Middle Technology Trap

Economics for the
Common Good
EU
INNOVATION
POLICY
HOW TO ESCAPE
THE MIDDLE
TECHNOLOGY
TRAP
A Report by the European

Policy Analysis Group
C. Fuest, D. Gros,
P.-L. Mengel,
G. Presidente and J. Tirole
Economics for the
Common Good
EU
INNOVATION
POLICY
HOW TO ESCAPE
THE MIDDLE
TECHNOLOGY
TRAP
ABSTRACT
The EU is losing the global innovation race. EU industry invests less than its peers in R&D, it lags
way behind in software, hardware and artificial intelligence, and its pharmaceutical component is
at risk. For over 20 years the same companies, mostly from the automotive sector, have dominated
EU innovation activity. We call this the middle technology trap.
Existing EU programmes to foster innovation, including those under the heading of the European
Innovation Council (EIC), are far from the gold standard – the US Advanced Research Projects
Agency (ARPA) model. They rarely finance disruptive innovation and instead bring funds to
rather mature entities. Their decision processes are still very political, they impose collaborations
instead of accompanying them, they devote too much of their limited resources to venture capital
investment rather than to supporting breakthrough innovation, and the few project managers are
over-stretched.
We propose an ARPA-style model of governance and a budget-neutral shift of resources to

support high-risk, high-return projects that are far from commercial application. Project selection
and management should be improved by increasing the scientific and engineering excellence
of the EIC Board and by delegating more to scientists. The current venture capital activities, if
maintained, should be outsourced to a specialised fund.
ACKNOWLEDGEMENTS
We wish to thank, without implicating them in our conclusions, Philippe

Aghion, Thomas Ebbesen, Christophe Hemous, Francesco Matteucci,
Charles-Henri Weymuller, as well as the members of the advisory group,
Ufuk Akcigit, Mario Monti, Monika Schnitzer, and John Van Reenen. We
thank participants of the April 10, 2024 presentation of the report in
Brussels, in particular Reinhilde Veugelers, as well as participants of the IEP
brainstorming workshop at Bocconi University, in particular Michele Polo,
which provided invaluable advice and suggestions. We also thank Anita
Dietrich, Florian Dorn, Silvan Hofer and Marta Zava for providing data and
additional material included in the report.
RECOMMENDED CITATION
Through its missions and governance, Horizon Europe does not meet
the innovation challenge and anchors our industry in the mid-tech range.
This report argues that current European efforts, while laudable, are
insufficient, in both quantity and quality. Important reforms are required
to enable Europe to compete in the value-creating space.
The three participating institutions do not take an institutional position. The opinions expressed in
this publication are those of the authors. Any inaccuracies or oversights are the sole responsibility
of the authors.
@2024 econpol@cesifo, IEP@BU Institute for European Policymaking@Bocconi University, and

Toulouse School of Economics - All rights reserved.
TABLE OF
CONTENTS
EXECUTIVE SUMMARY 4
I. INNOVATION SPENDING COMPARED:

DOES THE EU HAVE AN INNOVATION PROBLEM? 6
The middle technology trap 7
Patents as a measure of innovation output 11
Path dependency 13
II. CAN THE EUROPEAN INNOVATION COUNCIL

HELP EUROPE ESCAPE THE MID-TECH TRAP? 16
The ARPA model 17
Governance of EU Innovation Policy (EIC) 20
Budget 21
Who receives support from EIC programmes? 22
Role of programme managers 23
Eligibility requirements 24
Funding delays 25
III. ANATOMY OF HORIZON EUROPE 26

Pillar I 28
Pillar II 29
Pillar III 31
European Institute of Innovation & Technology 31
European Innovation Ecosystem 32
EIC Accelerator and the EIC Fund 32
Innovation Fund 34
IV. A BUDGET-NEUTRAL POLICY PROPOSAL

