Popular in Economics

Banks and Policy Organisations

Macroeconomic modelling of CBDC: a critical review

European Central Bank Ulrich Bindseil

Banks and Policy Organisations

Financing Private Credit

Federal Reserve Bank of New York Nina Boyarchenko

Economics

Failing Banks

Massachusetts Institute of Technology Emil Verner

Economic History

Golden Fetters

UC Berkeley Barry Eichengreen

Economic Inequalities

Visions of Inequality

City University of New York Branko Milanovic

Economics

School Finance Equalization Increases Intergenerational Mobility

Yale University Barbara Biasi

Economics

Olivier Blanchard on U.S. Pandemic-Era Inflation

Peterson Institute for International Economics Olivier Blanchard

Automation and the Rise of Superstar Firms

View more insights in Policy Organisations - Social Sciences - Economics - Banks and Policy Organisations - Labor Economics - Macroeconomics

View more figures from Federal Reserve Bank of San Francisco

View the page of Zheng Liu

Next in Banks and Policy Organisations

Geopolitical risk perceptions

Yevheniia Bondarenko | Deutsche Bundesbank

Industrial Policies for Innovation: A Cost-Benefit Framework

Daniel Garcia-Macia | International Monetary Fund

Zheng Liu discusses evidence that the rise in automation technology contributed to the rise of superstar firms in the past two decades. Liu further explains the empirical link between automation and industry concentration in a general equilibrium framework with heterogeneous firms and variable markups.

Read the Study

Image courtesy of interviewee. July 12, 2023

Log-in or Sign-up to Faculti
Currently viewing this subject insight as a guest. You have insight(s) remaining for this month. Login to view 8000+ figures on the platform.

Generating Workspace...

Export as PDF

The paper examines two significant trends in the U.S. economy over the past two decades: the rise of automation, particularly through robotics, and the concurrent increase in industry concentration dominated by a few large firms. These trends are observed to be interconnected, with automation technologies being adopted predominantly by larger firms, which in turn fosters further concentration in various industries.

The study poses questions about whether these trends are mere coincidences or if they have a causal connection, exploring how automation contributes to increased sales concentration among top firms while having an insignificant effect on employment concentration. It outlines a theoretical framework where firms make choices between adopting labor-only technology or automation technology, recognizing that larger firms are better positioned to invest in costly automation due to their greater productivity and capacity to absorb fixed costs.

Data indicate that the declining robot prices have primarily benefited larger firms, enhancing their productivity and leading to greater market power—thus increasing industry concentration. The authors utilize an instrumental variable approach informed by earlier robot adoption in European countries to establish causality in the observed relationships. The findings suggest that about 80% of the increase in sales concentration in U.S. manufacturing can be attributed to the decline in robot prices.

In considering policy implications, the paper discusses whether taxing robots would be beneficial. It suggests that while a tax could address market inefficiencies driven by large firms' market power, it may also hinder productivity gains. The authors propose that a small subsidy for businesses utilizing robots could enhance social welfare by balancing these conflicting interests. Overall, the research highlights the dynamic interactions between technological advancement and market structure, advocating for nuanced policy responses to maximize economic welfare.

Language:

Export as PDF

Export as PDF
Here are five key points from the video:

1. **Trends in the US Economy**: The paper discusses two significant trends in the US economy over the past 20 years: the rise of automation, particularly through increased robot adoption and declining robot prices, and the rise of industry concentration, with a few large firms dominating major sectors of the economy.

2. **Connection Between Automation and Industry Concentration**: The study investigates whether the trends of automation and industry concentration are merely coincidental or inherently linked. It finds that the adoption of automation technologies, especially by large firms, correlates with increased industry concentration, as these firms have the resources to invest in costly automation technologies.

3. **Causal Effects**: The authors employ instrumental variable techniques to establish that automation has a positive causal effect on sales concentration, measured by the share of top firms in industry sales, though the effect on employment concentration is minimal. This suggests that automation benefits larger firms more than smaller ones, contributing to their growing market dominance.

4. **Modeling Firm Behavior**: The theoretical framework developed in the paper presents firms of varying sizes and productivity levels making technology choices. Larger firms tend to adopt automation technologies more readily due to their financial capabilities, leading to economies of scale that further enhance their productivity and market power.

