Popular in Economics

Central Monetary Authorities

Macroeconomic modelling of CBDC: a critical review

European Central Bank Ulrich Bindseil

Central Monetary Authorities

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

Industrial Policies for Innovation: A Cost-Benefit Framework

View more insights in Global and Regional Policy Authorities - Macroeconomics

View more figures from International Monetary Fund

View the page of Daniel Garcia-Macia

Read the Study

Next in Global and Regional Policy Authorities

Dynamic Exports and Labor Markets for Inclusive Growth in Cambodia

Gladys Lopez-Acevedo | World Bank

Human-centric research and innovation for the manufacturing sector

Seán O'Reagain | European Commission

When and how should governments use industrial policy to direct innovation to specific sectors? Daniel Garcia-Macia and Alexandre Sollaci discuss a framework for analyzing the costs and benefits of industrial innovation policies, revealing that sector-specific fiscal support outperforms sector-neutral support only under stringent conditions while showing that most advanced economies, including China and the U.S., tend to over-subsidize innovation despite broadly targeting the right sectors.

Image courtesy of interviewee. November 13, 2024

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

This video discusses the increasing trend of industrial policies aimed at fostering innovation in specific sectors to enhance productivity growth while addressing issues such as security and climate change. It highlights various national programs such as the CHIPS Act (US), K Chips Act (Korea), and the EU's Green Deal, which focus on sectors like semiconductors and green technology.

Many of these policies include fiscal incentives, such as R&D subsidies and tax cuts, but pose risks due to their high costs, especially amid existing government debts. The authors propose a quantitative framework to help governments determine how and when to effectively implement industrial policies, emphasizing the importance of targeting sectors with significant knowledge spillovers and social benefits.

The paper also addresses potential pitfalls, such as misallocation of resources due to political influence or lack of information, which can diminish the effectiveness of these policies. Historical examples demonstrate that poorly executed industrial policies often fail. The authors advocate for cautious implementation, clear objectives, and an emphasis on measuring external benefits and domestic spillovers.

Key recommendations include strengthening technical capacity in policymaking, setting benchmarks for progress, and ensuring accessible support across firms to foster competition. Future research should explore the broader implications of industrial policies, such as trade barriers and their influence on international dynamics, especially in a context of geo-economic fragmentation. This video concludes that effective industrial policy is contingent upon clear goals, robust sector targeting, and preventing misallocation to maximize social welfare and innovation outcomes.

Language:

Export as PDF

Export as PDF
Here are 5 key points from the video regarding industrial policy and innovation:

1. **Rise of Industrial Policy**: There is a growing trend of countries implementing new industrial policies aimed at boosting innovation in specific sectors, intending to address productivity growth, security concerns, and climate change. Examples include the CHIPS Act in the U.S. and the Green Deal Industrial Plan in Europe.

2. **Framework for Evaluation**: The development of a quantitative framework is proposed to help governments effectively allocate resources for industrial policies by identifying which sectors benefit most from targeted innovation support. The framework should consider both the benefits and the risks, including potentially misdirected support due to insufficient information or political influence.

3. **Importance of Knowledge Spillovers**: Different sectors generate varying levels of research spillovers that can benefit the wider economy. Certain sectors, like basic science, are highlighted for their higher potential to create spillovers, and policies should be designed to maximize support for these high-spillover sectors.

4. **Risks of Misallocation**: Poor implementation of industrial policies can negate potential benefits, with risks such as political capture leading to misallocation of resources. Both political influence and random mistakes in policy execution can significantly hinder the effectiveness of industrial policies.

5. **Conditions for Effective Policies**: For industrial policies to be effective, several conditions must be met: the externalities of targeted sectors must be identifiable and measurable, domestic knowledge spillovers must be substantial, and government capacity must be strong enough to manage potential political capture. Additionally, there's a recommendation for governments to improve technical capacity and transparency when monitoring industrial policy performance.

Export as PDF
Here are some key quotes from the video that capture the main ideas and themes:

1. **Industrial Policy Trend:**
"Industrial policy is on the rise. Many countries have been implementing new industrial policy packages."

2. **Goals of Industrial Policy:**
"They try to boost innovation in specific sectors in the hope of reigniting productivity growth, but also addressing other goals like security issues or mitigating climate change."

3. **Examples of Initiatives:**
"To name a few examples, we have the CHIPS Act in the US or the K Chips Act in Korea, which promote research in semiconductors, but also the Green deal industrial plan in the European or IRA, which focus on green innovation."

4. **Resource Allocation Warning:**
"Governments should use these resources wisely in a way that maximizes productivity, growth, and social welfare."

5. **Quantitative Framework for Policymaking:**
"Our paper comes in to develop a quantitative framework that can guide governments as to when and how they should use industrial policy for innovation."

6. **Benefits of Targeting Innovation:**
"The framework should certainly capture the benefits of targeting innovation support to specific sectors as opposed to doing uniform support across the economy."

7. **Risks of Poor Implementation:**
"Poor implementation can quickly turn the potential gains of industrial policy into losses, making sector neutral support preferable."

8. **Importance of Clear Goals:**
"The goals of this policy need to be very clear, and well defined from the start."

9. **Economic Benefits of Targeted Support:**
"If it optimally targets innovation support to sectors with higher knowledge spillovers to other sectors, it can increase welfare by almost 3% relative to providing uniform support across sectors."

10. **Challenges of Political Capture:**
"A small amount of misallocation due to political capture can actually wipe out all the potential gains from industrial policy."

11. **Need for Government Capacity:**
"Government capacity must be high to avoid political capture."

12. **Recommendations for Policymakers:**
"Our recommendation is that they strengthen technical capacity in order to be able that they set clear benchmarks, so they can monitor progress and then discontinue any programs that don’t perform well."

