and distinguished career in the bank,
42:15 a highlight of which,
42:16 of course,
42:17 has been our former colleague in,
42:19 in,
42:20 uh,
42:20 in the research department.
42:22 Um,
42:22 so it's a pleasure to welcome Bill and,
42:24 you know,
42:25 5 to 10 minutes for,
42:26 uh,
42:26 reactions,
42:27 please.
42:30 I will be brief
42:34 Um
42:35 Not exactly sure why of all the talented trade economists in the bank,
42:39 I
42:40 got this privilege.
42:41 I can't believe it had to do with 3 papers in 2005.
42:44 Um,
42:45 I think it has more to do with the fact
42:46 that Erhan knows that I have a great appreciation for,
42:50 uh,
42:50 the kinds of things he's trying to do in this literature
42:53 and
42:54 Lest his presentation,
42:55 OK,
42:55 this paper,
42:56 this paper,
42:56 this paper,
42:57 come up a part of,
42:59 come across as too fragmented,
43:00 I just wanted to locate it
43:02 within
43:02 the literature that I see and then
43:04 um
43:05 see where we might push it forward.
43:07 So the context,
43:09 long-term problem of thinking about trade and inequality.
43:13 In the beginning,
43:14 uh,
43:15 there was Stopper Samuelson,
43:16 1941.
43:18 Useful framework.
43:20 Ashley could describe it to her mum.
43:23 Uh,
43:23 it's a 2 by 22 by 2 by 2 Edgeworth box,
43:26 very simple,
43:27 labor adjusts instantaneously,
43:29 and the intuition is clear.
43:31 The only problem is no one ever really liked it.
43:33 Though we keep using it,
43:34 it's worth remembering that it was rejected that year
43:38 by the AER
43:39 for having nothing to say about the real situations in
43:41 which the theory of international trade has to concern itself.
43:45 Um
43:46 and uh still,
43:47 it was used as the workhorse model of thinking about these questions
43:51 well through the 90s.
43:52 Um
43:54 maybe Davis and Mishra were the ones who pronounced it
43:57 uh formally
43:58 dead in 2007.
44:01 Um,
44:01 and what was it killed by,
44:02 it was killed by a bunch of stuff.
44:04 It was killed by introduction of intermediate inputs and models,
44:07 imperfect substitution of domestic and imported goods,
44:10 the continuum of goods,
44:12 um,
44:13 adding many countries,
44:14 heterogeneous firms,
44:15 and one thing that
44:17 Erhan has focused a lot on is exactly specific factors in labor immobility.
44:21 And this is kind of key,
44:22 because as long as you're assuming
44:24 that labor instantaneously adjusts across sectors,
44:26 then any shock,
44:27 as he said,
44:28 you,
44:28 you know,
44:28 there are no winners and losers because,
44:30 well,
44:30 let's put it this way,
44:31 among labor,
44:32 because you only have one kind of labor,
44:33 there's no winners and losers within labor because
44:36 wages are equated instantaneously across sectors and everybody's happy.
44:40 Everyone knew this was a lousy model from the beginning,
44:42 and that's why we went into
44:44 assumptions of the opposite end of the scale,
44:46 which was that uh.
44:48 That
44:49 labor was not mobile at all,
44:51 and this kind of came out of the observation that according to Stopper Samuelson,
44:55 labor and capital should always be on the opposite side of any political fight.
44:58 And in fact
44:59 we find them always on the same side of the fight.
45:01 They don't want a plant closing and so on and so,
45:03 and because of tariffs falling.
45:06 So both capital and labor get to protect themselves.
45:08 That makes sense if labor can't instantly go to some
45:11 other job that would pay as well or better.
45:14 And so we got the specific factors model out of that,
45:17 and that was the 70s.
45:18 Um and that's very useful,
45:19 and we all saw that in trade theory too,
45:21 and that was great too,
45:21 except
45:22 it's also at the other extreme.
45:24 Um,
45:25 so
45:26 that's one big problem we've had to deal with is how
45:28 do we deal with gradations or differing types of labor mobility
45:32 and how do we deal with the process through which labor markets adjust,
45:36 and that is where
45:37 this literature that Erhan is working in,
45:39 part of where it's centered,
45:41 and I think it's very important for that.
45:43 The other thing that I think is very important
45:44 is that we've all heard of the China shock,
45:47 but the China shock has also been a shock to empirical work.
45:50 We have to explain why we lost 5.8 million manufacturing jobs from 1999 to 2011,
45:56 and the biggest guns in the literature
45:58 are not in agreement.