5. **Policy Implications**: The study concludes by discussing the implications for policy, particularly regarding the taxation of automation (robots). While taxing robot usage could address market inefficiencies arising from large firms' market power, it could also hinder overall productivity growth. The authors recommend a modest subsidy for robot-using firms to balance these trade-offs and maximize social welfare.

Export as PDF
Here are some key quotes extracted from the video:

1. **Motivation for the Study**:
- "The paper is motivated by, uh, the observation that there have... been two remarkable trends in the US economy over the past 20 years."

2. **Trends Identified**:
- "One trend is the rise of, uh, automation... the number of industrial robots per thousand manufacturing workers... has quadrupled since 2000."
- "The real price of robots... has declined by over 40% during the same period."
- "The other trend that is interesting and remarkable is the rise of industry concentration... increasingly been dominated by a small number of superstar firms."

3. **Connection Between Trends**:
- "What's interesting to us is that these two trends appear to be connected."

4. **Causal Relationship Inquiry**:
- "Are these trends just coincidence or are they inherently connected?"

5. **Definition of Superstar Firms**:
- "A superstar firm is a company with a dominant presence in industry... typically have higher margins... and they have lower labor income share."

6. **Effects of Automation**:
- "Automation could also displace workers or reduce wages and labor income share as well."

7. **Analysis and Identification of Trends**:
- "There are two potential sources of the increasing concentration. One comes from the demand side, the other from the supply side."
- "Large firms are typically also more productive... since they are more capable of investing in new technologies such as automation."

8. **Implications for Policy**:
- "If the increasing concentrations primarily driven by factors such as market power, then it's not a good thing. It's inefficient, and it will call for policies such as, uh, antitrust regulations."
- "But if this consequence of, uh, more increased technology adoptions, then the policy prescription should obviously be different."

9. **Empirical Investigation**:
- "We further investigate the potential causal effects of automation on industry concentration."

10. **Quantitative Findings**:
- "We find that automation does have a causal positive effect on sales concentration measured by the top 1% firm's share."
- "A 40% decline in the robot price... could explain about 80% of the increase in sales concentration in the US manufacturing sector."

11. **Policy Recommendations**:
- "Taxing robots can actually help undo part of the inefficiency related to market power."
- "We find that a modest subsidy of about 5% for robot using firms would maximize social welfare."

These quotes highlight the main themes and findings of the presented study, including the trends of automation and industry concentration, their connection, and implications for policy-making.

Export as PDF
Here are some articles and studies that are similar in topic to the video you , which discusses the trends of automation, industry concentration, and their economic implications:

1. **"Rise of the Robots: Technology and the Threat of a Jobless Future" by Martin Ford** - This book explores the impact of automation and artificial intelligence on the economy and job market, including issues of job displacement and the concentration of wealth.

2. **"The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies" by Erik Brynjolfsson and Andrew McAfee** - This book examines how digital technologies are reshaping industries, creating wealth gaps, and influencing productivity and employment patterns.

3. **"Automation and Job Loss: How Expecting a Decrease in the Employment Rate Can Deal with the Problem" by Susan Houseman** - This article discusses empirical evidence regarding automation and its impact on employment, providing a perspective on labor market shifts as automation technologies are adopted.

4. **"Superstar Firms and the Global Economy" by Klaus Schwab and others** - This report highlights how a small number of firms are commanding a disproportionate share of the economy, discussing the implications for competition and market dynamics.

5. **"The Impact of Automation on Employment: A Review of the Evidence" by David Autor, Frank Levy, and Richard Murnane** - This paper reviews the relationship between technological change, automation, and labor market outcomes, assessing the effects on wages and employment structures.

6. **"Market Power, Technological Change, and the Decline of Labor Share" by John Grullon, Yelena Larkin, and Irem Tuna** - This academic paper investigates the relationship between market power, technology adoption, and labor income share, which parallels themes in your video.

7. **"Technology, Market Structure, and the Distribution of Income" by David Dorn, Lawrence F. Katz, and Cristina Patrinos** - This study explores the effects of technology and market structure on income distribution, discussing the implications for labor and capital.

8. **"The Automation Adventure: Understanding the Impacts of Artificial Intelligence on Business" by various authors** - This compilation of studies explores how AI and automation technologies are revolutionizing industries, impacting firm size, productivity, and market dynamics.

