00:00 Welcome,
00:00 everyone.
00:01 Thank you for joining us in the
00:03 new round of the
00:05 2024 WDR seminar series.
00:09 As many of you know,
00:10 the forthcoming WDR
00:12 focuses on economic growth in middle income countries,
00:16 particularly
00:17 trying to uncover the policies or the forces that make
00:20 transitioning from middle to high levels of income so difficult to achieve.
00:25 So in today's presentation,
00:27 uh,
00:27 we wanted to draw lessons from.
00:29 A country that is somewhat of an exception
00:32 to this empirical regularity which is South Korea.
00:36 So as we know,
00:37 Korea managed to transition from low to
00:39 high income levels almost without interruptions,
00:42 sustaining decades of,
00:44 uh,
00:44 economic growth.
00:46 So
00:47 questions that we're interested in for the purpose of WDR is how did Korea do it?
00:51 What.
00:52 What did they do,
00:53 what policies they followed,
00:54 and
00:55 what were the channels through which growth ensued
00:58 and didn't stagnate at uh some middle-income stages as in many other countries.
01:04 So,
01:04 uh,
01:05 following up on last week's presentation
01:07 in which the focus was more on the industrial policy aspect of Korea's growth,
01:12 today we have the pleasure of having Moon Su Lee
01:15 talk about,
01:16 uh,
01:17 offering a plant-level view of the drivers of Korea's growth miracle
01:22 and the more recent,
01:23 uh,
01:24 economic slowdown.
01:25 So let me tell you very briefly about the speaker.
01:27 So Moon is an assistant professor of economics
01:30 at the University of California in San Diego,
01:33 and
01:34 together with his co-author today,
01:36 John Shin,
01:37 who is a professor of economics at Washington University in Saint Louis,
01:41 have written extensively on the issues of
01:44 macroeconomic transitions in general,
01:46 but with a particular focus on,
01:49 on Korea's experience.
01:51 So we are very glad to have Moon today to,
01:53 to,
01:54 to present his research.
01:55 Uh,
01:56 the seminar rules,
01:57 I was thinking of allowing Moon speak.
02:00 With minimal interruption for 35 to 40 minutes unless of course
02:04 some pressing clarifying questions arise,
02:06 although uh Moon just tells me that please jump in
02:09 with any question you may have,
02:11 but still I think that uh with this time allocation,
02:14 we'll have plenty of time for Q&A
02:17 uh towards the end and
02:19 uh we can have some interesting discussions based on what we learned from,
02:23 from the talk.
02:24 So with that,
02:25 let me turn the floor to you,
02:26 Moon.
02:27 Uh,
02:27 thank you so much.
02:28 Take it away,
02:28 please.
02:30 OK.
02:31 So thank you,
02:32 Roberto,
02:32 the Jewish,
02:33 and the World Bank colleagues.
02:35 First of all,
02:35 for inviting me to this amazing team.
02:38 I'm preparing for the
02:39 WDR the next year.
02:42 So it has been a pleasure to work on this,
02:45 the issue,
02:45 and I'm glad to,
02:47 the,
02:48 the,
02:48 that you gave me the opportunity to share our findings.
02:52 So as Roberto say that this is the joint work with
02:55 the Jung Xin at the Washington University in Saint Louis.
03:02 Again,
03:02 I mean,
03:03 in front of this audience,
03:05 it may be redundant to say this again,
03:08 but South Korea is one of the rare economies that went from poor to rich
03:14 in one generation,
03:15 I mean,
03:15 literally within the 30 years.
03:19 So that's why it wasn't surprising that we haven't,
03:22 we have seen the several other,
03:25 the works in this event cities,
03:27 the revisiting
03:28 the experience of South Korea.
03:32 But then in this,
03:33 the study,
03:34 we want to take some balanced view,
03:37 the looking at both the gross miracle period,
03:40 say from the 1960s until the 1990s,
03:44 but at the same time the remarkable growth,
03:46 the slowdown
03:48 that has gotten the relatively less attention.
03:51 Because Korea's economic growth rate nowadays
03:55 is the 2 to 3%,
03:56 which is not much different from
03:58 the US.
04:00 So we will see
04:01 the one generation of the growth miracle,
04:03 but then the next generation of the slowdown.
04:09 But that's the macroeconomic phenomenon.
04:11 So the research question
04:13 that we have is
04:15 whether there is any systematic pattern at the micro level.
04:20 Which is at the plant level,
04:22 the behind the macro level grows miracle and slow down.
04:27 Let me first summarize our findings.
04:29 So we have two major findings.
04:31 The first,
04:33 we find no
04:34 clear relationship between the macro-level grows
04:37 and plant size distribution
04:40 or static measure of allocated efficiency.
04:43 So,
04:44 later,
04:44 I will tell you in detail why this happened in Korea.
04:48 But we also report
04:50 some evidence that
04:51 the,
04:52 the,
04:52 looking at the time series evidence and the cross-sectional evidence,
04:57 there might not be the,
04:59 the,
04:59 the very clear relationship between the cruise and
05:02 then plant level size distribution and allocated efficiency.
05:07 So I will get there
05:08 later.
05:09 And the second finding is that
05:11 I told you that
05:13 the,
05:13 the Korea experienced the growth slowdown for the recent 20 years,
05:17 and that coincides with the reduction in the business dynamism.
05:22 So the job creation and job destruction
05:25 or responsibilities to the productivity shocks,
05:28 so the business dynamics in general,
05:31 the declined over the slowdown period.
05:38 So let me first assure you the GDP per capita.
05:42 So the,
05:43 the orange line is the GDP per capita,
05:46 the covering
05:47 the 50 years,
05:49 and then the blue line is the value added per worker in manufacturing sector,
05:54 because in our study,
05:55 we will look at the manufacturing sector
05:57 where we have a high-quality data at the plant level,
06:01 but at the same time,
06:02 the manufacturing sector was one of the main drivers behind
06:06 the
06:06 Korea's economic growth
06:08 and slowdown too.
06:11 Korea experienced a gross miracle during the 1960s,
06:15 1970s,
06:15 198980s,
06:16 and 1919990s.
06:17 And then there was the Asian financial crisis
06:20 in 1997,
06:22 and after that,
06:23 the Korea experienced a growth slowdown both in the aggregate economy
06:28 and manufacturing sector.