46:00 So
46:01 the common
46:01 perception before was,
46:03 well,
46:03 we know we don't have the right model,
46:05 and here and there,
46:05 but the shocks from trade,
46:06 the flows are too small
46:08 to really make much of a difference.
46:10 So
46:11 trade's not causing what we see,
46:12 it must be something else,
46:13 right?
46:14 Well,
46:14 then China came in with the China shock and
46:17 it's big.
46:18 OK,
46:18 you can no longer say,
46:19 well,
46:19 it's a second order consideration,
46:20 it's first order.
46:22 Um,
46:22 and so the first,
46:23 well,
46:24 one of the first things came in,
46:25 you know,
46:25 Donald came in,
46:26 it's all automation.
46:27 Really,
46:28 the flows are still too small.
46:29 Uh,
46:29 we can only attribute 5,
46:31 560,000 of these job losses
46:34 to,
46:34 um,
46:35 to,
46:35 uh,
46:36 to China,
46:37 um,
46:37 and the rest,
46:38 the other 90% is automation.
46:41 That's a worry for a different reason,
46:43 um,
46:44 but,
46:44 uh,
46:44 but it certainly,
46:45 uh,
46:46 put one marker down,
46:47 and then Otor Dorn and Hansen came in in 2017 and said
46:49 actually it's more like 2 million.
46:51 So we're talking 40,
46:52 50% of the job losses are due to China.
46:55 Um,
46:55 that's a different animal,
46:57 and it's very different.
46:58 The difference between 10% and 50%,
47:00 um,
47:01 is,
47:01 is huge from a policy point of view.
47:03 Then Feinster came in and said,
47:04 well,
47:05 yeah,
47:05 we did lose that many,
47:06 uh,
47:07 our estimation is the same.
47:08 But in the end,
47:08 look at all the jobs that were created.
47:10 We've created 6 million new jobs,
47:12 uh,
47:12 2 million of those in manufacturing.
47:14 So net trade with China has been awesome.
47:17 OK,
47:17 so at the end of the day,
47:19 these are all big people,
47:20 these are all MBR grandees,
47:21 right?
47:21 So it's like,
47:23 what,
47:23 what are we supposed to think at the end of the day?
47:25 And,
47:25 uh,
47:26 and
47:26 what we're supposed to think at the end of the day is,
47:29 we really,
47:29 if we want to keep track of all these things,
47:31 we really need to be much more granular on what we're doing.
47:33 We need to write down models that really
47:35 isolate the different effects that are important.
47:38 And uh this again is where Erhan is going.
47:41 And just to emphasize that,
47:42 re-emphasize that last point,
47:44 um,
47:45 we're no longer talking about trade,
47:46 we're talking about globalization.
47:47 This means we get offshoring,
47:49 trading tasks,
47:50 fragmentation.
47:51 The whole thing becomes much muddier,
47:53 and hence the need to have models
47:54 which can simplify so we can see what's going on,
47:57 um,
47:57 and,
47:57 and credibly abstract from some of these other things
48:00 are very important.
48:01 Also,
48:01 it's nice if we can disaggregate.
48:04 Labor a bit in households so that we're not just saying labor does this,
48:09 labor does that.
48:10 Um,
48:10 so this is the context that I see this literature in
48:13 and why I thought,
48:15 think what Erhan does,
48:16 uh,
48:16 is,
48:17 is
48:17 really important,
48:18 um,
48:19 generally and for what we do at the bank.
48:21 And in particular,
48:22 he's getting down and dirty with the microempirics,
48:24 local labor markets.
48:25 OK,
48:25 you can move a lot within your labor market,
48:27 very hard to move among labor markets in the short term.
48:30 Um,
48:31 uh,
48:31 income distribution and household labor differentiation.
48:34 So we're gonna,
48:35 we're going to break apart households and see how,
48:37 um,
48:38 reducing tariffs.
48:40 On race affects rich households and poor households,
48:43 and that's uh becomes very important.
48:45 Combining empirics with modeling to isolate these
48:47 effects that I talked about before,
48:49 um,
48:49 estimating the actual costs of moving and
48:51 the implications of different degrees of mobility.
48:53 He's exactly buying us this intermediate space
48:56 that Stopper Samuel.
48:57 And the,
48:57 and the uh
48:58 and uh
49:00 uh
49:00 factor specificity models weren't uh uh specific factor models
49:04 were not allowing us to get at.
49:06 Um,
49:06 and then he has the three misconceptions,
49:08 which actually,
49:08 I'm not sure are the most important things of his literature,
49:11 of what he's doing,
49:11 but nonetheless,
49:12 they're important questions.
49:14 Um.