These articles delve into the themes of automation and its economic consequences, industry concentration, labor market implications, and policy responses—similar to the content you've shared.

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:06.900 --> 00:11.040
The paper is motivated by, the observation that the there have,

00:11.210 --> 00:16.080
there have been two remarkable trends in the US economy over the

00:16.110 --> 00:20.880
past 20 years. One trend is the rise of,

00:20.880 --> 00:24.960
automation. Uh, and, we can see that,

00:24.960 --> 00:29.680
there's a steady increase in robot adoptions and the steady

00:29.790 --> 00:33.970
decline in robot prices. And,

00:33.970 --> 00:38.250
this happens starting in early two thousands. Um, for example,

00:38.390 --> 00:43.090
the number of industrial robots per thousand manufacturing workers is aggregate

00:43.090 --> 00:48.050
data has quadrupled since 32,000. And,

00:48.050 --> 00:50.930
the real price of robots, which is inflation adjusted price,

00:51.270 --> 00:55.730
the cost of robots has declined by over 40% during the same period.

00:57.070 --> 01:01.890
The other trend that is interesting and remarkable is the rise of

01:02.210 --> 01:03.330
industry concentration.

01:04.840 --> 01:09.540
And each major sector of the economy has increasingly been

01:09.540 --> 01:12.940
dominated by a small number of superstar firms.

01:14.600 --> 01:18.540
And what's interesting to us is that these two trends appear to be connected.

01:18.950 --> 01:20.750
There is,

01:21.520 --> 01:26.510
there are some recent census studies that show that at the firm level in

01:26.510 --> 01:31.430
the United States adoption of automation technologies, in particular,

01:31.850 --> 01:36.550
AI and robotics are heavily skewed toward a

01:36.650 --> 01:39.550
few large firms. They're not widespread,

01:39.890 --> 01:43.470
but they're heavily concentrated in a few large firms.

01:45.250 --> 01:50.110
So our study is basically motivated by these observations. We, we,

01:50.450 --> 01:52.230
ask a question, you know,

01:52.410 --> 01:57.380
are these trends just coincidence or are they inherently

01:57.970 --> 02:02.100
connected? If they're connected, what might explain the connection?

02:02.870 --> 02:03.200
Well,

02:03.200 --> 02:07.960
a superstar farm is a company with a dominant presence in industry.

02:08.460 --> 02:12.720
You can think, of Apple or Walmart in their own industry.

02:13.590 --> 02:18.240
This farms usually account for disproportionally large share of revenues,

02:18.460 --> 02:20.520
in, in their own industries.

02:21.640 --> 02:26.180
And this superstar firms typically have higher margins,

02:26.180 --> 02:31.060
profit margins or markups, and they have lower labor income share.

02:31.600 --> 02:35.580
So, there are recent studies by order at all.

02:36.130 --> 02:40.100
They're saying, well, superstar firms are rising. And,

02:40.130 --> 02:44.940
then the conseque reallocation of resources towards these

02:44.940 --> 02:49.340
large farms with low labor share, how be explained,

02:49.960 --> 02:51.980
the decline, aggregate labor, income share,

02:52.480 --> 02:57.170
and the rise in aggregate markup. And,

02:57.170 --> 02:58.650
there are also other studies,

02:58.810 --> 03:01.660
I I'm gonna talk about the literature while we at this,

03:01.800 --> 03:03.780
on the superstar firm and automation.

03:04.250 --> 03:08.060
Some other empirical studies also find is,

03:08.430 --> 03:10.180
automation has some similar effect.

03:11.870 --> 03:16.760
Adopting automation technologies have improved productivity like grades might

03:16.760 --> 03:20.960
cause in 2018, showed that as IOUs travel also showed that,

03:21.420 --> 03:26.280
but automation could also displace workers or reduce wages and labor

03:26.280 --> 03:31.080
income share as well. So this, this studies,

03:31.080 --> 03:32.720
you know, suggest,

03:33.150 --> 03:38.080
both industry concentration increase and the

03:38.210 --> 03:42.880
automation rise have important consequences for productivity

03:43.540 --> 03:47.320
and the allocation income. And,

03:47.410 --> 03:51.800
under understanding the cost of the rise in industry concentration is very

03:51.800 --> 03:55.880
important for policy designs. Uh, and,

03:55.890 --> 04:00.240
there are two potential source of the increasing concentration.