06:35 So we will use the newly dispatched the mining and manufacturing survey,
06:40 the,
06:41 the covering,
06:42 the interesting period is starting from the 1967 until the 2019.
06:47 So there are several advantages and limitations like all
06:51 other microdata the researchers are using these days.
06:56 The first advantage is that this is the unique source of plant-level data,
07:01 the covering all plants with the
07:04 5 and more workers,
07:06 and the 10 and more workers starting from the 2007
07:10 in the manufacturing sector in Korea.
07:13 So like other manufacturing surveys in the US and other countries,
07:18 so we have detailed information on input,
07:20 labor and capital,
07:22 and output,
07:22 the value added,
07:24 and other information.
07:26 And when we aggregate this plant level data,
07:29 it replicates the aggregate statistics well.
07:35 There are several limitations.
07:37 The panel divention is only available after the 1981.
07:42 So before the 1981,
07:44 so we don't use the
07:46 panel structure.
07:47 Still,
07:47 we can show you some cross-section of the evidence.
07:51 And the capital stock,
07:52 one of the most important
07:54 input in the manufacturing sector,
07:57 is available only one year in the 1968,
08:01 and then it is available again every year after the 1978.
08:06 So that's the data limitation.
08:08 So
08:08 in later slide,
08:09 you may see some,
08:10 some jumps
08:12 when I report
08:13 the,
08:14 the set of the variables that requires the capital information.
08:24 Let me first report
08:26 the plant size distribution.
08:28 So this is the average size of the employment
08:33 for the manufacturing plant.
08:34 We dropped all the mining,
08:36 the sector.
08:37 So it started from
08:39 60.
08:40 So we have two graphs,
08:41 but I will look at the graph on the right,
08:44 the covering the old plants hiring more than 5 and more workers.
08:49 So the average employment,
08:50 average plant size was around 25 in the 1967,
08:56 and it increased up to over 70,
09:00 and it declined again towards the around the 15.
09:05 So it shows the invoice to you pattern.
09:07 So there was the increase during the 1960s and 191970s,
09:11 and then it declined
09:13 during the 1980s and 1919990s,
09:16 and then it has been
09:17 stabilized.
09:20 So it is somewhat different from the,
09:22 the other papers,
09:23 the finding that
09:25 the,
09:25 the,
09:25 the plant size,
09:27 the farm size increases over
09:30 the development.
09:32 So we believe this is somewhat related to
09:35 the,
09:35 the specific set of the policies in Korea.
09:38 The,
09:39 the Korea was pushing for the big push
09:41 industrial policy during the 1960s and 191970s,
09:45 and then there was the,
09:46 so-called the rationalization period.
09:49 The,
09:50 the,
09:50 the asking the private sector to read the grows,
09:54 the,
09:55 the by removing the set of the industrial policies.
09:59 But that's what we believe,
10:00 but the one message we want to convey here is that.
10:04 It's not that simple,
10:05 the relationship between the plant sites and the development of the country.
10:13 So another way of looking at the plant size distribution is
10:18 the quantifying the employment share of establishments
10:23 hiring
10:24 the 200 workers or the 500 workers.
10:27 This one represents the degree of the concentration.
10:31 Because the average plant size,
10:33 so it may mask the massive heterogeneity
10:37 on the
10:38 small farm size,
10:39 and the small,
10:40 small and the medium-sized corporations,
10:42 but at the same time,
10:43 large sized corporation.
10:45 But when we measure the degree of the concentration by looking at the share of the,
10:49 the employment hired by the large sized corporation,
10:52 it shows exactly the same
10:54 inverse the U shape.
10:59 The most transparent way of showing the plant size distribution is looking
11:04 at the so-called the log log plot.
11:07 So what is the log log plot?
11:09 The x-axis is the log of the employment,
11:13 and the y axis is the fraction of the establishments hiring more than
11:18 the specific number of the employment.
11:22 So whenever we see
11:24 The low lobe plot close to the flat line,
11:27 that means
11:28 the,
11:28 the farm,
11:29 farm size,
11:29 the,
11:30 the farm size is very large.
11:32 And then if you will see the low globe plot closer to the vertical line,
11:36 so that means the,
11:37 the,
11:38 the farm sizes are in general small.
11:41 There are
11:42 5 lines,
11:43 but I want you to look at only three lines,
11:46 the blue,
11:47 red,
11:48 and orange.
11:49 The blue is the 1967,
11:52 the red is the 1977,
11:55 and orange is the 1997.
11:59 So low low plot was the,
12:00 the shifting toward the right
12:02 and it came back to the original level.
12:05 So meaning that
12:07 the increase and decrease of the plant size were
12:11 broad-based.
12:12 So it was not driven by
12:14 the right tail of the distribution or even the left tail of the distribution.
12:19 So there was the increase and decrease of the plant size that were
12:23 broad-based.
12:27 So let me quickly give you the three related discussions.
12:31 So why it happened in Korea.
12:34 So the first question is
12:36 whether it was driven within industries,
12:39 I mean within the four-digit industries in manufacturing,
12:43 or
12:44 between the,
12:45 the sub-industries within manufacturing.
12:49 And the second question is,
12:50 I already hinted to you,
12:52 but do we find a similar inverse pattern in other countries,
12:56 especially in Asian countries that has experienced the gross miracle.
13:02 And third,
13:02 why do we see the inverse U pattern in Korea?
13:05 So
13:06 there I will,
13:07 the,
13:08 the,
13:08 the talk a little bit about the industrial policy during the 1960s and 19970s.
13:16 So if there's no questions,
13:18 so I will enter to the first point.
13:23 The first point is
13:26 quantifying
13:27 whether the increase and decrease of the farm
13:30 size came from within or between industries.
13:35 MT is the aggregate average employment defined as the
13:39 weighted sum of each industry's weight average employment.
13:44 So we use the four digit,
13:46 the industry,
13:47 then this is the employment share of the industry I,
13:52 and this is the weighted average of the
13:55 average employment in that industry.
13:58 We can decompose
14:00 the change in
14:02 the aggregate average employment
14:05 into 3 components.
14:07 The first component is
14:09 the,
14:09 the,
14:10 the fixing
14:11 employment share of the industry in the previous
14:15 year,
14:16 and then only taking into account the change within the industry.