49:15 Select findings in the static context,
49:17 we had this very nice finding from Ghana,
49:19 for instance,
49:19 that we can actually isolate the
49:22 the consumption effects
49:24 from the um
49:25 from the production effects
49:27 and this and who benefits and we find in the
49:30 Ghana case that the poor actually don't benefit from.
49:32 Um,
49:33 from reducing rice tariffs because they produce rice,
49:36 so
49:36 net they,
49:37 they suffer,
49:37 and that means,
49:38 uh,
49:39 Chico and all the friends in our poverty group in EFI
49:43 need to pay attention when we're talking about trade liberalization and like
49:46 to focus on them.
49:47 Um,
49:48 exports,
49:49 they make a really solid effort to try to see who benefits
49:52 from new export industries,
49:53 and that's really key,
49:54 and we find like really positive export effects
49:57 on informality in wages relatively at first,
49:59 um,
50:00 across the board,
50:01 except for rural people,
50:02 um,
50:03 in India,
50:04 um,
50:04 but it's very heterogeneous,
50:05 and this doesn't surprise me at all.
50:07 If you talk to Mexican,
50:08 uh,
50:09 people in Guadalajara about the impact,
50:10 for instance,
50:11 of the maquiladoras,
50:12 uh,
50:13 it's all.
50:13 Women
50:14 Employers want women.
50:15 Um,
50:15 they are not drunk,
50:16 they show up on time,
50:18 uh,
50:18 they have a good attitude.
50:19 They're not men.
50:21 Um,
50:21 so,
50:22 um,
50:22 so,
50:23 um,
50:23 so it,
50:24 it's very very funny because I had to give a
50:25 talk to a group of slightly left of center sociologists,
50:29 um,
50:30 and,
50:30 uh,
50:31 they were all talking about how bad the maquilas were,
50:33 and they couldn't get good help anymore either.
50:35 Um,
50:36 so it's like,
50:36 uh,
50:37 uh,
50:38 they,
50:39 they.
50:40 They didn't catch the irony of that.
50:41 Um,
50:42 so,
50:43 anyway,
50:43 uh,
50:44 so women are benefiting disproportionately,
50:45 we should probably tax them and
50:47 compensate the men,
50:48 um,
50:49 um.
50:51 This compensation issue is a really tough one,
50:54 and he,
50:55 he puts his finger
50:56 on the key question which is,
51:00 Let's assume that Daron is right and that 90% of these
51:03 job dislocation and job loss is actually due to automation.
51:06 No one's running around talking about,
51:08 oh,
51:08 we need to compensate all the people who
51:09 lost their jobs because of robots being introduced,
51:12 but they do in trade.
51:13 And what he highlights is this is a political economy question.
51:16 It's not actually a social justice question per se,
51:18 right?
51:19 Um,
51:19 and so then the question is,
51:21 if it's really a political economy question.
51:23 Are we,
51:23 is it sufficient to say how much do we need to
51:26 transfer to these characters so their utility remains the same?
51:29 Really,
51:29 we need to say how much are they gonna,
51:31 how much do we have to pay them so that they vote right in the next election.
51:34 Um,
51:35 and that could be much more,
51:36 given loss aversion and the fact that people don't really want
51:39 to move from where they are and all their families there.
51:41 Um,
51:41 but,
51:42 uh,
51:43 and,
51:43 uh,
51:43 so
51:44 I think it's actually maybe a little bit,
51:46 uh,
51:46 more complex than we're,
51:48 than even,
51:49 even this exercise is mapping out.
51:51 But what I wanted to push a little harder was
51:54 Uh,
51:54 what's making these costs so high,
51:56 and I think you said 68% of one year's
51:59 salary in your Brazil paper
52:01 to move
52:01 or something like that.
52:03 Uh,
52:03 yeah,
52:04 60,
52:04 something like 68%.
52:05 I mean,
52:05 that's a big chunk of change.
52:07 But in the United States context,
52:08 one of the,
52:09 one of the features of the last 20 years has been a slowing down of mobility.
52:12 So this old grapes of wrath thing,
52:14 well,
52:14 OK,
52:14 this is my,
52:15 my,
52:15 my farm collapsed,
52:16 so I'm going to move to California.
52:18 That's just happening much,
52:19 much less.
52:20 And of course,
52:21 the degree.
52:22 To which workers are resistant to moving
52:23 means that how much we're going to have to compensate them to be happy,
52:26 um,
52:27 is going to be much higher.
52:28 Um,
52:28 so I think we could think a lot more about it.
52:30 I know Gordon has theories that it has to,
52:32 Gordon Hansen has theories that it has to do with the breakup of the American family
52:35 before you pile all the Joads in your van and you go to California.