13. **Call for Granular Research:**
"Our paper provides general guidelines for industrial policy for innovation across sectors and countries. But we think that more granular research would be helpful in a few areas."

These quotes highlight the key points discussed in the video, including the challenges and considerations of implementing industrial policy for innovation in various sectors.

X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:90000

1
00:00:00.000 --> 00:00:01.765
Industrial policy is on the rise.

2
00:00:02.235 --> 00:00:04.645
Many countries have been implementing new

3
00:00:04.645 --> 00:00:06.005
industrial policy packages.

4
00:00:06.545 --> 00:00:09.765
And one common element in those packages is that they try

5
00:00:09.765 --> 00:00:13.525
to boost innovation in specific sectors in the hope

6
00:00:13.545 --> 00:00:15.285
of reigniting productivity growth,

7
00:00:15.665 --> 00:00:18.805
but also addressing other goals like security issues

8
00:00:19.385 --> 00:00:20.885
or mitigating climate change.

9
00:00:21.595 --> 00:00:25.295
To name a few examples, we have the CHIPS Act in the US

10
00:00:25.755 --> 00:00:27.655
or the K Chips Act in Korea,

11
00:00:28.105 --> 00:00:30.455
which promote research in semiconductors,

12
00:00:31.235 --> 00:00:34.295
but also the Green deal industrial plan in the European

13
00:00:34.295 --> 00:00:37.815
Union, or the Inflation Reduction Act in the US

14
00:00:37.995 --> 00:00:41.415
or IRA, which focus on green innovation.

15
00:00:42.525 --> 00:00:44.865
And all these new policies also add

16
00:00:44.965 --> 00:00:48.545
to longstanding policies in emerging market economies like

17
00:00:48.545 --> 00:00:52.505
China, which, for a long time have been trying to,

18
00:00:52.535 --> 00:00:56.385
advance domestic technology in priority sectors like

19
00:00:56.905 --> 00:00:59.265
electric vehicles or artificial intelligence.

20
00:01:00.235 --> 00:01:03.335
Now, most of these packages include fiscal incentives

21
00:01:03.475 --> 00:01:04.775
for targeted innovation.

22
00:01:05.315 --> 00:01:08.855
Uh, this, this take the form of research and development

23
00:01:08.915 --> 00:01:12.655
or r and d, subsidies, but also r and d tax cuts

24
00:01:12.875 --> 00:01:14.815
or public r and d.

25
00:01:15.565 --> 00:01:17.385
And these policies are costly

26
00:01:17.605 --> 00:01:20.985
and they come at a time when many governments are already

27
00:01:20.985 --> 00:01:21.985
highly indebted.

28
00:01:22.565 --> 00:01:26.425
Um, so governments should use these resources wisely

29
00:01:26.965 --> 00:01:29.105
in a way that maximizes productivity,

30
00:01:29.105 --> 00:01:30.505
growth, and social welfare.

31
00:01:31.245 --> 00:01:34.925
And this is precisely where our paper comes in.

32
00:01:35.585 --> 00:01:37.405
We realize that it would be helpful

33
00:01:37.465 --> 00:01:39.285
to develop a quantitative framework

34
00:01:39.675 --> 00:01:42.765
that can guide governments as to when

35
00:01:43.065 --> 00:01:46.565
and how they should use industrial policy for innovation.

36
00:01:47.675 --> 00:01:50.365
This framework should certainly capture the benefits

37
00:01:50.425 --> 00:01:54.165
of targeting innovation support to specific SEC sectors

38
00:01:54.665 --> 00:01:57.805
as opposed to doing uniform support across the economy.

39
00:01:58.345 --> 00:02:00.325
Uh, so some sectors,

40
00:02:00.625 --> 00:02:03.325
may generate more knowledge spillovers to other sectors,

41
00:02:03.945 --> 00:02:06.525
or innovation in certain sectors can also lead

42
00:02:06.525 --> 00:02:08.565
to other social benefits like,

43
00:02:08.885 --> 00:02:12.245
reducing carbon emissions in the case of green innovation.

44
00:02:13.245 --> 00:02:15.825
But the framework should also capture the costs.

45
00:02:16.485 --> 00:02:18.905
And here we're not just talking about fiscal costs,

46
00:02:19.565 --> 00:02:21.825
but also the risk that if,

47
00:02:22.265 --> 00:02:24.105
policies are not implemented correctly,

48
00:02:24.735 --> 00:02:28.665
they can end up misdirecting support to the wrong sectors.

49
00:02:29.565 --> 00:02:31.625
And this could happen, either

50
00:02:31.625 --> 00:02:32.745
because, you know,

51
00:02:32.745 --> 00:02:34.745
when the government doesn't have enough information

52
00:02:35.275 --> 00:02:38.545
about the economy, about the nature of, social benefits

53
00:02:38.545 --> 00:02:40.985
of different sectors, or,

54
00:02:41.205 --> 00:02:44.105
in the case when certain sectors have more political power

55
00:02:44.445 --> 00:02:47.905
and are able to capture more resources from the government.

56
00:02:48.725 --> 00:02:51.585
And these are not just theoretical considerations.

57
00:02:51.845 --> 00:02:55.185
Uh, if we look at the history of industrial policies

58
00:02:55.185 --> 00:02:57.945
for innovation, even in advanced economies,

59
00:02:58.365 --> 00:03:02.005
we see many examples of policies that didn't deliver

60
00:03:02.505 --> 00:03:05.245
and, programs that had to be abandoned.

61
00:03:06.335 --> 00:03:10.235
So our framework allows us to quantify this balance of costs

62
00:03:10.255 --> 00:03:11.915
and benefits of industrial policies

63
00:03:11.915 --> 00:03:14.035
for innovation across different countries,

64
00:03:14.575 --> 00:03:16.475
but also to assess whether

65
00:03:17.155 --> 00:03:20.355
existing policies are targeting the right sectors

66
00:03:20.855 --> 00:03:22.995
and whether they're doing so in the right amount.