14:20 So the first one is the weight adjustment.
14:24 And the second term is the fixing.
14:26 The average employment of the industry
14:30 only varying.
14:32 The size of the industry.
14:35 And then the 3rd
14:37 term is the residual.
14:39 So the first term is the within component and the 2nd term is the between component.
14:46 So this is the research.
14:48 So this is the cumulative contribution of the within,
14:51 between and residuals.
14:53 So
14:54 the between adjustment was contributing
14:57 toward the increase in
15:00 the,
15:00 the,
15:01 the average the plant size,
15:03 meaning that
15:05 the resources has been more allocated to
15:08 the industry
15:09 that had the larger the plant size.
15:13 But that inverse
15:15 U pattern was solely driven by the wid adjustment.
15:19 So this is the the inverse U pattern,
15:22 the increase and then decrease,
15:24 and then stabilization was solely driven by
15:28 the within the four-digit industry.
15:30 Again,
15:31 this inverse U-shaped pattern was the broad-based across the industries.
15:39 The next question,
15:40 natural question is whether it was a unique phenomenon in Korea.
15:46 The very well-known influential paper by Bob Lucas in 1978,
15:52 the Span of Control Paper,
15:55 but he showed,
15:55 he reported that the average farm size increased over time in the United States.
16:03 But evidence outside the US and Europe is limited because of the data,
16:09 the limitation,
16:09 you need the plant-level information
16:12 to calculate
16:13 average the plant or the farm size.
16:17 Interestingly,
16:18 there was a special issue of the journal,
16:21 The Small Business Economics in February 2002.
16:25 So the set of the authors,
16:26 the,
16:27 the,
16:27 the collectively,
16:29 the reported
16:30 the,
16:31 the,
16:31 the patterns in the plant and the farm size distribution,
16:35 the,
16:36 the historical,
16:37 the evolution.
16:39 So,
16:40 we checked those papers and then the Taiwan,
16:43 Korea,
16:43 I mean,
16:43 Korea,
16:44 naturally,
16:44 and Japan,
16:45 they show the similar,
16:47 the invoice,
16:47 the U-shaped pattern.
16:49 And then there were the other side of
16:51 the countries that showing the increasing the pattern,
16:54 and there was the one country that's showing
16:56 the stable the pattern.
16:59 So again,
17:00 the evidence on the relationship between
17:03 the plant farm size and economic development is mixed,
17:07 especially
17:08 in
17:08 the,
17:09 the Asian
17:10 countries.
17:15 We believe
17:17 in the case of the Korea,
17:18 I cannot generalize that into the case of,
17:21 uh sorry,
17:22 let me go back to the previous slide,
17:23 the Taiwan and Japan,
17:25 the,
17:26 the,
17:26 who has also experienced this inverse U pattern,
17:29 but our understanding in the Koreans,
17:32 the context is that
17:34 the industrial policy and its removal that contributed to
17:38 the inverse U-shaped,
17:40 the fam size distribution.
17:44 So the Roberto told me that this,
17:46 the,
17:46 the event series already covered the paper last week talking
17:51 about the industrial policy in Korea during the 1970s,
17:55 so I'll be very brief.
17:57 So,
17:57 there was a big push industrial policy targeting the side of
18:01 the industries and targeting some of the specific areas within the
18:06 the,
18:07 the,
18:07 the,
18:07 the,
18:08 the South Korean peninsula.
18:12 So this is the,
18:13 this is effective tax rate.
18:15 So the,
18:16 during the industrial policy period,
18:18 it
18:18 targeted a specific set of the industries represented by
18:22 primary metal,
18:23 the fabricated metal,
18:25 machinery,
18:26 and the equipment,
18:28 and the chemical industry.
18:30 Those were the targeted industries.
18:32 And then the Korea Development Bank,
18:34 the state-owned bank that made the more
18:38 the loan toward the heavy and the chemical industries.
18:41 So this industrial policy targeted
18:43 heavy and chemical industries.
18:47 At the same time,
18:48 the Korean government
18:51 built
18:51 the 9 industrial complexes,
18:54 those are concentrated in the south and eastern
18:58 area of South Korea
19:01 that had the advantage toward
19:04 the access to the largest port,
19:06 which is Busan.
19:10 And this is an example of Changwon.
19:12 Changwon is the name of the city,
19:14 and the Korean government built a machinery industry cluster there.
19:19 So in 1974,
19:21 it was empty,
19:23 the territory,
19:25 but within the two years,
19:27 the factories had been built and then they started the production.
19:35 But this industrial policy was short-living,
19:38 so it unexpectedly ended within the 6 years
19:43 after the assassination of the President Park Chung-hee in October 1979.
19:51 And then there was a period of so-called the rationalization
19:55 by
19:56 the new military junta in 1980
20:00 to distance them,
20:01 to differentiate themselves from the President Park Geun-hye's regime,
20:05 and to blame the economic
20:07 contraction
20:09 in 1979 and 1980.
20:13 Mostly by the second OSHA,
20:16 the,
20:16 to blame the,
20:17 the,
20:17 the economic contribution on the parks,
20:20 the,
20:20 the heavy chemical industry policy and
20:23 the low utilization.
20:25 So by the way,
20:26 this the 1980 was the first negative growth since
20:29 the Korean War that ended in 1953 in the modern Korean
20:35 history.
20:37 So as an outcome,
20:39 the new establishment entered at a faster rate in the 1980s,
20:44 driving down the average plant size while the aggregate economy grew
20:49 steadily.
20:52 So this is somewhat
20:53 unrelated to the plant size distribution I'm talking about,
20:58 but
20:58 I personally believe
21:01 the
21:01 the the Korean industrial policy was successful partly because
21:06 it was the short-lived.
21:09 Why do I believe that?
21:10 So,
21:11 if I look at the macroeconomic indicators during the 1979,
21:17 the government debt was increasing,
21:19 the inflation rate was
21:21 15% per year.
21:23 So,
21:24 and there was the,
21:26 the,
21:26 the,
21:27 the,
21:27 the,
21:27 the increase in,
21:29 uh no,
21:30 no,
21:30 there was the,
21:31 the,
21:31 the decrease in the utilization rate.
21:33 So hinting was that
21:34 there has been some,
21:36 the redundant,
21:37 the investment,
21:39 and then the government.