52:39 Now you've got 3 or 4 families from
52:41 various different wives or girlfriends or whatever,
52:43 it becomes more and more complicated.
52:45 Um,
52:45 um,
52:46 um,
52:47 the other thing is,
52:48 uh,
52:49 Uh,
52:49 disability insurance,
52:50 right?
52:51 We've,
52:51 you know,
52:51 the big welfare reform
52:54 under Clinton has meant a lot of people instead
52:56 of being on welfare are now on disability,
52:58 which means you can't declare that you found
53:00 a job now because you'll lose disability.
53:01 Either you're disabled or you're not.
53:03 Um,
53:03 so in a sense,
53:04 that's a,
53:05 that is a subsidy
53:06 to being immobile,
53:08 um,
53:08 which is also probably,
53:09 um,
53:10 causing troubles.
53:11 Looking into how we can make people more mobile again and make this adjustment more
53:16 feasible.
53:17 Uh,
53:17 strikes me as,
53:18 uh,
53:19 part of his agenda,
53:20 uh,
53:20 that we want to push him on further.
53:22 Just a couple of things,
53:24 because I've,
53:24 I've set up this thing as Darren's saying,
53:26 it's all technology and,
53:28 um,
53:28 and,
53:29 uh,
53:30 outsource becoming more
53:31 favorable to saying a lot of it has to do with trade.
53:33 Um,
53:34 this technology angle is not.
53:37 Disjunct from the trade question and I,
53:39 and somehow we have to integrate these issues more,
53:42 and I just wanted to show this
53:43 quick,
53:44 uh,
53:45 quick graph which I don't wanna show that yet,
53:47 um,
53:47 which shows,
53:48 Uh,
53:49 what happens to domestic employment as you hire more robots?
53:52 And the orange line is the advanced countries,
53:54 and the more robots you have,
53:55 the less manufacturing employees you have as a share of your workforce.
53:59 Except the blue line is for developing countries and it's upward sloping.
54:03 And so there's not a displacement effect there,
54:05 and then the question is why should that be?
54:07 And looking at those countries,
54:09 immediately what pops to mind is this is.
54:11 Probably an offshoring story more than anything is that Hungary,
54:14 when,
54:15 when the Germans take their factories and for medical
54:17 equipment or electronic equipment to Hungary or to Slovenia,
54:21 they take robots with them,
54:23 um,
54:23 and then they export from those bases to the rest of the world.
54:26 Um,
54:26 in Mexico,
54:27 the United States,
54:27 when,
54:28 when Ford opens a plant in Hermosillo,
54:30 it takes the best technology with robots and everything to Hermosillo,
54:33 and so the robots are there and They hire manufacturing workers there.
54:36 There's technological transfer in doing that,
54:38 clearly,
54:39 um,
54:39 but it's part of a trade story,
54:41 um,
54:41 and,
54:41 uh,
54:42 it's becomes a little bit hard to separate out those,
54:44 those issues over time and somehow over,
54:46 you know,
54:46 the this displacement effect versus the,
54:49 the offshoring effect through trade
54:50 are competing in our countries and we probably need to understand them better.
54:54 One more
54:54 little thing,
54:55 and then I will stop taking up oxygen.
54:58 Um,
54:58 is,
54:59 uh,
55:01 that distribution is often related to,
55:04 uh,
55:05 risk.
55:05 I mean,
55:06 trade,
55:07 uh,
55:07 uh,
55:08 increase,
55:08 uh.
55:10 Um,
55:11 increased risk
55:12 is welfare reducing in itself,
55:15 um,
55:15 but it also shows up as worsening distribution,
55:17 which is the
55:18 notional,
55:19 uh,
55:19 topic of this talk.
55:21 Um,
55:22 uh,
55:22 Roderick made a big point.
55:24 20 years ago about saying that
55:26 globalization was really increasing the,
55:28 the risk that workers face and that should show up technically
55:31 in uh
55:32 in distribution and directly through welfare.
55:35 Empirical results to date are,
55:36 are,
55:37 are not super supportive
55:38 of that,
55:39 but nonetheless,
55:40 I think this issue of,
55:41 I mean workers feel more exposed,
55:43 they feel more fear of losing their jobs,
55:45 they feel they feel.
55:46 That they're going to have to uproot their family,
55:47 that life is going to have to change a lot more
55:49 um because of trade,
55:51 and that's a risk story.
55:52 Um,
55:52 so,
55:53 uh,
55:53 we want to keep our eyes on that as the bank as well.
55:56 Anyway,
55:57 I think this agenda is incredibly important.
55:59 I'm so glad he's doing it.