04:00.780 --> 04:04.440
One comes from the demand side, the other comes from the supply side.

04:05.190 --> 04:08.730
On the demand side, large firms have high market power.

04:10.060 --> 04:14.630
They want to maintain their market power by producing less,

04:15.630 --> 04:19.860
by raising price, or else being equal on the supply side. However,

04:20.630 --> 04:23.380
large firms are typically also more productive.

04:25.560 --> 04:29.900
So this high productivity help them to contain costs.

04:31.610 --> 04:35.510
And since they are more capable of investing in new,

04:35.610 --> 04:38.690
new technologies such as automation,

04:39.430 --> 04:42.090
and those technologies are costly.

04:42.670 --> 04:44.770
So by investing these new technologies,

04:45.070 --> 04:48.170
it helps 'em to improve productivity further.

04:50.180 --> 04:55.120
So if the increasing concentrations primarily driven by factors such as

04:55.120 --> 04:59.520
market power, then it's not a good thing. It's inefficient,

04:59.780 --> 05:04.080
and it will call for policies such as, antitrust regulations.

05:05.270 --> 05:07.370
But if this consequence of,

05:07.560 --> 05:10.370
more Inc or increased technology adoptions,

05:11.000 --> 05:14.050
then the policy prescription should obviously be different.

05:14.350 --> 05:16.410
You should not tax them. Instead,

05:16.410 --> 05:20.410
you should perhaps encourage 'em to do more of the technology adoption,

05:20.830 --> 05:22.290
expand their production further.

05:23.650 --> 05:28.030
So there's a recent strand of literature that our studies closely related to

05:28.650 --> 05:30.340
that literature argue that

05:31.940 --> 05:36.880
the rise of industry concentration is linked to technology adoptions,

05:37.270 --> 05:40.960
such as automation through an economy of scale effect,

05:41.140 --> 05:43.120
or sometimes known as network effect.

05:44.060 --> 05:48.120
And the order at the co-authors that we cite in the paper and the Uber

05:48.270 --> 05:51.080
registrable and the all, they, they all

05:52.990 --> 05:56.800
explain this idea. So the basic idea is large firm

05:56.820 --> 06:01.350
Has sufficient resources to invest in new technologies that might be too costly

06:01.350 --> 06:06.190
for small firms. And, by yeah, gaining a,

06:06.310 --> 06:10.470
a new, advantage in new technology, a large firm can grow even larger.

06:11.450 --> 06:16.310
So this, this is, this idea, we build on this idea. And,

06:16.690 --> 06:21.350
just to dig deeper in the specific area,

06:21.350 --> 06:26.110
of automation. Main data source that we rely on is,

06:26.180 --> 06:28.070
comp compute stat,

06:28.200 --> 06:32.190
which is a dataset that contain all the publicly listed firms.

06:33.050 --> 06:34.430
Um, and,

06:34.810 --> 06:39.710
we construct a measure of industry concentration based on the share

06:39.930 --> 06:43.870
of sales of the top 1% of firms. Uh,

06:43.870 --> 06:45.830
we also do the same for employment,

06:45.890 --> 06:50.670
the share of employment of the top 1% of firms. And,

06:51.010 --> 06:52.070
we measure,

06:52.080 --> 06:56.670
automation by the number of industrial robots per thousand workers

06:57.290 --> 07:00.790
in each manufacturing industry. And the, the,

07:00.790 --> 07:03.430
the data is limited to manufacturing because,

07:03.430 --> 07:08.270
we take the robot data from the International Federation of Robotics or if r

07:08.970 --> 07:13.950
and, that data source give us mostly manufacturing,

07:14.040 --> 07:17.790
industries. And, our,

07:18.530 --> 07:22.700
if you look at the raw correlations, you construct robot den density,

07:22.870 --> 07:25.540
which is a number of robots per thousand workers.

07:26.240 --> 07:30.740
And you correlated with those, measures of industry concentration.

07:31.440 --> 07:35.900
You can see a significantly positive correlation between robot density

07:36.400 --> 07:40.140
at the industry level and the top 1% share in sales.