21:41 Was
21:42 the leading debt investment,
21:44 so the government was the,
21:45 the,
21:46 the,
21:46 the accumulating the debt and increasing the money supply.
21:50 So during the 1978,
21:52 1979,
21:53 we see some
21:55 warning indicators,
21:56 the warning,
21:57 the macroeconomic indicators.
21:59 But industrial policy ended and the six-year period was enough
22:04 for
22:04 the entrepreneurs to make the necessary
22:07 investment in these heavy and the chemical
22:11 industries.
22:12 And then they didn't need the subsidy
22:15 anymore.
22:16 So during the 1980s still,
22:18 they were able to
22:20 produce the output and then
22:23 becoming the competitive in the global market.
22:27 Sorry,
22:27 let me come back to
22:29 the,
22:30 the,
22:30 my original presentation,
22:32 uh,
22:33 and move to the other findings we have from the plant level data.
22:38 So so far we looked at the size distribution.
22:41 The second issue is the static allocated efficiency,
22:45 whether
22:46 the resources,
22:47 the labor and capital
22:49 are allocated efficiently
22:52 across the plants within the narrowly defined industry
22:56 and how that changes over time.
22:59 So we will use the Shankle of 2009 papers,
23:02 the Methodology,
23:04 differentiating the TFPQ and TAFPR.
23:08 So it is very natural to see the dispersion in the TFPQ,
23:12 but under certain assumptions,
23:14 the,
23:15 the we shouldn't see the dispersion in the TFPR.
23:19 So whenever we see the dis dispersion in the TFPR,
23:23 so that is the telltale sign of the,
23:26 the,
23:26 the distortions that each plants are facing.
23:35 So then we calculate the elasticity,
23:37 the correlation between the TFPR and TFPQ.
23:41 The TPPQ is the productivity and TPPR is the size of the idiosyncratic distortion.
23:49 So then what do we mean by the correlation?
23:51 So if the correlation is low,
23:54 so
23:55 the,
23:55 the,
23:55 uh,
23:55 OK,
23:56 let me talk about the corre when the,
23:57 the,
23:57 the opposite case.
23:58 When the correlation is high,
24:00 so that means
24:02 the,
24:02 the highly productive,
24:03 the plants are facing the higher
24:06 idiosyncratic distortion.
24:09 So,
24:09 which is the,
24:10 the case of the higher
24:13 degree of misallocation.
24:14 And it's the opposite in the case of the low correlation.
24:20 The level of the the the aggregate misallocation,
24:24 when I say aggregated,
24:24 it's again,
24:25 within manufacturing
24:27 has been somewhat facilitating.
24:29 The during the 1960s,
24:31 1970s,
24:32 and the 1980s,
24:34 and then it
24:35 increased during the 200 2010.
24:39 So,
24:41 It means that
24:43 improvement in the allocated efficiency was not
24:46 the source behind the Korea's growth miracle,
24:50 but
24:51 The increase in
24:53 the degree of the misallocation was certainly behind the growth,
24:58 the slowdown.
25:00 So it doesn't explain
25:01 or it is not
25:02 consistent to the growth miracle,
25:04 but at least it is consistent to
25:06 the growth slowdown.
25:11 And we can also quantify
25:13 how much the economy is losing by comparing
25:17 actual output
25:19 and the model-based,
25:20 the efficient output.
25:22 So if there is no distortion at all,
25:25 this ratio should be one,
25:27 but in general this ratio is
25:29 below one.
25:31 And this is the mirror image,
25:33 the,
25:34 the economic loss coming from
25:37 the,
25:37 the,
25:38 the,
25:38 the plant-specific idiosyncratic distortion
25:42 has increased during the 202,010
25:46 when Korea experienced a growth slowdown.
25:52 So for remaining.
25:53 10 minutes.
25:55 So I will talk more about this,
25:58 the close,
25:58 the slowdown
26:00 period,
26:01 the 2020 test.
26:04 So 2
26:06 The
26:07 quantify
26:09 how the business dynamism has changed in Korea.
26:13 So I will use two measures.
26:16 The first measure is the job creation and destruction,
26:19 and the second measure is the responsiveness to productivity.
26:24 So I saw that in this event series,
26:27 the,
26:27 the,
26:28 the Javier Miranda already
26:30 presented his great work quantifying the job
26:33 creation and destruction and responsiveness to productivity
26:37 by using
26:38 the European data.
26:40 So when I checked his presentation slide,
26:42 I saw some similarities and the difference.
26:44 differences
26:45 between what he found and what I
26:46 found,
26:47 but I mean collectively
26:49 now
26:50 we can see how
26:52 the plant level,
26:53 the business dynamism is at least correlated or not
26:57 with
26:58 the positive economic growth or
27:00 the negative economic growth or the slowdown in the growth rate.
27:05 But I will define in the later slide what I mean by job creation,
27:09 destruction,
27:09 and responsiveness to
27:11 productivity.
27:13 And again,
27:14 due to the data limitation,
27:16 I can only report the result after the 1982.
27:20 But still,
27:21 the 1980s and early the 1990s,
27:24 the Korean,
27:24 Korean economy grew by 7%,
27:27 8% per year.
27:28 So still,
27:29 that period was the part of the gross miracle.
27:36 So we calculate
27:38 the Davis Hart Wenger growth rate of employment.
27:42 So this employment is the definition of the DHA,
27:45 the growth rate,
27:46 but it is very similar to the,
27:49 the percent growth rate,
27:51 but the advantage of that is
27:53 we can
27:55 The incorporated entry and exit decision.
27:59 Because
27:59 if we calculate the,
28:00 the growth rate in,
28:03 then we cannot include the,
28:05 the establishment that is newly showing up and establishment
28:08 exceeding because we don't have a number from there.
28:13 So I'm showing you
28:15 the DHS,
28:16 the growth rate,
28:17 it's histogram.
28:19 The x axis is from -2 xit
28:23 to 2,
28:23 which is the entry,
28:25 and this scale is closer to the low growth rate if you are familiar with it.
28:31 I'm showing you the four panels,
28:34 but I want you to look at only the top left,
28:38 the 1982,
28:40 5 years after that.
28:42 And the bottom right,
28:44 2013 and five years after that.
28:48 By looking at
28:49 the,
28:49 how the histograms
28:51 look like,
28:53 the first,
28:54 we have two observations.