56:00 It's it's important for
56:01 um
56:03 for the profession and it's very important for us at the bank.
56:06 Um,
56:06 so thank you very much.
56:15 Thank you very much,
56:16 Bill,
56:16 uh,
56:17 for that,
56:17 um,
56:18 discussion.
56:19 So we've
56:20 cleared up a whole lot of facts and misconceptions,
56:22 including that
56:24 women are not men.
56:25 I think we have it the right way around.
56:26 Um,
56:27 it's,
56:28 yes,
56:29 and,
56:29 uh,
56:29 but if there are remaining
56:31 questions about facts or misconceptions,
56:33 it's a good time to open up for,
56:35 uh,
56:35 Q&A.
56:36 Asla,
56:36 you have a burning question.
56:38 It's not a burning question,
56:39 but,
56:40 uh,
56:40 I'm,
56:41 I would like to hear a burning reaction.
56:44 So listening to you,
56:46 I am struck by the similarities between trade and immigration.
56:52 Um,
56:53 it's also brought to my,
56:55 uh,
56:56 memory some of the reactions that we've heard from the Nobel Prize winners,
57:00 right,
57:01 Esther Defoe and Abhijit Benaji,
57:03 where they were taking a very,
57:06 um,
57:06 negative
57:08 perspective on the trade implications,
57:11 whereas a very positive
57:14 perspective on the impact of migration.
57:18 Or at least not negative,
57:19 and I,
57:20 uh,
57:20 listening to you,
57:21 I think one of the key elements distinguishing between these things
57:26 is mobility,
57:27 the fact that they're probably underestimating the role of mobility,
57:31 but I wanted to hear your view on What you
57:34 think about these different takes on the migration and trade,
57:38 where,
57:39 you know,
57:39 analysis would suggest very similar implications,
57:43 where,
57:43 you know,
57:44 the
57:45 uh prominent people in the field seem to
57:47 be seeing it from a very different perspective.
57:51 Let's take a few more questions um
57:54 uh for,
57:54 for Erhan.
57:57 Chico.
57:59 Thanks,
57:59 Erhan,
58:00 for a great talk.
58:01 Um,
58:02 I,
58:02 I,
58:02 I have a
58:03 question about,
58:04 uh,
58:05 dynamic gains from trade,
58:08 um,
58:08 in the,
58:09 in the sense that I used to hear about,
58:11 but I,
58:11 I,
58:11 I think the discipline doesn't
58:13 Talk so much about
58:15 anymore,
58:15 which is this possibility
58:17 that actually you absorb,
58:18 you know,
58:18 you absorb new technology and you become,
58:21 you know,
58:21 the production function changes.
58:23 The production function,
58:24 you,
58:25 the efficiency of the whole economy changes.
58:27 The sort of stuff that um
58:29 uh
58:30 Haman,
58:31 Roderick,
58:31 and Velasco used to hypothesize about if there
58:33 are monkeys jumping from tree to tree,
58:35 or,
58:36 um,
58:36 you know,
58:36 Justin Lin used to talk about flying geese.
58:38 The idea that China is doing
58:41 Huawei now
58:42 has to do with trade,
58:44 and has to do with the fact that they
58:46 were trading,
58:46 you know,
58:47 textiles first,
58:48 and then other things,
58:48 and so on and so forth.
58:49 And it seems to me that everything you talked about,
58:52 uh,
58:52 when you talked about
58:53 um
58:54 The aggregate gains that are difficult to identify when you don't have mobility.
58:58 In my head,
58:59 they were still the static gains,
59:00 the sort of gains from efficiency from reallocation,
59:03 rather than these potentially,
59:06 you know,
59:06 more far-fetched,
59:07 um,
59:08 productivity gains.
59:09 Is that something the literature looks at?
59:11 Is it just even more difficult to identify?
59:13 Is there any evidence of that kind of thing?
59:15 And if there is,
59:16 you know,
59:16 does it make
59:17 the bar
59:19 um
59:20 against trade even higher in the sense that
59:22 there's additional benefits that we're not looking at?
59:27 Interesting.
59:29 Um,
59:29 thanks,
59:29 Erhan,
59:30 great talk.
59:30 Um,
59:31 my question is about compensation,
59:33 so if
59:34 I agree that
59:35 we'd like to compensate people who are losers,
59:38 um,
59:38 what do you see as the best way of,
59:39 of going about identifying them?
59:42 Um,
59:42 and,
59:42 and then specifically how do you do that
59:44 if,
59:45 um,
59:45 you're trying to anticipate losses that will be in the future.
59:49 Um,
59:49 so,
59:50 you know,
59:50 a,
59:50 a transition as,
59:51 as a result of China's growth,
59:52 we think.