07:41.840 --> 07:46.220
But the correlation with employment concentration is, less significant. Uh,

07:46.840 --> 07:50.900
it is, statistically it's not different from zero,

07:51.840 --> 07:54.900
of course, you know, the correlations are not,

07:54.900 --> 07:59.620
necessarily causal relations. Um, so,

07:59.880 --> 08:01.700
mainly because, you know,

08:01.810 --> 08:06.660
both robot density and the industry concentration could be driven.

08:06.730 --> 08:09.500
Some other factors,

08:09.510 --> 08:14.020
could be changes in productivity or changes in demand. So,

08:14.360 --> 08:19.220
we further investigate the potential causal effects of automation

08:19.720 --> 08:24.660
on industry concentration. To do that, we need,

08:24.660 --> 08:27.380
some instrumental variable,

08:27.380 --> 08:30.540
an instrumental variable in simple language is something that is,

08:30.750 --> 08:32.820
correlated with, robot density,

08:33.440 --> 08:37.700
but not directly correlated with the industry concentration.

08:39.280 --> 08:41.300
And, the idea of constructing this,

08:41.340 --> 08:45.820
I IV or instrumental variable is based on,

08:45.900 --> 08:49.500
a study by ACLU and RA in 2020.

08:50.330 --> 08:52.980
They noted that, there's a,

08:53.220 --> 08:56.740
a wide variation across industries of robot adoptions.

08:57.400 --> 09:01.380
Not every industry adopt robots quickly. And,

09:01.610 --> 09:06.260
there's a common set of industries like automobile in both US

09:06.400 --> 09:11.180
and Europe experienced fast robot adoptions in the past two

09:11.180 --> 09:14.770
decades. So importantly,

09:15.690 --> 09:19.710
the European economies adopted robots well ahead of the us.

09:21.130 --> 09:25.750
Uh, like Germany, France, they have, earlier,

09:25.750 --> 09:30.550
their earlier adopters of robots. So that gives us information,

09:31.010 --> 09:35.550
from the, the European robot, adoption,

09:35.850 --> 09:40.590
or you measure it by robot density in the industry should

09:40.700 --> 09:42.790
lead that of the US adoption.

09:44.470 --> 09:48.310
So we use the lagged European robot density

09:49.410 --> 09:54.270
in the industry as an instrument for the

09:54.300 --> 09:58.850
same, so robot density in the same industry in the us. Uh,

09:59.080 --> 10:03.890
this is, it turns out to be, a valid, and good instrument,

10:04.350 --> 10:09.290
by statistical measures. So if you do the IV this, this way,

10:09.910 --> 10:12.050
the in instrumental variable estimation,

10:12.400 --> 10:17.330
then we find that automation does have a causal positive causal effect

10:17.870 --> 10:21.890
on sales concentration measured by the top 1% firm's share.

10:22.790 --> 10:26.730
But, the causal effect on employment concentration is,

10:26.960 --> 10:31.090
insignificant. We try to understand what might be,

10:31.090 --> 10:35.890
might explain the link between automation and industry concentration,

10:36.250 --> 10:40.410
documented, in, in our, dataset. And for that purpose,

10:40.590 --> 10:43.450
we construct a theoretical framework,

10:43.600 --> 10:48.410
that features firms with different sizes and different levels of productivity,

10:49.270 --> 10:53.890
and they make decisions between, which technology to use,

10:54.190 --> 10:58.730
how much to produce. Uh, the technologies available are two types.

10:59.110 --> 11:03.970
One is, labor only technology that doesn't require any automation equipment.

11:04.550 --> 11:05.490
The other is,

11:05.490 --> 11:10.210
automation technology that requires both labor and automation

11:10.210 --> 11:11.043
equipment.

11:11.110 --> 11:15.450
And I just call it robots as inputs and robots.

11:15.450 --> 11:19.130
And the workers are, substitutable inputs,

11:19.390 --> 11:22.050
but they're not perfect substitutes just to be realistic.

11:22.350 --> 11:26.530
And we calibrate the elasticity of substitution based on other studies.