28:56 The first,
28:57 we see
28:58 the decline in the mass of the entry,
29:01 so we see the decline in the entry of the new plants,
29:07 and at the same time,
29:08 we see the second observation is that we see
29:12 the more mass,
29:13 the near zero.
29:15 So through the lens of the creative destruction model,
29:19 endogenous,
29:19 the,
29:20 the grows like uh the series of the models,
29:23 the papers,
29:23 the Upuk oxygen has
29:25 the,
29:26 the worked on.
29:29 Higher entry
29:31 and then the larger mass in this histogram,
29:34 the far from the zero,
29:37 that represents the healthy,
29:39 the,
29:39 the,
29:40 the,
29:40 the meaningful,
29:42 the degree of the creative destruction.
29:45 The singing
29:47 The,
29:48 the,
29:48 the slightly,
29:50 the,
29:50 the lower mass and then the higher mass,
29:52 the near zero can be interpreted as
29:55 the decline in
29:58 the,
29:58 the degree of the creative destruction in the Korean manufacturing sector.
30:05 Another way of quantifying the degree of business dynamism is
30:10 looking at
30:11 how the plants are responding to their own productivity shock.
30:17 The following Decker Harte Wenger,
30:19 the Harmin,
30:20 and the Miranda papers,
30:22 so we estimate the responsiveness of the businesses to the shocks.
30:27 So the left-hand side is the DHS employment growth
30:30 I showed you before.
30:32 And then the right-hand side,
30:33 the main variable is this one,
30:36 which is the productivity.
30:38 So I will show you the one version using the TFPR.
30:42 As a proxy for the low productivity,
30:45 but in the background paper that we are providing,
30:48 you can also see that the result is
30:50 robust to the other,
30:51 the use of the other alternative to productivity measures.
30:57 But then the key component here is the time specific.
31:02 The coefficient.
31:03 So I will first show you the beta one.
31:06 When the plants get the positive productivity shock,
31:09 not surprisingly,
31:10 they,
31:11 they become bigger.
31:12 So I will show you
31:14 the positive significant the beta one,
31:17 but we are more interested in how that correlation has changed over time.
31:22 So I will report this coefficient
31:25 during the 1980s,
31:27 the 19990s,
31:28 2000s,
31:28 and 2010.
31:31 So this is the research.
31:33 Again,
31:34 not surprisingly,
31:35 the employment growth and the productivity at
31:39 the plant level is correlated and significant.
31:42 We also looked at the capital growth.
31:44 Not surprisingly,
31:45 it is the positive and significant,
31:48 but I want you to look at
31:50 the period by period
31:52 degree of correlation.
31:54 So it started from 0.2
31:57 or capital growth,
31:58 I mean sorry,
31:58 0.02 or capital growth to 0.2,
32:02 but it kept
32:04 decreasing.
32:05 So,
32:06 during the 2010,
32:08 the,
32:08 it was,
32:09 say,
32:10 the 30% lower than the level in the 1980s
32:14 and capital growth-wise,
32:16 it was the 50% lower than the level in the 1980s.
32:21 So that means
32:22 The plants that became
32:25 the less responsive to,
32:27 to their own productivity shock.
32:29 This is again another,
32:31 the,
32:31 the evidence that
32:33 the degree of misallocation in this sector has increased during the,
32:38 the slowdown period.
32:44 OK,
32:44 so this is my,
32:45 the last slide.
32:46 So I will give you again,
32:48 the two taking stocks from my presentation.
32:52 The first,
32:53 again,
32:54 There's no clear relation correlation between the macro level growth and
32:58 plant size exhibition or a static measure of allocated efficiency,
33:03 especially during the Korea's growth miracle period.
33:07 And second,
33:08 the growth slowdown
33:10 after the 2000 that coincides with the reduction in dynamism.
33:17 So now we need a more empirical research on
33:19 business dynamism over time and across the countries,
33:23 but to do so one needs the micro panel data for this purpose.
33:26 In this event series,
33:27 you guys have already seen the evidence from the European countries,
33:31 and I'm adding the evidence from Korea.
33:35 And then another,
33:36 the,
33:37 the,
33:37 the,
33:37 the research needed is the identifying
33:40 the frictions.
33:41 So,
33:41 here,
33:42 that we documented that
33:44 the business dynamics has declined,
33:46 but our research doesn't tell
33:48 the why that happened.
33:50 So,
33:51 the weather it was due to
33:53 the adjustment cost,
33:55 including some tax by the government or the creative cons the credit constraint,
34:00 so that is still the,
34:01 the,
34:02 the remaining area for
34:04 future
34:05 research.
34:07 OK,
34:07 so I will stop here.
34:10 And
34:11 many thanks,
34:11 Musso.
34:13 Definitely fascinating.
34:14 Uh,
34:16 I myself have
34:17 a lot of questions to ask,
34:19 uh,
34:19 but let me not be selfish here and open up the floor.
34:23 To the full audience,
34:24 um.
34:26 Uh,
34:27 if I see some.
34:29 Uh,
34:29 delays in people to,
34:31 uh,
34:31 raise their hands,
34:32 then I,
34:32 I,
34:33 I will start with my own conclusions and,
34:35 and doubts mostly,
34:37 um.
34:39 Yeah,
34:40 so why don't,
34:41 why don't I,
34:42 I,
34:42 I start then.
34:43 So,
34:45 Some parts of the presentation,
34:47 I think that
34:48 help you rationalize certain things that,
34:50 that was,
34:50 were happening and some of them I still think as puzzling that
34:54 I cannot understand how could this be going on
34:57 at the micro level.
35:00 With,
35:00 with just a little bit of a slowdown in the aggregate,
35:03 uh,
35:04 I,
35:04 I'm,
35:04 I'm referring to the later years
35:06 and the evidence you showed about skyrocketing,
35:10 skyrocketing like largely increasing misallocation
35:14 and definitely
35:16 the evidence,
35:18 uh,
35:18 points towards contributing to the
35:20 decline in,
35:21 in,
35:21 in,
35:21 in growth in slowdown,
35:23 but,
35:23 but let me,
35:24 let me begin with the first year.
35:25 11 of the things we,
35:27 we push in the WDR.