67
00:03:23.855 --> 00:03:25.995
The point of having industrial policies, which

68
00:03:26.615 --> 00:03:29.635
for us in our framework is mainly just different subsidies,

69
00:03:30.095 --> 00:03:32.605
for different sectors, is that

70
00:03:33.745 --> 00:03:36.995
different sectors can have different research spillovers

71
00:03:37.215 --> 00:03:38.235
to other sectors.

72
00:03:38.695 --> 00:03:41.075
But research spillovers, we mean just, you know,

73
00:03:41.075 --> 00:03:42.635
when someone comes up with a new idea

74
00:03:42.695 --> 00:03:44.715
or a new product, it's gonna make it easier

75
00:03:44.775 --> 00:03:46.915
for future researchers to,

76
00:03:47.125 --> 00:03:49.595
build their own new research projects

77
00:03:49.595 --> 00:03:51.715
because they can, of course, you know, build on top of,

78
00:03:51.815 --> 00:03:53.035
of what's being done before.

79
00:03:54.465 --> 00:03:57.755
When you put this in a sectoral perspective, it's

80
00:03:57.755 --> 00:03:59.515
very possible that some sectors

81
00:03:59.515 --> 00:04:00.835
provide more spillovers than others.

82
00:04:00.975 --> 00:04:03.035
And this is usually the case for basic science.

83
00:04:03.055 --> 00:04:05.435
For example, basic scientists tend

84
00:04:05.435 --> 00:04:08.555
to generate higher spillovers than in other sectors.

85
00:04:08.695 --> 00:04:10.355
You can think of the laser as an example,

86
00:04:10.355 --> 00:04:13.395
which really just started out, as research done

87
00:04:13.395 --> 00:04:16.115
by theoretical physicists that were trying

88
00:04:16.115 --> 00:04:17.635
to understand the properties

89
00:04:17.635 --> 00:04:19.835
of electromagnetic magnetic radiation.

90
00:04:20.575 --> 00:04:24.845
Um, and nowadays it has applications that go from,

91
00:04:25.305 --> 00:04:28.085
you know, fiber optic communication to wide surgery

92
00:04:28.265 --> 00:04:29.805
and basically everything else in between.

93
00:04:30.865 --> 00:04:34.795
Um, and then if we take a step back economically,

94
00:04:34.795 --> 00:04:37.195
what this means is that if we're trying

95
00:04:37.215 --> 00:04:40.355
to provide an optimal policy, this policy should then

96
00:04:40.865 --> 00:04:43.875
intuitively give more benefits to these sectors

97
00:04:43.875 --> 00:04:46.035
that create higher spillovers.

98
00:04:47.355 --> 00:04:49.655
Um, but then to take into account all

99
00:04:49.655 --> 00:04:50.775
of the spillovers created

100
00:04:50.915 --> 00:04:54.655
and received by every single sector in the economy, we sort

101
00:04:54.655 --> 00:04:56.935
of need a model that's going to,

102
00:04:56.935 --> 00:04:58.655
filter all this information in

103
00:04:58.995 --> 00:05:00.655
and then give us a measure of welfare

104
00:05:00.875 --> 00:05:03.295
or something like that, that we can understand

105
00:05:03.555 --> 00:05:05.815
and then rank different policies based on that.

106
00:05:06.355 --> 00:05:09.335
And once we have this, the advantage is

107
00:05:09.335 --> 00:05:11.215
that we can then also figure out

108
00:05:11.805 --> 00:05:15.655
what is the best configuration of subsidies across sectors

109
00:05:15.925 --> 00:05:17.655
that is going to give us the higher welfare,

110
00:05:18.275 --> 00:05:19.415
in the economy.

111
00:05:20.475 --> 00:05:24.455
So this is what was done in, in a recent paper by Ernest Lu

112
00:05:24.455 --> 00:05:28.975
and Song ma, which is a base of our own analysis,

113
00:05:30.365 --> 00:05:33.425
but there's a, another issue that comes into play,

114
00:05:33.425 --> 00:05:35.945
which is the implementation of these policies.

115
00:05:36.125 --> 00:05:37.745
So it's not uncommon for policies

116
00:05:37.745 --> 00:05:41.705
to face implementation issues, as they're trying to

117
00:05:41.725 --> 00:05:43.145
to be rolled out by governments.

118
00:05:43.205 --> 00:05:45.305
And this could come in the form,

119
00:05:46.695 --> 00:05:49.555
of having politically powerful sectors able to sort

120
00:05:49.555 --> 00:05:51.475
of gather more benefits to themselves

121
00:05:51.895 --> 00:05:54.955
or simply just random mistakes that policy makers might do,

122
00:05:55.255 --> 00:05:57.325
when they're trying to define,

123
00:05:57.615 --> 00:05:59.125
which sector gets what.

124
00:06:00.125 --> 00:06:04.145
Either way, both of these things are going to move the

125
00:06:04.985 --> 00:06:08.405
actual policy or the actual implemented policy away,

126
00:06:08.555 --> 00:06:09.685
from the optimal level

127
00:06:10.515 --> 00:06:13.255
and sort of reduce how efficient this policy is

128
00:06:13.275 --> 00:06:15.575
and actually generating more innovation.

129
00:06:16.235 --> 00:06:19.575
And then I think the last important point to, to make is

130
00:06:19.575 --> 00:06:23.055
that when you're defining which policy you're going

131
00:06:23.055 --> 00:06:26.255
to implement, the goals of this policy need

132
00:06:26.255 --> 00:06:28.935
to be very clear, and well defined from the start.