TO BOOST EUROPEAN DEEP-TECH 35
Reallocation towards disruptive innovation 36
V. SUMMARY AND CONCLUSIONS 39
References 41
Appendix 44
Governance of European research/innovation bodies 50
EU INNOVATION POLICY
EXECUTIVE
SUMMARY
Encouraging innovation has long been a priority for European policymakers, with the implicit aim
of reaching the technology frontier represented by the US. This goal has not been achieved. The
Innovation Scoreboards regularly published by the European Commission have consistently found
that the EU lags behind the US on many indicators. The most recent Scoreboards indicates that the
transatlantic gap has widened.
This report argues that current European efforts, while laudable, are insufficient, in both quantity
and quality. Important reforms are required to enable Europe to compete in the value-creating
space.
The disappointing European performance might surprise the reader, given that government support
for research and development (R&D) has grown gradually over the last two decades and is now at
about the same level as in the US (around 0.7% of GDP). It is in the private sector where one finds a
large quantitative difference between the US and EU. Business expenditure on R&D (BERD) in the
EU, at 1.2% of GDP, represents about half that of the US (2.3% of GDP).
Moreover, European business R&D is concentrated in mid-tech sectors, like the automotive industry.
These sectors compete by applying the latest technological advances to production, but they do
not require the same R&D intensity or offer the same growth potential as high-tech industries that
produce the newest technologies. The main reason for US private R&D being twice that of Europe
is therefore the much higher weight of high-tech industries in the US.
European specialisation in mid-tech, the ‘middle technology trap’, has persisted for two decades.
The largest EU companies in terms of R&D expenditures are almost invariably car producers, whereas
in the US car producers, which were important 20 years ago, have been supplanted by software
companies. The EU’s comparative advantage in cars is worrisome, as despite its massive investment
in R&D, the EU automotive industry now risks being leapfrogged by US producers and increasingly
by Chinese ones. Foreign producers can build on their leadership in electric and autonomous-driving
technologies.
4
Today, the transatlantic gap is particularly wide in software development, where US companies
account for 75% of the global total, compared with 6% for the EU (less than China). A large portion
of the growth in US corporate R&D spending over the last decade has come from software-related
companies, underpinning US dominance in the latest advances in artificial intelligence (AI). The quasi-
monopoly position of US high-tech sector also applies to next generation of upcoming software
(most cutting edge LLMs are US) and hardware (see Nvidia for semi-conductors etc). And China is
rapidly catching up in terms of high-tech R&D spending.
A shift in the direction of EU innovation towards high-tech industries thus appears highly desirable.
Escaping the middle technology trap would foster growth and increase the geopolitical weight of
the EU. But to achieve this, governance must match ambitions. Simply put, the EU does not have the
institutions it takes to meet the 21st century innovation challenge.
The EU contributes through the Horizon Europe1 programme, which earmarks about €11-12 billion
per year to support broadly-defined innovation, research and development.
However, less than 5% of Horizon Europe supports breakthrough innovation, which has the potential
to create new markets but is remote from commercial applications. Distinguishing between types of
R&D is important (yet overlooked), as projects aimed at bringing known technologies closer to the
market cannot be expected to deliver disruptive innovation.
The recent creation of the European Innovation Council (EIC) was a positive step towards redirecting
R&D efforts, but it is hampered by several limitations. First, is too dependent on the European
Commission. Second, it only marginally targets breakthrough research, which is still substantially
underfunded
1 Horizon Europe is the name of the 9 th Framework Programme for Research and Technological Development (2021-2027).
5
EXECUTIVE
SUMMARY
The EIC seems more focused on remedying perceived capital market imperfections than on
promoting innovation, as a substantial share of its spending supports the capital structure of
small to medium-sized enterprises (SMEs) and, to a more limited extent, startups. Moreover,
serious governance issues may undermine its mission of boosting breakthrough innovation: the
EIC is mostly led by EU officials rather than top scientists; some eligibility criteria impose severe