35:29 Is the idea that,
35:30 that at the,
35:32 at the
35:33 early stages of development of a country,
35:37 it's kind of somewhat expected that large incumbents might take the lead
35:41 in driving growth.
35:43 Uh,
35:44 why?
35:44 Well,
35:44 because we argue that
35:46 those are the firms that managed to have acquired some capabilities
35:50 in a very distorted environment.
35:53 So when the country
35:55 does some sort of policy that it's pro-growth,
35:58 then
35:59 Naturally,
36:00 those would be the firms to leverage the most,
36:03 uh,
36:03 the,
36:03 the new context.
36:04 So
36:05 I interpreted,
36:07 the industrial policy years as
36:10 through direct subsidization
36:12 allowing for these firms to,
36:14 to emerge
36:15 and there's no puzzle there.
36:17 We,
36:17 we understand that the,
36:19 the rising side of the size distribution is directly attributable
36:24 to the industrial policy.
36:26 Now,
36:26 a good news I would say is that as soon as you shut down the policy,
36:31 There's no problem in,
36:33 in dynamism to
36:35 emerge,
36:36 you know,
36:36 because some,
36:37 some might wonder,
36:38 well,
36:38 maybe once you create these large firms they become so entrenched
36:42 that thereafter,
36:43 even when,
36:45 when,
36:45 when the industrial policy is stopped,
36:48 then dynamism wouldn't reignite because there will be so much,
36:52 uh,
36:52 connect,
36:53 you know,
36:53 barriers to entry that,
36:55 that it,
36:55 that it would block any entrance.
36:57 To wanna come in and,
36:59 and enjoy the like the,
37:00 the,
37:00 the,
37:01 the,
37:01 the,
37:01 the level playing field.
37:03 So is this a fair statement that
37:05 somehow as soon as the industrial policy uh stopped
37:10 uh.
37:12 New entrants and,
37:13 and other firms were able to flourish
37:15 or,
37:16 or is the decline in the average size coming from the,
37:18 the,
37:19 the fact that the the Shaibos themselves started to shrink?
37:22 That's my,
37:23 my first question.
37:25 OK,
37:26 great.
37:27 So,
37:28 the first of all,
37:28 yeah,
37:29 you are
37:30 totally right.
37:31 So as an outcome of the industrial policy,
37:34 the,
37:35 the,
37:35 the,
37:36 I mean,
37:36 the,
37:36 I mean,
37:37 doing the industrial policy,
37:38 the government naturally the supported the large incumbent
37:42 and that increased the size of the,
37:45 the farms.
37:47 So,
37:48 let me go back to the basic because now we are seeing
37:51 the revival of the industrial policy in the US and other countries.
37:56 So,
37:57 the,
37:57 the textbook,
37:58 the style,
37:59 the justification for the industrial policy
38:02 is that whenever there is the externality,
38:04 the individual plants do not,
38:07 the internalize the benefit they are making by
38:11 the,
38:11 the putting,
38:12 the making,
38:12 the huge,
38:13 the fixed cost,
38:14 the,
38:14 the,
38:15 the,
38:15 the fixed investment.
38:17 But the government subsidized that.
38:19 Now
38:20 the company is able to make that high investment at that
38:24 generate the benefit
38:26 the toward the nearby plants,
38:28 and then the plants that connected through the input output,
38:31 the metrics.
38:32 So,
38:33 totally right.
38:34 So it is
38:35 the,
38:35 the,
38:36 it is natural to see
38:37 the.
38:38 The pattern in other countries too,
38:41 but what is interesting in the Koreas,
38:43 the context is that we also see the end of the industrial policy,
38:47 which is not common because
38:50 we know the several
38:52 other cases of industrial policy,
38:53 mostly in Latin America,
38:55 that failed at the end.
38:57 So
38:58 my counterfactual scenario
38:59 So if the Korea continue to do the industrial policy for 10 years,
39:03 then 20 years,
39:04 so then
39:05 you are building this political connection between Jaber and
39:08 the President Park Chung-ye for 10 to 20 years.
39:12 So then it is likely to be,
39:14 I mean,
39:14 corrupted.
39:15 I mean,
39:15 that relationship,
39:16 it's unlikely to be healthy after 10 to 20 years.
39:20 So the European Council knows about what happened after the end,
39:24 the unexpected,
39:25 the end of the industrial policy.
39:28 So,
39:29 we don't see
39:31 the,
39:31 the,
39:32 the,
39:32 the,
39:32 the decrease in the size of
39:35 the zebras.
39:36 So larger farms are,
39:37 I mean,
39:38 still
39:39 the,
39:39 the large,
39:40 but they stopped growing,
39:42 so they,
39:43 they grew a lot in the industrial policy,
39:46 but they stopped the,
39:47 the,
39:48 the growing the further,
39:49 but the,
39:50 the,
39:51 the many,
39:51 the small
39:53 and the productive,
39:54 the,
39:54 the plants and the farms enter to the market.
39:58 So the mechanically,
39:59 the decline in the average size was mostly,
40:02 were mostly driven by
40:04 the new,
40:05 the entrant.
40:07 And then my understanding is that
40:10 They could not enter during the industrial policy period,
40:14 uh,
40:14 but the,
40:15 the,
40:15 when
40:16 the,
40:16 the,
40:17 the,
40:17 during the rationalization period,
40:19 they could enter
40:20 and then they have
40:21 contributed to the next 10 to 20 years of the,
40:25 the growth miracle without the industrial policy.
40:30 Yeah,
40:31 no,
40:31 I,
40:31 but
40:32 I think it's a great insight that uh it's not well known and
40:36 of course it's not well known because it required data like this to,
40:40 to make the point
40:41 which is the short-lived nature of industrial policy
40:45 is one sort of necessary condition for
40:48 let's say argue in favor of,
40:50 of this kind of government interventions.
40:53 Uh,
40:53 the other one I would say,
40:54 and it didn't come up today too much although it did uh last week uh in New Hu's talk.
41:00 Is the
41:01 gearing of the subsidization
41:04 not only to particular sectors but also by
41:07 um a requirement if you will of export
41:11 so
41:12 in a way.
41:14 Forcing and disciplining the recipients of the subsidies
41:18 to go have to compete and export
41:20 and,
41:20 and kept in check by the globally competitive nature
41:24 of the industries where they operated I think.