133
00:06:29.635 --> 00:06:31.855
Um, so for example, if you're just trying to design

134
00:06:32.925 --> 00:06:36.295
some policy that maximizes the amount of innovation

135
00:06:36.525 --> 00:06:37.615
that is done in the economy,

136
00:06:38.555 --> 00:06:41.175
you might not really care too much about

137
00:06:41.245 --> 00:06:44.095
what are the green sectors, what aren't the green sectors.

138
00:06:44.515 --> 00:06:46.295
On the other hand, if you want

139
00:06:46.295 --> 00:06:49.135
to actually implement more innovation in the greener sectors

140
00:06:49.135 --> 00:06:52.455
because of other reasons, environmental, for example, the

141
00:06:53.015 --> 00:06:55.255
optimal policy scheme changes in order

142
00:06:55.315 --> 00:06:57.535
to reflect this choice that you had at the beginning.

143
00:06:58.255 --> 00:07:02.005
And this is true for any other issue that you can think of.

144
00:07:02.345 --> 00:07:04.485
So we also show in the paper, for example,

145
00:07:04.865 --> 00:07:06.485
on our baseline optimal policy,

146
00:07:06.515 --> 00:07:07.885
just trying to maximize innovation.

147
00:07:08.975 --> 00:07:12.355
Um, the subsidy to each sector is going to be

148
00:07:12.905 --> 00:07:16.075
virtually uncorrelated with the exposure to AI that

149
00:07:16.075 --> 00:07:18.075
that sector has, which is something

150
00:07:18.075 --> 00:07:20.315
that we might care about, especially going to the future.

151
00:07:20.775 --> 00:07:24.395
And if we do, it needs to then be, implement

152
00:07:24.735 --> 00:07:26.035
or sort of taken into account

153
00:07:26.135 --> 00:07:28.275
as this policy is being defined in the first place.

154
00:07:28.775 --> 00:07:30.395
Our model simulations show

155
00:07:30.395 --> 00:07:33.275
that the large advanced economy think of the us

156
00:07:34.015 --> 00:07:37.595
if it optimally targets innovation support to sectors

157
00:07:37.665 --> 00:07:40.195
with higher knowledge spillovers to other sectors,

158
00:07:40.775 --> 00:07:44.675
it can increase welfare by almost 3% relative

159
00:07:44.675 --> 00:07:47.435
to providing uniform support across sectors.

160
00:07:48.015 --> 00:07:49.115
So this is equivalent

161
00:07:49.115 --> 00:07:51.995
to a permanent 3% increase in consumption,

162
00:07:52.805 --> 00:07:55.425
and these gains can go up to 6%.

163
00:07:55.885 --> 00:07:59.065
If in addition, the government also takes into account the

164
00:07:59.065 --> 00:08:00.425
goal to increase green innovation

165
00:08:01.205 --> 00:08:04.665
by reallocating more support to sectors with a higher share

166
00:08:04.925 --> 00:08:06.065
of green patents.

167
00:08:06.635 --> 00:08:09.325
However, and this is a very important caveat,

168
00:08:10.575 --> 00:08:13.865
poor implementation can quickly turn the potential gains

169
00:08:13.865 --> 00:08:16.105
of industrial policy into losses,

170
00:08:16.645 --> 00:08:19.225
making sector neutral support preferable.

171
00:08:19.885 --> 00:08:23.185
Uh, poor implementation can result either from political

172
00:08:23.185 --> 00:08:25.745
capture by certain sectors that are more influential

173
00:08:26.605 --> 00:08:28.985
or simply because the government doesn't have enough

174
00:08:28.985 --> 00:08:31.585
information either about the social benefits

175
00:08:31.585 --> 00:08:35.145
of cer certain sectors or about, firm's activities.

176
00:08:35.605 --> 00:08:37.905
So let me discuss these, these two possibilities.

177
00:08:37.965 --> 00:08:40.705
In turn, regarding political capture,

178
00:08:41.205 --> 00:08:44.305
we will use two alternative methods, to try

179
00:08:44.305 --> 00:08:46.785
to measure the political influence of sector.

180
00:08:47.645 --> 00:08:50.835
First we look at the markup in sector,

181
00:08:51.655 --> 00:08:55.525
and the idea which is based on previous literature is

182
00:08:55.525 --> 00:08:58.525
that sectors with, larger markets, sorry, firms

183
00:08:58.555 --> 00:09:00.005
with larger market shares

184
00:09:00.385 --> 00:09:04.045
or higher markups also tend to employ more politicians.

185
00:09:04.745 --> 00:09:08.565
And firms that employ more politicians are also able to,

186
00:09:08.565 --> 00:09:11.005
influence the allocation of government support

187
00:09:11.385 --> 00:09:15.205
and capture more, more benefits from the government.

188
00:09:16.285 --> 00:09:19.945
And second, as an alternative, we also look at the share

189
00:09:20.165 --> 00:09:24.425
of, lobbying expenses by each sector in the us,

190
00:09:24.445 --> 00:09:28.305
in order to gauge how much each sector is investing in,

191
00:09:28.325 --> 00:09:29.785
in building political influence,

192
00:09:30.685 --> 00:09:32.985
but actually under either methods,

193
00:09:33.325 --> 00:09:34.705
the results are quite similar.

194
00:09:35.285 --> 00:09:38.785
And the result is that a small amount of misallocation due

195
00:09:38.785 --> 00:09:42.105
to political capture can actually wipe out,

196
00:09:42.165 --> 00:09:44.465
all the potential gains from industrial policy.

197
00:09:45.465 --> 00:09:47.285
Now, turning to the other case,

198
00:09:47.335 --> 00:09:50.405
which is when the government makes, random mistakes,

199
00:09:50.715 --> 00:09:53.645
when allocating innovation subsidies,

200
00:09:53.675 --> 00:09:55.485
because it doesn't have enough information.