constraints, rendering the selection mechanisms highly bureaucratic; collaborations are mandated
rather than accompanied; and the disbursement of funding is slow.
Institutional change is thus needed to boost the development of EU high-tech industries. Specifically,
we propose:
(i) giving leading scientists a more central role on the EIC Board and in selecting projects;
(ii) shifting decision-making power from the European Commission to a larger number of
independent project managers; and
(iii) drawing resources from underperforming programmes of Horizon Europe and other parts of
the EU innovation ecosystem to expand the size and scope of programmes actually devoted
to breakthrough research, without changing the existing Multiannual Financial Framework.
Our budget-neutral, yet radical reform of the EIC could give the EU an innovation engine along the
lines of the US ARPAs (advanced research project agencies). ARPAs have been widely credited with
supporting advances in several breakthrough technologies and the development of the American
biotechnology, software and hardware industries. A flourishing European innovation ecosystem
would create the right incentives, crowd in the missing private investment and stimulate the growth
of high-tech industries, helping the EU to escape its middle technology trap.
This report is concerned with the allocation and management of European funds aimed at filling
our R&D gap. Before getting to the heart of the matter, let us say what we do not do.
6
■ First, we focus on European-level financing. Unlike in the US, much of the R&D support (actually
nine tenths) is brought by Member states. Let us content ourselves here with noting that there
are good reasons why the subsidiarity argument does not apply to our context. Moving part
of the funding to the EU level might bring a number of benefits. It would enlarge the pool
of projects to choose from, and the pool of independent experts who can give an informed
opinion on these projects. It would increase the (physical, cultural, social) distance between
principal investigators and reviewers, conferring more integrity on the review process. Finally,
and specifically for the case of disruptive innovation, the fact that ambitious R&D is bound to
fail with non-negligible probability means that it is hard to conduct advanced research without
risking a mediatic backlash if very few projects succeed. The law of large numbers at the EU level
makes this occurrence less likely.
■ Second, we are looking only at a small part of the picture – this is both a strength and a
limitation of this report. Solving the EU innovation problem will take much more than changing
the governance and mission of the EIC. The European innovation ecosystem is hampered by
multiple other problems: the inequation of European labour market regulations to the start-
up environment, the insufficient level of universities, the absence of entrepreneurship courses
in many of them, the limitations of the single market (from the portability of pensions to the
complexity of 27 labour laws, the imperfect capital market and digital unions, the segmentation
of procurement practices, etc.).
7
I.
INNOVATION SPENDING
COMPARED:
DOES THE EU HAVE
AN INNOVATION PROBLEM?
8
E uropean leaders perceive that Europe has an innovation problem. An
expenditure on R&D of 3% of GDP has been an official goal of the EU
since the launch the Lisbon Strategy in 2000. However, gross domestic
expenditure on R&D in the EU is still roughly 2% of GDP, lower than in other major
economies such as the US, Japan and China.
The reason why the EU lags behind other regions is not that governments (national
and EU) spend less on R&D than its rivals. In 2020, government-funded R&D
amounted to €110 billion in the EU (mostly by national governments2) and €150
billion in the US, accounting for a very similar percentage of GDP, around 0.7%.3 In
the other regions of the world, government R&D spending is somewhat lower at
0.5% of GDP.
The key reason for the overall transatlantic difference is the lower engagement in
R&D by the business sector, whose spending amounts to only 1.2% of GDP in the
EU, versus 2.3% of GDP in the US.
These oft-cited OECD figures, however, do not allow us to decompose them
among the different sectors. To analyse in more detail the sectoral composition of
R&D, we use data from the EU Industrial R&D Scoreboard, which are based on the
accounts of the 2,500 largest companies in the world in terms of R&D spending.4
A cross-check with the OECD data shows that these 2,500 companies account
for 80 to 90% of the R&D spending in the regions we analyse, making our data
representative of the actual total R&D spending.
THE MIDDLE TECHNOLOGY TRAP