41:27 That necessary condition which I think is more well understood
41:30 combined with what you teach us today that the short lived
41:33 is important are are two ingredients that
41:36 I think whoever is gonna keep advocating for
41:38 industrial policy and of course in the report we're gonna touch on this
41:42 uh we want to emphasize
41:44 um
41:45 so
41:45 again as as I don't see anyone else uh please
41:49 feel free to jump in,
41:50 um.
41:52 But I,
41:52 I,
41:53 I see a question here from
41:56 Matheus.
41:57 Matthew,
41:59 so Matthew,
41:59 please take the floor.
42:04 Yeah.
42:04 Hi,
42:05 hi,
42:05 everyone.
42:06 Um,
42:07 I'm just wondering one thing where you presented uh
42:11 some evidence there on the static allocative efficiency.
42:16 Uh,
42:17 And you presented this correlation,
42:20 uh,
42:21 increasing.
42:22 So my question to you is,
42:26 What
42:27 could be the role of quality upgrading here.
42:30 So there's a question of markups.
42:33 There's one part.
42:34 But the point is,
42:36 couldn't this positive and increasing correlation
42:41 be such that more productive firms are increasing quality
42:45 as they compete in international markets so that increase
42:49 This correlation.
42:51 Thanks.
42:54 OK,
42:55 so the,
42:56 the,
42:57 thank you for the question,
42:57 Matthias.
42:59 So,
43:00 my first answer is that,
43:02 so we don't know yet,
43:04 so why it
43:06 increased a lot during the,
43:07 the recent period.
43:09 So that's something
43:10 the Youngs and I,
43:12 the,
43:12 the plant
43:13 to work on.
43:14 And your question was the weather that is related to
43:19 the quality of grading.
43:21 So,
43:22 like
43:23 many other,
43:24 the data,
43:25 so we don't see the price.
43:27 So unfortunately,
43:28 in the data,
43:29 we don't have a direct measure
43:31 to quantify
43:32 whether this is coming from the change in the market
43:36 or change in the quality.
43:38 I mean,
43:38 even
43:39 with the price information,
43:41 if
43:41 the highly highly productive farms are now
43:44 Charging the higher price,
43:45 that could be interpreted as the markup or that
43:48 could be also interpreted as the higher quality.
43:51 So your question was already incorporating these two possibilities.
43:56 So
43:56 the first of all,
43:57 one is the price data,
43:59 and then the second one is the model
44:02 to separately quantify the role of the Markov and the quality.
44:06 I mean,
44:07 that's really the interesting,
44:09 the,
44:09 the research area.
44:10 Unfortunately,
44:11 the Korean data is not ideal
44:13 to answer that question,
44:15 but there must be the other countries,
44:17 the
44:17 context that giving you the price information,
44:20 say Colombia.
44:21 So the Colombian,
44:22 the manufacturing plant data,
44:24 so they have uh price information separately,
44:27 then at least you can overcome the data limitation,
44:29 but still it's the question.
44:31 The whenever you see the increase or decrease in the,
44:34 the price,
44:35 the,
44:35 whether you want the,
44:37 the attribute that to the change in the markup or quality.
44:41 Great question,
44:41 but unfortunately,
44:43 so we don't have an answer.
44:46 Yeah,
44:46 thanks.
44:47 And I,
44:47 I was just puzzled that to interpret this as a sign
44:52 that uh there's the allocative efficiency is not improving.
44:59 So I,
44:59 I,
45:00 maybe I misunderstood,
45:01 but,
45:02 yeah,
45:02 as you said that this,
45:04 the data is not well suited to answer that.
45:07 So
45:08 I,
45:08 I understand the constraints.
45:10 I mean,
45:11 thanks.
45:12 Thanks.
45:19 So if I,
45:19 if,
45:20 I mean,
45:20 if it's OK,
45:21 so let me add just one additional point because
45:24 the,
45:25 because now we are talking about the allocated efficiency.
45:28 So even though I reported that during the industrial policy period,
45:32 the aggregate allocated efficiency didn't change,
45:35 so that was the,
45:36 the,
45:37 the what I showed you.
45:38 During the industrial policy period,
45:39 the aggregate
45:41 level of allocated efficiency didn't change much,
45:44 but I have another paper with Jung's and
45:47 the Dr.
45:48 Minho Kim.
45:49 The comparing
45:51 the
45:52 change in the located efficiency between the
45:54 targeted industries and the non-targeted industries.
45:58 And we found that
46:00 the degree of the misallocation increased in the targeted industries,
46:05 but in the definitive sense,
46:06 comparing the targeted versus the non-targeted,
46:09 the industrial policy was somewhat.
46:13 Distorting the resource allocation across the plants.
46:17 When the government,
46:18 uh,
46:18 the,
46:18 put the resources
46:20 to the plants that,
46:22 those were not necessarily the most productive plants in the
46:26 economy.
46:27 So this
46:29 aggregate.
46:30 The
46:30 constant allocated efficiency during the industrial
46:33 policies is the masking this heterogeneity
46:36 between the targeted and non-targeted
46:38 sectors.
46:39 So,
46:40 even though
46:41 today I focused on the facts,
46:43 the aggregate facts,
46:45 but yeah,
46:45 we need more research
46:47 on looking at
46:48 how that is,
46:49 has been different across the sub-sectors of the
46:53 manufacturing.
46:55 And then may,
46:56 I may have a better answer to Matteo,
46:58 to your question because you may have some
47:00 user suspect,
47:01 the set of the sectors,
47:02 you may anticipate the,
47:04 the larger,
47:05 the quality of grading during certain period.
47:17 Anyone else wants to
47:20 make a point or a question?
47:24 I think that either way,
47:26 I,
47:26 whatever is the actual mechanism
47:29 driving that rise in the TFPR TFPQ.
47:33 Uh,
47:35 so far the stories that have been ventured,
47:38 if it's markups or if it's,
47:39 uh,
47:40 some distortion of another kind,
47:42 like,
47:43 uh,
47:44 my impression of what I have seen happening in other countries
47:49 where you can document the time series evolution of distortions
47:54 is that
47:55 you can
47:56 see a bigger response in the aggregate.