59:53 Industries will continue to decline in the United States.
59:56 Um,
59:57 is there any good way of identifying them now?
1:00:03 Let me add one other question and then we'll,
1:00:05 uh,
1:00:06 uh,
1:00:06 turn over Erhan,
1:00:07 to you to answer a batch of them.
1:00:09 Um,
1:00:10 and I was wondering when you talked a lot
1:00:12 in the beginning about the fact that you're not
1:00:14 able to identify sort of the winners from trade,
1:00:17 um,
1:00:18 and I was wondering whether you were
1:00:20 inadvertently or unintentionally using the word identify in the
1:00:23 econometric sense rather than in the plain English sense.
1:00:26 And I was thinking that it might be the former because
1:00:28 you're emphasizing the fact that the gains are uncorrelated with income,
1:00:31 which makes them difficult to identify in an econometric sense,
1:00:34 but that doesn't mean that
1:00:35 they're necessarily hard to
1:00:37 Identify in the plain English sense,
1:00:38 right?
1:00:38 And I thought that the modeling work from the first paper,
1:00:41 it's very clear,
1:00:42 if you have some idea of what consumption shares are,
1:00:44 and either you assume consumption shares don't change
1:00:46 a huge amount,
1:00:47 or if they do,
1:00:48 we have some idea of what demand systems look like,
1:00:50 then it doesn't seem very hard to get those sort of
1:00:53 big welfare gains that are coming from the consumption side,
1:00:56 even though they're not correlated with income and therefore
1:00:58 not sort of identifiable in an econometric sense.
1:01:01 Any,
1:01:01 you've got plenty to
1:01:02 go on,
1:01:03 so.
1:01:09 OK,
1:01:09 so I think the questions are
1:01:13 starts with a very difficult question and gets easier,
1:01:15 so I will
1:01:16 answer them in the reverse order,
1:01:19 putting the less difficult question to the last,
1:01:22 so
1:01:23 er.
1:01:24 So yes,
1:01:26 er,
1:01:26 our question is,
1:01:27 is,
1:01:28 is,
1:01:28 uh,
1:01:29 his explanation makes a lot of sense,
1:01:31 so it's like,
1:01:32 it's exactly what I,
1:01:33 what I meant
1:01:34 by identification,
1:01:35 I meant econometric identification.
1:01:38 So if you have data and some framework
1:01:41 on the,
1:01:43 uh,
1:01:44 income shares,
1:01:45 consumption shares,
1:01:46 it is possible to identify
1:01:48 winners and losers from the English sense,
1:01:51 but it is difficult to identify them.
1:01:53 From the econometric sense,
1:01:54 so we cannot really use statistical tools,
1:01:56 but it is possible to identify them.
1:01:59 But the problem with that is we have to rely on
1:02:04 the model and make very strong assumptions like the one
1:02:06 that I did in the very beginning there is no.
1:02:10 The
1:02:12 reallocation of resources.
1:02:13 So if we have
1:02:14 a strong model,
1:02:15 like a model with strong assumptions,
1:02:18 then it is possible to identify winners and losers,
1:02:21 but then
1:02:22 we can,
1:02:23 Naturally give evidence,
1:02:24 so in order to give evidence we have to have both,
1:02:27 uh,
1:02:28 econometrics,
1:02:29 statistics and
1:02:31 some
1:02:31 quantitative model and it is,
1:02:33 I think it's very difficult to have them both at the same time,
1:02:35 so
1:02:36 identification from econometric sense and English sense is,
1:02:39 is not.
1:02:40 Is not
1:02:41 so easy.
1:02:43 So Tristan's er question,
1:02:45 is it possible
1:02:47 er to,
1:02:48 so how do we identify
1:02:51 the winners and losers,
1:02:53 So what are the things that we have to
1:02:55 think about?
1:02:56 So there is a big debate.
1:02:58 So if you just read the local labor markets approach,
1:03:01 then you will think that
1:03:03 Only the region matters,
1:03:05 so other than region,
1:03:06 other things don't matter as much,
1:03:08 but if you look at
1:03:09 other
1:03:10 pieces of literature,
1:03:12 it would be industries or it could be firms,
1:03:14 it could be occupations,
1:03:15 so depending on where you look at,
1:03:17 you'll find different
1:03:18 answers,
1:03:19 so.
1:03:20 I think the only possible way to identify the
1:03:23 precisely who is losing and who is winning,
1:03:25 we have to
1:03:26 like
1:03:27 kind of combine all these approaches and try to
1:03:30 find the correlation of winning and losing respect to
1:03:33 the attributes.