Figure 1 shows the sectoral composition of business R&D spending (BERD) in
nominal terms for businesses headquartered in the four regions, plus a residual,
the rest of the world (ROW).5 In the US, high-tech industries – mostly software
& computer services and pharmaceuticals & biotechnology – account for 85% of
BERD. In the EU, by contrast, mid-tech industries – especially automobiles & parts
– account for roughly 50% of BERD, a much higher share for mid-tech industries
2 National funding is very concentrated, with Germany alone accounting for over 40% and France for almost 20% of the
EU (and national) total. It is sometimes argued that a selection of projects based on merit alone could favour the richer
Member States. But this is not the case. Appendix Figure 11 shows that across EU countries there is very little correlation
between income per capita and EU research funding as a percentage of GDP.
3 There are no substantial differences in government tax support for BERD either, since the EU spends 0.1% of GDP, while the
US spends 0.12%. See the OECD data here.
4 The data are taken from the EU Industrial R&D Scoreboard: https://iri.jrc.ec.europa.eu/scoreboard/2022-eu-industrial-
rd-investment-scoreboard. Representativeness should not be compromised, as global R&D is concentrated among the
world’s top R&D investors. Other sources provide scant data on R&D spending by industrial sectors. The OECD provides
some, but the classification varies and some EU member countries are either absent or have missing data for key industries/
years.
5 The figure does not include Amazon, which does not provide information on R&D expenses in its financial statement.
As an approximation we refer to the aggregate category “Technology and content”, that merges R&D spending with
infrastructure costs.
9
than in the US.6 The sectoral composition of corporate R&D spending by EU-
headquartered firms is more similar to that of Japan and China than to the US.7
Figure 1.
BERD by technology level 2022 (Top 2,500 companies)
BERD by Tech−level 2022 (Top 2500 companies)
500
R&D expenses (in billion €)
400
300
200
100
0
EU US China Japan ROW
Other High−tech Software & Computer Services Pharmaceuticals & Biotechnology