47:59 To this micro-level evidence of,
48:02 of,
48:02 of frictions and
48:04 um which in the case of Korea,
48:05 it might just tell us that
48:07 There are many other forces that we don't know that are keeping growth afloat
48:11 and therefore
48:12 the slowdown
48:14 is the marginal contribution from this channel and therefore it's notable
48:18 because these other forces were so strong for 30
48:20 years that now they're slowing down and they are,
48:23 they are there.
48:24 But
48:25 what is really puzzling to me
48:27 is the actual level to which the reversion of the size distribution
48:32 converged.
48:33 Uh,
48:33 like if you put Korea,
48:35 I mean,
48:35 if I take your latest year
48:38 of the average firm size in Korea and I put it in a cross section across countries
48:42 and plotted against GDP.
48:46 I would expect
48:47 based on Korea's average size of less than 2030 workers
48:51 to be as poor as,
48:53 uh,
48:53 you know,
48:53 Mexico or India,
48:55 uh,
48:55 you know,
48:56 even the conditioning on firms with 10+ or 5+ workers,
49:00 which is your sample.
49:02 Uh,
49:03 so that to me is really,
49:04 is really,
49:05 uh,
49:05 puzzling.
49:06 Like how is it that the size distribution makes such a big turn
49:10 to the downside
49:11 and you don't see in the aggregate,
49:14 uh,
49:14 uh,
49:15 uh,
49:15 an equivalent decline,
49:16 but,
49:16 um.
49:20 And it's,
49:20 it's let me quickly address that question by showing the slide.
49:24 So for other audience,
49:25 Robert,
49:26 the,
49:26 the,
49:26 you already know
49:28 the that sides,
49:29 the facts on the top of your head,
49:32 but you're totally right.
49:33 So this is the figure on the left is comparing the Average
49:38 employment
49:39 of Korea,
49:40 which is the yellow line,
49:41 which is below
49:43 the New Zealand,
49:44 the Mexico,
49:45 Brazil,
49:46 those are the set of the countries that allowed us to do
49:50 the April to April comparison using the OECD data.
49:53 So it is so true that
49:56 where Korea is in terms of the average employment is much lower than
50:01 other countries,
50:03 including the medium income countries.
50:06 So again,
50:07 there could be the multiple,
50:09 the hypothesis.
50:10 The one hypothesis is the concentration,
50:13 the degree of the concentration.
50:15 So
50:16 the,
50:18 the,
50:18 the,
50:18 the,
50:18 the big conglomerates are leading
50:21 the development of the manufacturing
50:24 sector,
50:25 and the mirror image of that is that
50:28 the rea is suffering from the self-employment.
50:31 What do I mean by the self-employment?
50:33 The,
50:34 the
50:35 When the manufacturing workers lose their job.
50:39 It is,
50:40 they struggle to find the new employer.
50:43 When you lose the job from the manufacturing sector at the age of 40 to 50,
50:48 the labor market is not that dynamic.
50:51 And then
50:52 they.
50:54 Many of them,
50:55 the,
50:55 the,
50:56 the,
50:56 the,
50:57 the become the self-employed.
50:59 So if they have a specific skill,
51:01 so they just open the small-scale store,
51:05 or
51:06 that they open the,
51:07 the,
51:07 the small scale,
51:08 the restaurant or the supermarkets.
51:11 So
51:12 that,
51:13 the large,
51:13 the size of the self-employment,
51:15 self-employment rate in Korea is the largest among the OECD countries.
51:19 It's comparable to Mexico actually.
51:21 So,
51:22 the,
51:23 the one side,
51:24 the,
51:25 the,
51:25 the,
51:26 there is the,
51:26 there are
51:27 the largest sites,
51:28 the
51:29 big conglomerates,
51:30 and on the other side,
51:32 there are large sites of the self-employment.
51:35 I think there is so much
51:37 the,
51:37 the pushing these size of the average employment down compared to
51:42 the other countries,
51:43 but again,
51:44 that's the one hypothetis we haven't scientifically
51:48 tested it.
51:49 But you are totally right.
51:50 So that's another interesting
51:52 observation from the Korea's experience.
51:59 Yeah,
52:00 very interesting.
52:00 Thanks for bringing up the fear.
52:02 Uh,
52:02 I put numbers to
52:04 what I was trying to
52:06 convey to the broad audience,
52:07 but
52:12 And related to the allocative efficiency,
52:16 the puzzle,
52:16 I mean,
52:17 to me,
52:17 it's the puzzle,
52:18 so it's very fair to say the puzzle.
52:19 But what Jungs and I,
52:21 we work first on is the
52:24 trying the different measurements.
52:25 So you know that there has been the development after the Shanreno paper.
52:30 So when there is the measurement error in the output input,
52:33 so how they overcome that,
52:34 the base reno,
52:35 they have a paper,
52:36 and then we can also use the decreasing return
52:38 to scale instead of the constant return to scale.
52:41 So I think the first thing
52:43 we need to do is
52:45 the,
52:45 the,
52:45 the,
52:46 the checking whether this huge increase is coming from the measurement error
52:52 or not.
52:52 So,
52:53 the,
52:53 we will,
52:54 the use other the methodologist,
52:56 I mean.
52:56 Extension of the Shang reno
52:58 and check whether we still see
53:01 that
53:02 we don't expect to change the,
53:04 the,
53:04 in the,
53:05 the result the qualitatively,
53:07 but quantitatively the result may change
53:09 so we may,
53:10 yeah,
53:10 we want to check that.
53:14 OK,
53:15 great moon,
53:16 um.
53:17 I believe that
53:19 people have had enough time to raise more hands,
53:22 so if they didn't,
53:24 I,
53:24 I take it as we are ready to.
53:27 To close the seminar,
53:28 um.
53:30 Thank you so much.
53:31 We've learned a lot and most importantly,
53:33 we have a lot more to keep learning,
53:35 so that's the best seminars,
53:36 the ones that teach you something but also keep you thinking.
53:39 So,
53:39 um,
53:40 thank you,
53:41 everyone,
53:41 uh,
53:42 who made it until now for joining and please stay tuned for
53:46 more seminars,
53:47 uh,
53:47 from the WDR series.
53:49 See you all soon.
53:50 See you.
53:51 Thank you.
53:52 Thank you so much.
53:53 I'm looking forward to the further engagement with the WDR.
53:57 Bye-bye.
53:58 Awesome.
53:59 Thanks.
53:59 Bye.
54:00 Thank you.
- add-style
- lp-body-content