201
00:09:56.385 --> 00:09:59.605
The result here, somewhat surprisingly, is

202
00:09:59.605 --> 00:10:04.365
that random mistakes are as harmful as a political capture

203
00:10:04.745 --> 00:10:07.245
for social welfare, if not worse.

204
00:10:07.705 --> 00:10:12.125
Uh, because in the case of political capture, what we see is

205
00:10:12.125 --> 00:10:15.245
that sectors that generate more knowledge spillovers

206
00:10:15.245 --> 00:10:18.765
to other sectors are also sectors that tend to have,

207
00:10:18.765 --> 00:10:21.085
high markups and, and more political power.

208
00:10:21.825 --> 00:10:25.485
So in a way, when, the allocation of resources is driven

209
00:10:25.545 --> 00:10:28.965
by political influence, it may not be as bad

210
00:10:29.385 --> 00:10:32.045
as when it's simply a result of random mistakes

211
00:10:32.705 --> 00:10:33.845
due to lack of information.

212
00:10:34.065 --> 00:10:36.325
To answer this question, it's useful to just, you know,

213
00:10:36.355 --> 00:10:38.365
step a step back a little bit

214
00:10:38.905 --> 00:10:42.005
and, go back to what is the reason

215
00:10:42.025 --> 00:10:44.805
to have industrial policies in our framework to begin with.

216
00:10:45.425 --> 00:10:47.805
Um, and the reason is that you have sectors

217
00:10:47.805 --> 00:10:50.005
that create more spillovers to other sectors.

218
00:10:50.395 --> 00:10:54.045
Well, the point is that not all spillovers come from

219
00:10:54.765 --> 00:10:56.365
research that is done in your own country.

220
00:10:57.265 --> 00:10:59.655
So, and we can actually measure this

221
00:10:59.655 --> 00:11:03.575
by looking at patent citations, across, countries.

222
00:11:03.635 --> 00:11:06.735
So parents work a lot like academic papers

223
00:11:07.235 --> 00:11:09.695
or books in the sense that they will cite all

224
00:11:09.695 --> 00:11:12.015
of the prior research that's relevant to them.

225
00:11:12.955 --> 00:11:16.135
And because we can link all the patents to their authors

226
00:11:16.135 --> 00:11:17.775
and their authors to their location,

227
00:11:18.825 --> 00:11:20.245
we can actually create this network

228
00:11:20.745 --> 00:11:24.805
of patent citations across countries, sector by sector.

229
00:11:25.345 --> 00:11:29.685
And this also tells us things like, about 30%

230
00:11:30.145 --> 00:11:33.525
of all the citations from patents done in the US

231
00:11:34.345 --> 00:11:36.925
go from research that was done in other countries.

232
00:11:37.385 --> 00:11:39.045
And this is by the way, the largest,

233
00:11:39.545 --> 00:11:40.765
share that we get in the data.

234
00:11:40.765 --> 00:11:44.685
So the typical European country, for example, will have 90%

235
00:11:44.685 --> 00:11:48.885
of those citations go to research done in other countries.

236
00:11:50.445 --> 00:11:52.535
Then this creates a bit of a wrinkle for,

237
00:11:52.675 --> 00:11:54.045
for the policymaking process,

238
00:11:54.515 --> 00:11:57.285
because industrial policies, at least in the sense

239
00:11:57.285 --> 00:12:00.565
that we define them as these sub sector specific RD

240
00:12:00.565 --> 00:12:02.045
subsidies, will mostly

241
00:12:02.905 --> 00:12:06.725
affect the domestic output research output or inputs

242
00:12:06.725 --> 00:12:08.045
or, you know, however you want to define them.

243
00:12:08.825 --> 00:12:13.405
But the point is that, if the bulk,

244
00:12:14.305 --> 00:12:15.885
of the spillovers

245
00:12:15.885 --> 00:12:18.565
or the citations for scientists in say,

246
00:12:18.565 --> 00:12:22.525
country A are actually coming from other research done in

247
00:12:22.765 --> 00:12:25.085
countries B, C, or D,

248
00:12:25.155 --> 00:12:28.485
then the gains from industrial policy in country A are going

249
00:12:28.485 --> 00:12:31.685
to be greatly reduced, essentially because, the r

250
00:12:31.685 --> 00:12:33.325
and d done in this country A,

251
00:12:33.985 --> 00:12:36.885
is not really affecting the bulk of the spillovers received

252
00:12:36.905 --> 00:12:40.245
by the companies or researchers in country a as well.

253
00:12:41.285 --> 00:12:43.975
Empirically, we can actually compare the results

254
00:12:43.975 --> 00:12:48.015
of the actual policies, in several advanced economies,

255
00:12:48.435 --> 00:12:51.095
across the world with what the model says.

256
00:12:51.365 --> 00:12:55.295
It's the optimal, allocation of r and d resources.

257
00:12:55.715 --> 00:12:58.415
And if we do that, we do find a positive correlation,

258
00:12:58.685 --> 00:13:03.565
meaning that, policymakers are somewhat able to identify

259
00:13:04.375 --> 00:13:06.805
which are the sectors that provide the highest spillovers

260
00:13:06.805 --> 00:13:08.205
for the economy,

261
00:13:08.385 --> 00:13:10.525
and subsidize these sectors at higher rates.

262
00:13:11.615 --> 00:13:13.875
But we also find that the concentration

263
00:13:13.875 --> 00:13:17.995
of these resources in the data tends to be much higher than

264
00:13:17.995 --> 00:13:19.115
what the model will suggest.

265
00:13:19.895 --> 00:13:23.555
Um, meaning that scaling back of some

266
00:13:23.555 --> 00:13:26.155
of these policies might be beneficial from a lot

267
00:13:26.355 --> 00:13:29.355
of countries, that, that we look at in the sense

268
00:13:29.355 --> 00:13:32.035
that they're over subsidizing a few of those sectors

269
00:13:32.035 --> 00:13:33.835
that they think are going to be,

270
00:13:34.295 --> 00:13:35.795
the key sectors in the future.