Other Mid−tech Automobiles & Parts Other
Source: Industrial R&D Investment Scoreboard (2023).
Not surprisingly, high-tech industries are much more R&D-intensive than mid-tech
industries, as shown in Figure 2. Therefore, the larger share of high-tech industries
in the US contributes to explaining why BERD is so much higher than in other
economies. What is more, evidence suggests that public-sector support is more
likely to crowd out business R&D in low R&D-intensity industries (e.g. Marino et al.,
2016; Szücs, 2020), which might explain the low business-sector multiplier in the EU
relative to the US.8 9
6 For the purpose of this exercise, we have used three broad categories – high- and mid-tech plus the remainder, ‘other’,
mostly including services and utilities. Our classification is similar to that adopted by Eurostat and the OECD:
• High-tech includes aerospace & defence, alternative energy, electronic & electrical equipment, health care equipment
& services, pharmaceuticals & biotechnology, software & computer services, and technology hardware & equipment.
• Mid-tech includes automobiles & parts, chemicals, financial services, fixed line telecommunications, industrial
engineering, industrial metals & mining, industrial transportation, leisure goods, mobile telecommunications, and
personal goods.
• Other includes banks, beverages, construction & materials, electricity, food & drug retailers, food producers, forestry
& paper, gas, water & multiutilities, general industrials, general retailers, household goods & home construction, life
insurance, media, mining, nonlife insurance, oil & gas producers, oil equipment, services & distribution, real estate
investment & services, support services, tobacco, and travel & leisure.
7 The high-/mid-tech classification reflects the idea that mid-tech industries might use advanced technologies as inputs
(e.g. autonomous driving software in autos), but the latter are created in “upstream” sectors.
8 One reason might be that R&D-intensive industries need resources far exceeding the typical amounts of a grant.
9 In the EU, €1 of public-sector spending is associated with €2 spent by the private sector. In the US, the private-sector
multiplier is equal to 3.
10
I. INNOVATION SPENDING COMPARED
Does the EU have an innovation problem?
Figure 2 shows that US and EU R&D intensity within high- and mid-tech industries
was similar until 2013. This implies that up to that point most of the difference in
aggregate BERD was driven by differences in the sectoral composition. After 2013,
however, the R&D intensity of US high-tech industries (and of others) began an
upward trend. In China, the R&D intensity increased even more: starting from close
to zero, it has now reached the EU level.10
Figure 2.
R&D intensity by technology level (% of sales)
R&D Intensity by technlogy level (% of sales)

EU US
15 15
10 10
High−tech High−tech
5 5
Mid−tech Mid−tech
Other Other
0 0
2003 2008 2013 2018 2023 2003 2008 2013 2018 2023
Year Year
Japan China
15 15
10 10
High−tech High−tech
5 5
Other Other
0 0
2003 2008 2013 2018 2023 2003 2008 2013 2018 2023
Year Year
Note: Sector lines weighted by share of total 2023 net sales. Sector R&D Intensity is the ratio of R&D expenditures
withinNote: Sectortolines
a sector theweighted bysales.
total net share of total net sales. Sector R&D Intensity is the ratio of R&D expenditures within a sector
to the total net sales.
Source: Industrial R&D Investment Scoreboard.
We decompose the difference in R&D intensity between the US and EU and find
that in 2022, 60% of it was due to differences in composition – the R&D-intensive
high-tech industry is larger in the US – and the remaining 40% to a generally higher
R&D intensity in the US across all industries.11 Before 2013, the composition effect
explained more than 90% of the disparity in average R&D intensity between the
US and EU.
10 This is not due to a reallocation across sectors within high-tech industries. More broadly, we observe very similar trends if
we focus on the main sectors separately.
11 Specifically, we apply a decomposition of the R&D intensity (RDI) gap between the US and EU using a standard shift-
share formula:
where i indexes high-, mid-tech and other industries, and S denotes the respective shares in both regions.
11
Clearly, specialising in mid-tech industries limits growth opportunities. While there