11:28.570 --> 11:33.220
Well, in the model adopting automation technology requires first

11:33.830 --> 11:37.830
purchasing a robot at the market price. But it also,

11:38.370 --> 11:40.790
in order to operate in the automation technology,

11:41.370 --> 11:44.230
it requires the firm to pay not just workers,

11:44.930 --> 11:49.710
but also the cost of fixed cost of maintaining the robots

11:50.130 --> 11:54.060
or keep it running, like using energy. Those we don't model.

11:54.320 --> 11:58.540
So we just assume it's a fixed cost for, using the op,

11:58.540 --> 12:03.100
automation technology. And, with this,

12:03.100 --> 12:06.940
fixed cost of automation of using automation technology,

12:07.430 --> 12:12.400
there comes a economy of scaling, fact that I just mentioned.

12:13.540 --> 12:17.040
And, large firms are able to pay the fixed cost.

12:17.430 --> 12:19.160
They're more capable of paying that,

12:19.780 --> 12:23.400
and therefore they're more likely to use the automation technology.

12:24.210 --> 12:28.550
But using that technology help them to improve productivity cuz they can produce

12:28.550 --> 12:33.440
more output for any given number of workers. And,

12:33.590 --> 12:36.320
that leads to an economy of scale effect.

12:36.700 --> 12:41.160
That's crucial for explaining the link with automation and industry

12:41.480 --> 12:44.960
concentration To just think a little deeper in the model mechanism,

12:46.250 --> 12:47.550
you think about on average,

12:47.760 --> 12:51.720
large firms are more productive and they're more profitable.

12:52.260 --> 12:56.800
So they're better capable of investing in this automation technology cuz they

12:56.800 --> 13:01.320
can afford to fix cost. And now when you think about a,

13:01.680 --> 13:04.880
a robot price decline, as we observe in the data,

13:04.930 --> 13:08.680
robot price decline steadily, a decline.

13:08.770 --> 13:13.400
Robot price benefits primarily those large firms that do use robots.

13:14.880 --> 13:19.040
A small firm doesn't use robot. If it doesn't use robot, it does not matter. Uh,

13:19.160 --> 13:23.680
a robot price decline might induce some firms at the margin to increase robot

13:23.880 --> 13:28.320
adoption, become a robot user. But those existing robot users,

13:28.570 --> 13:32.080
which are large farms, will benefit most.

13:33.700 --> 13:38.320
And the increased user robots would improve those farms labor productivity

13:38.700 --> 13:42.960
and allow the large farms to become even larger. And this,

13:42.960 --> 13:47.880
economy scale effect leads to an increase in concentration and explain

13:48.020 --> 13:52.000
the connection between robot use and,

13:52.420 --> 13:56.080
the increase in the share of top 1% firms in sales.

13:56.780 --> 14:01.000
But now come to the empirical fact that the connection between automation

14:01.580 --> 14:05.400
and employment concentration is not significant.

14:05.700 --> 14:07.320
We try to find,

14:07.580 --> 14:12.440
use the same approach to see the correlation or the causal relation between

14:12.440 --> 14:16.240
automation and employment concentration. We don't find any in the data.

14:16.860 --> 14:21.240
So the model has, has also correctly predicts that pattern because,

14:22.420 --> 14:24.320
for large firms that use robots,

14:24.740 --> 14:27.720
robots can substitute partially substitute for labor.

14:28.910 --> 14:33.610
And the expansion of those large farms therefore relies more

14:33.870 --> 14:36.250
on robot use than worker use.

14:37.310 --> 14:41.530
So their sales share rises quickly that that's because they're using more

14:41.550 --> 14:43.650
robots. When the robot price declines,

14:44.290 --> 14:47.220
they don't need to rely on workers as much.

14:47.640 --> 14:50.140
So the employment concentration does rise,

14:50.560 --> 14:55.370
but it rise by much less than sales concentration. So,

14:55.750 --> 14:58.890
this is connected to the literature I mentioned earlier.

14:59.230 --> 15:02.330
The concept of the scale economy is not new.

15:02.830 --> 15:05.330
Is this associated with new technology adoption? You know,

15:05.520 --> 15:09.690
when the large firms are more capable of investing in new technologies in

15:09.690 --> 15:12.450
general, and that allowed them to grow fast,

15:13.350 --> 15:17.180
our contribution is mainly quantitative.