1:03:34 So
1:03:34 it's not easy and we have to like have a
1:03:38 a broader approach.
1:03:40 Uh,
1:03:40 so
1:03:41 Chico's question is,
1:03:43 uh,
1:03:44 do we actually
1:03:46 Have good evidence about dynamic effects.
1:03:50 So
1:03:51 it is unfortunately a challenge.
1:03:52 So there were recently some papers about identifying
1:03:55 procompetitive games and they did not find much.
1:03:58 And there are some papers about
1:04:00 the
1:04:01 uh closing of
1:04:03 the different channels and then
1:04:04 how it affects growth.
1:04:06 So
1:04:07 basically to identify the
1:04:09 dynamic effects,
1:04:10 you need
1:04:10 time series
1:04:12 because it is dynamic.
1:04:13 And then time series
1:04:15 at the same time usually as
1:04:17 you mentioned,
1:04:18 there are lots of shocks happening at the same time like technology
1:04:22 and,
1:04:23 Uh,
1:04:24 automation,
1:04:24 other things at the same time,
1:04:25 so it is much more difficult
1:04:28 to
1:04:29 have a time series variation and to identify
1:04:32 the dynamic effects.
1:04:34 Uh,
1:04:34 so
1:04:36 Yes,
1:04:36 I think you would need some model
1:04:38 in order to identify econometrically it's difficult,
1:04:41 but then
1:04:42 even with models,
1:04:43 it is,
1:04:43 it is not so easy.
1:04:46 Uh,
1:04:47 so
1:04:48 Arcelor's question is,
1:04:48 I think the most difficult one,
1:04:50 the,
1:04:51 uh,
1:04:51 the impact of,
1:04:53 Trade and
1:04:54 immigration,
1:04:55 so.
1:04:58 The
1:04:59 The impact of
1:05:01 trade and immigration are
1:05:02 definitely they they look similar,
1:05:04 they are both related to mobility.
1:05:06 One of them is about mobility of workers,
1:05:08 the other is the mobility of goods,
1:05:10 but,
1:05:12 I think there was
1:05:13 something that Adam Smith say,
1:05:15 like,
1:05:16 Hundreds of years ago he was saying that you cannot have
1:05:19 actual.
1:05:21 Uh,
1:05:22 free trade without having,
1:05:24 uh,
1:05:25 free labor mobility,
1:05:26 so it's it's,
1:05:27 it's a it's a factor,
1:05:28 it's a production factor,
1:05:28 it's a part of the production process,
1:05:30 so you have to have
1:05:32 both
1:05:33 free trade of
1:05:34 goods and services and also the workers,
1:05:37 so,
1:05:39 But I think the reason why we have
1:05:42 different effects is that
1:05:44 In the
1:05:46 migration literature,
1:05:47 we usually think about
1:05:49 the
1:05:50 locals,
1:05:51 so
1:05:52 how,
1:05:53 how the
1:05:54 host countries are affected,
1:05:55 and the migration policies are determined by them,
1:05:58 not the ones who are taking
1:06:00 benefit,
1:06:01 essentially migrants.
1:06:03 And
1:06:04 also the migration is much more
1:06:07 difficult because the frictions are much higher,
1:06:09 so that's why you have
1:06:12 much larger frictions
1:06:14 and.
1:06:15 I think that's the,
1:06:16 that's the main difference
1:06:18 uh between the,
1:06:20 the,
1:06:20 the migration literature and the trade literature,
1:06:22 but
1:06:23 as I said,
1:06:24 many of the things I mentioned could be
1:06:26 applied to
1:06:27 migration literature as well.
1:06:30 Uh,
1:06:30 but
1:06:31 the political economic questions are much
1:06:33 more difficult in the migration literature,
1:06:35 so it's not,
1:06:36 I think the,
1:06:37 the policies are more sophisticated and more difficult to answer.
1:06:44 Thanks,
1:06:44 Sarahan.
1:06:44 Um,
1:06:45 we have time for another,
1:06:46 uh,
1:06:47 round.
1:06:48 Ivan,
1:06:49 um,
1:06:50 and I'm sorry,
1:06:51 I,
1:06:52 yes.
1:06:54 Hi.
1:06:55 Uh,
1:06:55 I have
1:06:56 two related questions on,
1:06:58 on compensation.
1:07:00 So,
1:07:00 um,
1:07:01 from what you presented,
1:07:02 uh,
1:07:03 on compensation,
1:07:04 it is seen
1:07:06 from the monetary point of view that,
1:07:08 no,
1:07:08 this is the amount of money we need to compensate for the losses.