are examples of successful high-tech industries in the EU (see Box I for aerospace
and defence), the status quo is problematic because mid-tech industries have only
limited potential for sustained growth. We document that sales and profits tend
to grow faster in high-tech industries, as Appendix Figure 14 and Figure 15 show,
respectively.
BOX I
The EU aerospace & defence industry
This industry represents one of the few sectors where EU companies spend almost as much on R&D as those
in the US (€8.7 billion versus €9.4 billion). The main difference is the size of the market, as can be seen from net
sales of €260 billion for the US and €140 billion for the EU: defence spending is much higher in the US and most
of this spending naturally benefits US companies.
The near equality in R&D efforts (in terms of capital expenditure, the figures would be similar) suggests that EU
industry should be competitive in the open market. That is, EU firms should be able to satisfy a large share of
the increases in defence expenditure that are required immediately to support Ukraine and also those in the
medium and longer term, as Member States ramp up defence expenditures.
The European Defence Industrial Strategy can therefore count on a strong industrial base. The problem here
pertains much more to the market fragmentation within the EU and the limited role of joint procurement.
The recently-created European Defence Fund should at least partially address these long-standing problems.
But its main focus will be technologies close to commercial application, which carries the risk that it might
substitute, rather than complement, existing industrial R&D.
At least some of its budget could be devoted to bold transformative projects, like the Defense Advanced
Research Projects Agency (DARPA) in the US.
The mid-tech trap is a piece of a related and broader European problem: most
of the largest market caps in Europe are firms that (often under other names)
incorporated 50 or 100 years ago (or even more). This is not the case in the US.
This suggests a limited contestability that favors incumbents. A loose competition
policy enforcement does not seem to be the explanation for this incumbency
advantage, at least for the recent past: Europeans have become at least as tough
antitrust enforcers as the Americans. For the purpose of this report, we did not
have the time to investigate the respective roles of insufficient disruptive R&D and
of political lobbying in generating such industrial inertia.
12
PATENTS AS A MEASURE OF INNOVATION OUTPUT

R&D spending is an important indicator of public- and private-sector innovation
efforts. Another key measure of innovation output is patent activity. Although what
matters is the most important (usually often-cited) patents, the sheer quantity of
patents and its composition tell us an interesting story.
In 2022, around 270,000 Patent Cooperation Treaty (PCT) applications were

filed with the World Intellectual Property Organization (WIPO) from all over the
globe. Four regions, the EU, US, Japan and China, dominate patent activity with a
combined total of 220,000 filings. Figure 3 shows the shares of all patents filed in
the various regions under the PCT.12
This figure shows that in the same year, the EU accounted for 17% of all applications,
down from 24% in 2012. A similar decline is observed for the US, which accounted
for 27% in 2012 and 21% in 2022.
Figure 3.
International patents applications (PCT)
International patents applications (PCT) (% of global applications)
(% of global applications)
30%
27%
25%
25% 24%
22%
21%
20% 18%
17%
15%
10%
10%
5%
0%
EU US Japan China
2012 2022
Source: WIPO.
12 If the patent is accepted in the ‘international phase’, applicants can (but do not have to) simultaneously seek patent
protection for their invention in ‘national phases’ in a large number of countries, so that it de facto becomes an
‘international’ patent. Further information can be found at https://www.wipo.int/portal/en/index.html. See also the
statistical database of the WIPO: https://www3.wipo.int/ipstats/pmh-search/pct.
13
China’s share, by contrast, soared from 10% in 2012 to 25% in 2022. Still, a number
of sources suggest that this increase is the result of domestic subsidies, potentially
exceeding the cost of filing a patent (see Box II). For this reason, the figures for
China need to be taken with caution. Since it is not possible to disentangle the real
from subsidy-induced Chinese patent activity, we concentrate on data for the EU,
US and Japan.
BOX II
Subsidies for patents and trademarks in China
China appears to lead in many patent statistics, at least in terms of the sheer number of patents granted. Yet,
the Chinese numbers are most probably inflated by government subsidies.
A recent report (USPTO, 2021) discusses a number of measures introduced by the Chinese government since
2008, including patent subsidies, government mandates, bad-faith trademark applications and defensive
countermeasures, which together have contributed to inflating the number of patent filings in the region. In
particular, the report provides examples of subsidies that exceed the cost of registering a trademark, therefore
leading Chinese economic actors to pursue a trademark application even without any intention to use it
commercially.
Along similar lines, Beebe and Fromer (2020) present empirical evidence that in 2017, almost 70% of the patents
registered at the USPTO originating from China included fraudulent specimens, and that a substantial fraction
of these ended up being published.
Thus, non-market factors can explain why, despite the inflation of Chinese patents over the period of analysis,
few Chinese inventors file for patent protection overseas. They also explain the low commercialisation rate of
patented Chinese inventions.
All this does not imply that China is making little progress in science and technology, but that one has to look
more closely at the quality than the quantity of output.
Thus, in terms of aggregate innovation output and with the previous caveat
that we look at patent quantity rather than quality, the EU does not appear to
have a significant innovation problem. All the same, a closer look at the specific
technologies being patented reveals a different picture.
14
Figure 4 shows that each region tends to specialise in a different technological