271
00:13:36.155 --> 00:13:38.555
O overall the results of our paper such as

272
00:13:38.555 --> 00:13:41.595
that governments should be cautious, when it comes

273
00:13:41.595 --> 00:13:43.195
to industrial policies for innovation,

274
00:13:43.585 --> 00:13:46.795
because those only beneficial under fairly

275
00:13:46.905 --> 00:13:48.235
restrictive conditions.

276
00:13:48.895 --> 00:13:51.475
Uh, we identify three such conditions.

277
00:13:52.005 --> 00:13:53.595
First, that externalities

278
00:13:53.735 --> 00:13:57.715
or social benefits of targeted sectors must be easily

279
00:13:58.475 --> 00:13:59.715
identified and measurable.

280
00:14:00.255 --> 00:14:02.675
And reducing carbon emissions is one example

281
00:14:02.775 --> 00:14:04.155
of when that is possible.

282
00:14:04.895 --> 00:14:07.795
Second, the domestic knowledge spillovers

283
00:14:07.975 --> 00:14:10.315
of targeted sectors must be strong enough.

284
00:14:11.055 --> 00:14:14.115
And third, government capacity must be high

285
00:14:14.455 --> 00:14:17.195
to prevent misallocation, for example, due

286
00:14:17.195 --> 00:14:20.995
to political capture as, my coauthor Alex Explains.

287
00:14:21.455 --> 00:14:24.355
Um, we do find that, countries,

288
00:14:24.895 --> 00:14:28.555
or leading economies are actually targeting broadly the

289
00:14:28.555 --> 00:14:30.755
right sectors that may be,

290
00:14:30.775 --> 00:14:32.715
to an excessive degree in some cases.

291
00:14:33.495 --> 00:14:35.835
And so they should pair back the size

292
00:14:35.835 --> 00:14:37.075
of their industrial policies.

293
00:14:37.925 --> 00:14:39.545
But for governments that do want

294
00:14:39.545 --> 00:14:43.425
to implement industrial policies, our recommendation is

295
00:14:43.495 --> 00:14:46.825
that they strengthen technical capacity in order to be able

296
00:14:46.825 --> 00:14:48.705
to vet the projects they subsidize,

297
00:14:49.375 --> 00:14:52.525
that they set clear benchmarks, so they can monitor

298
00:14:53.045 --> 00:14:56.165
progress and then discontinue any programs that

299
00:14:56.195 --> 00:14:59.045
that don't perform well as opposed to letting sort

300
00:14:59.045 --> 00:15:01.525
of policy, policy, inertia take over.

301
00:15:02.355 --> 00:15:05.255
And governments should also conduct an exhaustive assessment

302
00:15:05.275 --> 00:15:09.055
of fiscal costs and risks to avoid fiscal surprises.

303
00:15:10.055 --> 00:15:12.475
And more broadly, they should make sure

304
00:15:12.475 --> 00:15:15.955
that programs are accessible across all firms in order

305
00:15:16.015 --> 00:15:17.435
to foster competition.

306
00:15:18.175 --> 00:15:20.915
Now, thinking about future research,

307
00:15:21.175 --> 00:15:24.755
our paper provides general guidelines for industrial policy

308
00:15:24.755 --> 00:15:27.755
for innovation across sectors and countries.

309
00:15:28.295 --> 00:15:31.115
But we think that more granular research would be

310
00:15:31.115 --> 00:15:32.155
helpful in a few areas.

311
00:15:33.035 --> 00:15:35.855
For example, governments are not always

312
00:15:35.875 --> 00:15:38.335
so transparent when they report the size

313
00:15:38.335 --> 00:15:39.655
of their industrial policies.

314
00:15:39.995 --> 00:15:44.095
So more empirical work, is needed here to, to quantify

315
00:15:44.095 --> 00:15:45.295
that across countries.

316
00:15:46.385 --> 00:15:49.755
Also, our paper focused on innovation subsidies,

317
00:15:50.175 --> 00:15:52.995
but industrial policies include other instruments

318
00:15:53.145 --> 00:15:56.955
that could also affect innovation, like trade barriers,

319
00:15:57.475 --> 00:15:59.475
regulatory advantages for certain sectors

320
00:16:00.055 --> 00:16:02.035
or, even directed allocation

321
00:16:02.175 --> 00:16:03.755
of credit in some countries.

322
00:16:04.895 --> 00:16:07.475
But also they, they, include policies

323
00:16:07.475 --> 00:16:09.475
that benefit certain firms,

324
00:16:09.695 --> 00:16:11.835
in a given sector more than others, for example,

325
00:16:11.865 --> 00:16:15.555
when governments try to create industrial champions.

326
00:16:15.935 --> 00:16:17.355
So looking at, at the effects

327
00:16:17.355 --> 00:16:18.955
of those policies would be interesting.

328
00:16:19.455 --> 00:16:22.035
And finally, I think in, in our current context

329
00:16:22.135 --> 00:16:24.075
of geo-economic fragmentation,

330
00:16:24.695 --> 00:16:27.275
it would also be very important to try to understand

331
00:16:27.875 --> 00:16:29.835
strategic interaction between countries.

332
00:16:30.415 --> 00:16:33.355
So strategic considerations mean that a country

333
00:16:34.055 --> 00:16:38.235
can take certain policy actions that do not benefit it,

334
00:16:38.305 --> 00:16:42.155
immediately, but that are taken in the hope

335
00:16:42.175 --> 00:16:45.715
of influencing the behavior of other competitor countries.