15:18.480 --> 15:19.160
We,

15:19.160 --> 15:23.940
our paper shows that in a carefully calibrated the model with realistic

15:24.300 --> 15:25.133
features like,

15:25.570 --> 15:30.300
heterogeneous firms and who are making technology choices

15:30.450 --> 15:33.460
between automation, technology, labor only technology.

15:34.760 --> 15:37.540
And this model, if you calibrate it carefully,

15:37.540 --> 15:42.340
you can actually match some real world observation such as the one

15:42.580 --> 15:46.140
documented by the census study, recent sensor study by ular,

15:46.160 --> 15:51.020
all which is automation is highly concentrated in a few

15:51.310 --> 15:53.420
large firms, it's not widespread.

15:54.640 --> 15:59.620
And the ability of the model to match this firm level distribution of

15:59.620 --> 16:01.580
automation is actually,

16:03.460 --> 16:05.220
evidence we interpret as,

16:05.230 --> 16:09.900
suggesting that model is a plausible framework for addressing the

16:10.330 --> 16:13.340
link between, between automation and industry concentration.

16:14.240 --> 16:17.940
So now I come to the quantitative conclusion. Um,

16:18.480 --> 16:20.940
we find that in the calibrated model,

16:21.880 --> 16:26.840
a 40% decline in the robot price comparable to the magnitude to observe

16:26.900 --> 16:28.360
in the past 20 years,

16:29.490 --> 16:34.000
would expl could explain about 80% of the increase

16:34.820 --> 16:38.240
in sales concentration in the US manufacturing sector,

16:39.060 --> 16:43.720
and also explain about 60% of the diverging trends between

16:43.770 --> 16:48.680
sales concentration and employment concentration. So,

16:48.860 --> 16:52.800
we also study the policy question,

16:52.800 --> 16:57.280
based on the Calgary model. We ask a question, should,

16:57.660 --> 17:00.840
should we tax robots? Well,

17:01.340 --> 17:05.640
our framework is able to provide an answer, because the,

17:05.640 --> 17:10.550
the model implies that large firms are more likely robot

17:10.550 --> 17:15.090
users, they have a higher markup, but markup is not a good thing.

17:15.710 --> 17:17.650
That's a potential source of inefficiency.

17:18.600 --> 17:23.180
So taxing robots can actually help undo

17:23.330 --> 17:26.980
part of the inefficiency related to market power.

17:28.680 --> 17:30.420
But at the same time,

17:31.850 --> 17:33.880
large firms are also more productive.

17:35.910 --> 17:40.530
Taxing robot using large firms would reallocate

17:40.850 --> 17:45.250
resources away from the more productive firms to lose less productive farms.

17:45.590 --> 17:47.650
And that reduce aggregate productivity

17:49.060 --> 17:53.910
Gain, you know, mitigating the markup distortion, you lose,

17:54.570 --> 17:55.990
aggregate productivity.

17:57.010 --> 18:01.960
And this tradeoff between markup and productivity implies

18:02.020 --> 18:06.640
an interior optimum of tax or subsidy rates,

18:06.640 --> 18:09.600
automation. And, with our calibration,

18:10.180 --> 18:14.880
we find that a modest subsidy of about 5% or under 5%

18:15.260 --> 18:19.120
for robot using firms would maximize social welfare.

18:20.300 --> 18:22.840
Uh, of course, we, we, we are aware that the,

18:22.990 --> 18:26.800
this particular result is specific to our calibration,

18:27.990 --> 18:31.710
although we argue that our calibration is empirically plausible.

Faculti AI can make mistakes

Faculti

Popular in Economics

Macroeconomic modelling of CBDC: a critical review

Financing Private Credit

Failing Banks

Golden Fetters

Visions of Inequality

School Finance Equalization Increases Intergenerational Mobility

Olivier Blanchard on U.S. Pandemic-Era Inflation

Automation and the Rise of Superstar Firms

Next in Banks and Policy Organisations

Guide

Platform and Category Pages

Contact Us

FAQs

What is Faculti?

How do you select interviews?

Interview Process

Page Structure

Popular in Economics

Guide

Platform and Category Pages

Contact Us

FAQs

What is Faculti?

How do you select interviews?

Interview Process

Accessibility Options

Contrast Adjustment

Text Size Adjustment

Text Spacing Adjustment

Line Height Adjustment

Cursor Size Adjustment

Page Structure Options

Page Structure