1:07:12 I'm wondering if you are aware of any work that has been looking at whether
1:07:17 Actually,
1:07:17 the losers may
1:07:19 prefer,
1:07:20 you know,
1:07:20 to get a job rather than a check with the equivalent amount of money.
1:07:26 And I mean,
1:07:26 this is just a question,
1:07:27 are you aware of some kind of welfare
1:07:29 analysis that takes into account this difference?
1:07:31 And related to that,
1:07:32 um,
1:07:34 My question is,
1:07:35 can some of these neo-protectionist policies we are seeing these days.
1:07:39 Being seen as an inefficient way of doing this in the sense of,
1:07:43 OK,
1:07:44 we,
1:07:44 we are going to
1:07:45 raise tariffs to generate some
1:07:47 inefficient employment,
1:07:49 but that is going to,
1:07:50 uh,
1:07:51 you know,
1:07:51 keep these people employed
1:07:54 rather than taking them out of the labor market and just
1:07:57 giving them a check.
1:07:58 That's
1:07:59 two questions.
1:08:00 Thank you.
1:08:03 OK,
1:08:03 so my question is almost exactly the same.
1:08:05 I think it's because we are working on a paper which is about this actually
1:08:09 together.
1:08:09 So yeah,
1:08:10 exactly.
1:08:10 So going back to compensation and the non-monetary aspects like uh
1:08:14 because losing a job has effects on health,
1:08:16 on family composition,
1:08:18 on.
1:08:18 Violence and drug behavior and all those things and,
1:08:21 and yes,
1:08:21 similarly,
1:08:22 I was wondering if
1:08:23 we can pull out from the econometric or empirical literature to have like a
1:08:27 much better idea actually of like the total impacts of like these big losses of
1:08:32 of jobs.
1:08:38 All right,
1:08:39 Arahan,
1:08:39 you get the last word.
1:08:41 OK,
1:08:41 so,
1:08:42 uh,
1:08:42 thank you very much for the question.
1:08:44 So I'm going to combine Matilda and Ivan's questions.
1:08:48 So,
1:08:48 uh,
1:08:50 Why should we actually write a check to the workers?
1:08:53 Can't we just give some
1:08:55 non-monetary
1:08:56 compensation like jobs or training?
1:08:58 So
1:08:59 the famous program was trade
1:09:01 adjustment assistance in the US
1:09:04 and there are lots of empirical research about this and
1:09:07 usually what they find is that it did not
1:09:09 work well.
1:09:11 So
1:09:14 Trying to give
1:09:15 workers some.
1:09:18 Like training or finding jobs,
1:09:20 it's not so easy because.
1:09:23 It requires identifying precisely
1:09:27 who is losing,
1:09:28 which is not.
1:09:29 Easy as the local labor market literature showed,
1:09:32 like if you're a service worker in a,
1:09:34 in a region
1:09:35 where you're,
1:09:37 uh,
1:09:38 where it is
1:09:39 highly concentrated import competition,
1:09:41 you can be worse off.
1:09:43 So because your job will be threatened by the ones who are uh
1:09:47 unemployed in the other industries,
1:09:49 so it is not really easy to identify.
1:09:52 So the challenges that we face as researchers
1:09:56 are the same challenges by the policy makers,
1:09:58 so they cannot really pinpoint.
1:10:00 So the policy that actually work really well is the flexicurity in Denmark,
1:10:04 but as I said it is
1:10:06 uh
1:10:07 not
1:10:08 probably feasible
1:10:09 in many cases.
1:10:10 And one thing about the compensation,
1:10:12 this uh this another idea,
1:10:14 uh rather than
1:10:16 writing checks,
1:10:17 can you just give conditional compensation like,
1:10:21 Uh,
1:10:21 some aid
1:10:23 to let workers move,
1:10:25 but it is not clear
1:10:27 whether these labor mobility frictions
1:10:30 are due to some market failure,
1:10:33 or
1:10:33 are they like really distortions,
1:10:35 are
1:10:36 they are just frictions.
1:10:37 Maybe it's the case that
1:10:38 people just don't want to move.
1:10:40 So there are some studies that show that
1:10:42 if actually you try to compensate people to move,
1:10:45 there is a welfare loss.
1:10:48 So
1:10:50 Since identification is difficult,
1:10:52 compensation is difficult,
1:10:54 not
1:10:54 only because it's costly,
1:10:56 but because we don't know whom to
1:10:59 er compensate.
1:11:03 All right.
1:11:03 Well,
1:11:03 thank you very much everyone for coming.
1:11:08 And uh before you all run out,
1:11:10 just one word of advertisement that on Friday in the same time,
- add-style
- lp-body-content