336
00:16:46.495 --> 00:16:49.675
Now, of course, we think that this is a slippery slope,

337
00:16:49.935 --> 00:16:53.195
not only because it can lead to retaliation, but also

338
00:16:53.195 --> 00:16:54.595
because as we mentioned,

339
00:16:55.015 --> 00:16:58.035
all countries rely on innovation don abroad.

340
00:16:58.575 --> 00:17:00.475
But still, it is very important

341
00:17:00.495 --> 00:17:03.395
to understand the incentives of countries for doing so.

342
00:17:04.015 --> 00:17:07.755
If we are to prevent, geo-economic fragmentation

343
00:17:08.025 --> 00:17:11.635
that can end up reducing the world's innovative capacity

344
00:17:11.975 --> 00:17:12.715
and productivity.

Rising Industrial Policies: Fostering Innovation and Addressing Global Challenges

00:00:00 - Industrial policy is on the rise. Many countries have been implementing new industrial policy packages. And one common element in those packages is that they try to boost innovation in specific sectors in the hope of reigniting productivity growth, but also addressing other goals like security issues or mitigating climate change. To name a few examples, we have the CHIPS Act in the US or the K Chips Act in Korea, which promote research in semiconductors, but also the Green deal industrial plan in the European

00:00:37 - or IRA, which focus on green innovation. And all these new policies also add to longstanding policies in emerging market economies like China, which, for a long time have been trying to, advance domestic technology in priority sectors like electric vehicles or artificial intelligence. Now, most of these packages include fiscal incentives for targeted innovation.

00:01:08 - or r and d, subsidies, but also r and d tax cuts or public r and d. And these policies are costly and they come at a time when many governments are already highly indebted. so governments should use these resources wisely in a way that maximizes productivity, growth, and social welfare. And this is precisely where our paper comes in.

Guiding Industrial Policy for Innovation: Benefits, Costs, and Risks

00:01:37 - to develop a quantitative framework that can guide governments as to when and how they should use industrial policy for innovation. This framework should certainly capture the benefits of targeting innovation support to specific SEC sectors as opposed to doing uniform support across the economy. so some sectors, may generate more knowledge spillovers to other sectors, or innovation in certain sectors can also lead

00:02:08 - reducing carbon emissions in the case of green innovation. But the framework should also capture the costs. And here we're not just talking about fiscal costs, but also the risk that if, policies are not implemented correctly, they can end up misdirecting support to the wrong sectors. And this could happen, either because, you know, when the government doesn't have enough information about the economy, about the nature of, social benefits

00:02:41 - in the case when certain sectors have more political power and are able to capture more resources from the government. And these are not just theoretical considerations. if we look at the history of industrial policies for innovation, even in advanced economies, we see many examples of policies that didn't deliver and, programs that had to be abandoned. So our framework allows us to quantify this balance of costs

Enhancing Innovation Through Targeted Industrial Policies

00:03:11 - for innovation across different countries, but also to assess whether existing policies are targeting the right sectors and whether they're doing so in the right amount. The point of having industrial policies, which for us in our framework is mainly just different subsidies, for different sectors, is that different sectors can have different research spillovers to other sectors. But research spillovers, we mean just, you know,

00:03:42 - or a new product, it's gonna make it easier for future researchers to, build their own new research projects because they can, of course, you know, build on top of, of what's being done before. When you put this in a sectoral perspective, it's very possible that some sectors provide more spillovers than others. And this is usually the case for basic science. For example, basic scientists tend to generate higher spillovers than in other sectors. You can think of the laser as an example, which really just started out, as research done

00:04:16 - to understand the properties of electromagnetic magnetic radiation. and nowadays it has applications that go from, you know, fiber optic communication to wide surgery and basically everything else in between. and then if we take a step back economically, what this means is that if we're trying to provide an optimal policy, this policy should then intuitively give more benefits to these sectors

Modeling Spillovers for Effective Policy Implementation

00:04:47 - but then to take into account all of the spillovers created and received by every single sector in the economy, we sort of need a model that's going to, filter all this information in and then give us a measure of welfare or something like that, that we can understand and then rank different policies based on that. And once we have this, the advantage is that we can then also figure out what is the best configuration of subsidies across sectors

00:05:18 - in the economy. So this is what was done in, in a recent paper by Ernest Lu and Song ma, which is a base of our own analysis, but there's a, another issue that comes into play, which is the implementation of these policies. So it's not uncommon for policies to face implementation issues, as they're trying to to be rolled out by governments. And this could come in the form,

00:05:49 - of gather more benefits to themselves or simply just random mistakes that policy makers might do, when they're trying to define, which sector gets what. Either way, both of these things are going to move the actual policy or the actual implemented policy away, from the optimal level and sort of reduce how efficient this policy is and actually generating more innovation. And then I think the last important point to, to make is

Defining Clear Objectives for Effective Innovation Policies

00:06:23 - to implement, the goals of this policy need to be very clear, and well defined from the start. so for example, if you're just trying to design some policy that maximizes the amount of innovation that is done in the economy, you might not really care too much about what are the green sectors, what aren't the green sectors. On the other hand, if you want to actually implement more innovation in the greener sectors because of other reasons, environmental, for example, the

00:06:55 - to reflect this choice that you had at the beginning. And this is true for any other issue that you can think of. So we also show in the paper, for example, on our baseline optimal policy, just trying to maximize innovation. the subsidy to each sector is going to be virtually uncorrelated with the exposure to AI that that sector has, which is something that we might care about, especially going to the future. And if we do, it needs to then be, implement

00:07:26 - as this policy is being defined in the first place. Our model simulations show that the large advanced economy think of the us if it optimally targets innovation support to sectors with higher knowledge spillovers to other sectors, it can increase welfare by almost 3% relative to providing uniform support across sectors. So this is equivalent to a permanent 3% increase in consumption, and these gains can go up to 6%.

Challenges in Implementing Industrial Policy for Green Innovation

00:07:59 - goal to increase green innovation by reallocating more support to sectors with a higher share of green patents. However, and this is a very important caveat, poor implementation can quickly turn the potential gains of industrial policy into losses, making sector neutral support preferable. poor implementation can result either from political capture by certain sectors that are more influential

00:08:28 - information either about the social benefits of cer certain sectors or about, firm's activities. So let me discuss these, these two possibilities. In turn, regarding political capture, we will use two alternative methods, to try to measure the political influence of sector. First we look at the markup in sector, and the idea which is based on previous literature is that sectors with, larger markets, sorry, firms

00:09:00 - or higher markups also tend to employ more politicians. And firms that employ more politicians are also able to, influence the allocation of government support and capture more, more benefits from the government. And second, as an alternative, we also look at the share of, lobbying expenses by each sector in the us, in order to gauge how much each sector is investing in, in building political influence,

Impact of Political Capture and Mistakes on Industrial Policy Effectiveness

00:09:33 - the results are quite similar. And the result is that a small amount of misallocation due to political capture can actually wipe out, all the potential gains from industrial policy. Now, turning to the other case, which is when the government makes, random mistakes, when allocating innovation subsidies, because it doesn't have enough information. The result here, somewhat surprisingly, is that random mistakes are as harmful as a political capture

00:10:07 - because in the case of political capture, what we see is that sectors that generate more knowledge spillovers to other sectors are also sectors that tend to have, high markups and, and more political power. So in a way, when, the allocation of resources is driven by political influence, it may not be as bad as when it's simply a result of random mistakes due to lack of information. To answer this question, it's useful to just, you know,

00:10:38 - and, go back to what is the reason to have industrial policies in our framework to begin with. and the reason is that you have sectors that create more spillovers to other sectors. Well, the point is that not all spillovers come from research that is done in your own country. So, and we can actually measure this by looking at patent citations, across, countries. So parents work a lot like academic papers

Global Patent Citations and Industrial Policy Impacts

00:11:09 - of the prior research that's relevant to them. And because we can link all the patents to their authors and their authors to their location, we can actually create this network of patent citations across countries, sector by sector. And this also tells us things like, about 30% of all the citations from patents done in the US go from research that was done in other countries.

00:11:39 - share that we get in the data. So the typical European country, for example, will have 90% of those citations go to research done in other countries. Then this creates a bit of a wrinkle for, for the policymaking process, because industrial policies, at least in the sense that we define them as these sub sector specific RD subsidies, will mostly affect the domestic output research output or inputs or, you know, however you want to define them.

00:12:14 - of the spillovers or the citations for scientists in say, country A are actually coming from other research done in countries B, C, or D, then the gains from industrial policy in country A are going to be greatly reduced, essentially because, the r and d done in this country A, is not really affecting the bulk of the spillovers received by the companies or researchers in country a as well.

Guidelines for Effective Industrial Policy and R&D Allocation

00:12:43 - of the actual policies, in several advanced economies, across the world with what the model says. It's the optimal, allocation of r and d resources. And if we do that, we do find a positive correlation, meaning that, policymakers are somewhat able to identify which are the sectors that provide the highest spillovers for the economy, and subsidize these sectors at higher rates.

00:13:13 - of these resources in the data tends to be much higher than what the model will suggest. meaning that scaling back of some of these policies might be beneficial from a lot of countries, that, that we look at in the sense that they're over subsidizing a few of those sectors that they think are going to be, the key sectors in the future. O overall the results of our paper such as that governments should be cautious, when it comes to industrial policies for innovation,

00:13:46 - restrictive conditions. we identify three such conditions. First, that externalities or social benefits of targeted sectors must be easily identified and measurable. And reducing carbon emissions is one example of when that is possible. Second, the domestic knowledge spillovers of targeted sectors must be strong enough. And third, government capacity must be high

Guidelines for Effective Industrial Policy Implementation

00:14:17 - to political capture as, my coauthor Alex Explains. we do find that, countries, or leading economies are actually targeting broadly the right sectors that may be, to an excessive degree in some cases. And so they should pair back the size of their industrial policies. But for governments that do want to implement industrial policies, our recommendation is that they strengthen technical capacity in order to be able

00:14:49 - that they set clear benchmarks, so they can monitor progress and then discontinue any programs that that don't perform well as opposed to letting sort of policy, policy, inertia take over. And governments should also conduct an exhaustive assessment of fiscal costs and risks to avoid fiscal surprises. And more broadly, they should make sure that programs are accessible across all firms in order to foster competition.

00:15:21 - our paper provides general guidelines for industrial policy for innovation across sectors and countries. But we think that more granular research would be helpful in a few areas. For example, governments are not always so transparent when they report the size of their industrial policies. So more empirical work, is needed here to, to quantify that across countries. Also, our paper focused on innovation subsidies,

Impact of Trade Policies on Innovation and Geoeconomic Relations

00:15:53 - that could also affect innovation, like trade barriers, regulatory advantages for certain sectors or, even directed allocation of credit in some countries. But also they, they, include policies that benefit certain firms, in a given sector more than others, for example, when governments try to create industrial champions. So looking at, at the effects of those policies would be interesting. And finally, I think in, in our current context

00:16:24 - it would also be very important to try to understand strategic interaction between countries. So strategic considerations mean that a country can take certain policy actions that do not benefit it, immediately, but that are taken in the hope of influencing the behavior of other competitor countries. Now, of course, we think that this is a slippery slope, not only because it can lead to retaliation, but also

00:17:11 - all countries rely on innovation don abroad. But still, it is very important to understand the incentives of countries for doing so. If we are to prevent, geo-economic fragmentation that can end up reducing the world's innovative capacity and productivity.

Faculti AI can make mistakes

Research

Policy

Analysis