field. We consider four broad technological domains: ICT (computer and digital);
electronics (electric, semiconductor and audio-visual); pharmaceutical and biotech;
and transport-related technologies, which together accounted for roughly 20% of
global applications in 2022.
Figure 4.
PCT applications byPCT applications
technology classby
(%technology classin all regions, 2022)
of applications
(% of applications in all regions, 2022)
70%
58%
60% 55%
50%
50% 48%
40%
29%
30% 24% 26% 26%
24% 23%
21%
20% 16%
10%
0%
Computer and digital Electrical, semiconductors Pharmaceuticals and Transport
and audi-visual Biotech
EU US JPN
Source: WIPO
The US accounts for more than half of the patents filed across these regions in two
sectors: ICT, and pharmaceuticals and biotech. Japan leads in electronics. Thus, the
EU lags behind other major economies in patenting in these key high-tech fields.
Instead, the EU dominates in patents filed for mobility technologies, accounting for
almost 50% of the total. Thus, the evidence on patents echoes the evidence of the
previous section about the specialisation of EU R&D activity.
Furthermore, the specialisation of the EU in mobility-related technologies has

increased over time. EU patent filings in the other three industries considered have
stagnated over the last 10 years. By contrast, US patent filings in ICT and pharma
have continued to grow strongly over this period (see Appendix Figure 12).
The EU seems stuck in a middle technology trap not only in terms of innovation
efforts, but also in terms of innovation output, as measured by patent counts.
PATH DEPENDENCY
How can one interpret such a large difference (in level and trend) in industry
composition of R&D efforts and patents between the EU and US?
Prospective profit margins are usually a key driver of R&D investment. One possible
explanation for the strong R&D effort by automotive producers in Europe could
be that their profit margin is high, thus justifying the strategic orientation of EU
producers. But this does not seem to be the case. Appendix Figure 16 shows that
15
in all regions, mid-tech industries tended to deliver lower profit margins than high-
tech ones.13
This difference in profitability should be the main incentive to move from the mid-
tech to the high-tech sector. This incentive was much lower in Europe, where on
average over the last twenty years, the profit margin of high-tech industries was
only about 2 percentage points higher than mid-tech ones, whereas in the US the
difference between high-tech and mid-tech industries was about 7 percentage
points. More recently, many European software firms tended to report losses,
further reducing the incentive to invest in such a key high-tech industry.
It is possible that the higher profit margins of US high-tech firms at least partially
reflect the near-monopoly position of US software giants in their respective markets.
But this does not alter the fact that the availability of higher profit margins for
US firms presented a strong incentive to invest in these industries. R&D-intensive
industries can be considered natural oligopolies, in which a few market leaders
emerge, sustained by the dynamics of large market shares fuelling R&D, which in
turn sustain large market shares in a virtuous cycle leading to dominant positions.14
In these industries, sales and R&D expenditures follow a similar pattern (Sutton,
2007).
The evolution of profits in our data reflects these patterns of natural oligopoly
formation. In 2003 (the first year for which data are available), profits for US high-
tech firms were already 3 times larger than for their mid-tech counterparts. By 2022,
US high-tech firms made 6 times more profits. By contrast, in 2003 the profits of EU
high-tech firms amounted to only a quarter of mid-tech ones and by 2022, this ratio
had only marginally increased. Looking at sales likewise reveals stark differences.
Table 1 shows the top-3 R&D spenders and their industries over time as a further
illustration of the diverging development of industries.15 It gives the top-3
companies in terms of R&D spending and their respective industries over a recent
20-year period in the US, EU and Japan.16 In the US, Microsoft is the only company
appearing more than once among the top-3 R&D spenders. Meanwhile, in the EU
and Japan, Volkswagen (VW), Mercedes and Toyota remain in the top-3 over the
20 years, while Panasonic, Bosch and Honda appear at least twice.
13 If one removes the software industry, the difference goes to around 4 percentage points. Thus, it is not exclusively the
software industry that drives the higher profitability.
14 We wish to thank Michele Polo for pointing this out.
15 As noted above, the table does not include Amazon, which in its 2022 financial statement reported €69 billion for
“Technology and content”, which comprises infrastructure costs in addition to R&D and cannot be disentangled. For
comparison, Alphabet reported €37 billion for R&D – three times what the EU spends annually with Horizon Europe, which
is described further below.
16 We do not include China because some companies there have changed their reference industry over the years.
16
Interestingly, in the US two of the three top R&D spenders in 2003 were also in the
automotive industry, but this changed over time. The software industry became
increasingly important over the years; by 2022, all top-3 spenders produced
software. In the EU and Japan, the auto industry tended to dominate throughout
the 20-year period. These patterns are consistent with the literature on path
dependence in innovation and industrial specialisation (e.g. Acemoglu, 2023;
Aghion et al., 2021; Aghion et al., 2016).17 18
Table 1.
Top-3 R&D spenders and their industries compared over time
2003 2012 2022
US Ford (auto) Microsoft (software) Alphabet (software)

Pfizer (pharma) Intel (hardware) Meta (software)
GM (auto) Merck (pharma) Microsoft (software)
EU Mercedes-Benz (auto) VW (auto) VW (auto)

Siemens (electronics) Mercedes-Benz (auto) Mercedes-Benz (auto)
VW (auto) Bosch (auto) Bosch (auto)
JPN Toyota (auto) Toyota (auto) Toyota (auto)

Panasonic (electronics) Honda (auto) Honda (auto)
Sony (electronics) Panasonic (electronics) NTT (telecom)
Source: Industrial R&D Investment Scoreboard (2004, 2013 and 2023).
The initial advantage of the US in high-tech was magnified over time, whereas EU
(and Japanese) industries remained in their specialisation pattern. Breaking this
path dependency might justify public-sector intervention to provide the seeds for
an alternative model of specialisation.
Appendix Figure 17 shows how global R&D shares in the automotive and software
sectors have evolved over the years. US companies have always dominated in
software and accounted for about three-quarters of worldwide BERD in that
category in 2022, while the EU is almost absent. The EU has long had a strong
position in automobiles. Unsurprisingly, Japan, its biggest competitor in the sector
so far, is also a major R&D spender in this area. Thus, the EU has much less of a
stronghold in the automotive industry than the US has in software.
Recent developments furthermore show that the large weight of the automotive
sector exposes it to vulnerabilities, such as the risk of being leapfrogged by vehicles
from the US and China. This illustrates the broader danger for the EU of being locked
17 Typically, in these models increasing returns to scale resulted in past advances in a given sector technology facilitating
further advances in the same sector.
18 These patterns are also consistent with evidence of declining business dynamism in many sectors around the world
(e.g. Akcigit, 2024; Biondi et al., 2023; Decker et al., 2020), and the existence of within-industry frictions hampering the
process of creative destruction (e.g. Cincera and Veugelers, 2014; Veugelers, 2021). Analysing these aspects goes beyond
the purpose of this study.
17
in a middle technology trap, where the current policy programmes are focused on
incremental improvements of mature technologies, rather than on industries with
greater potential for radical innovation. The purpose of EU innovation policy should
be to provide the initial impetus for the growth of new sectors and to support the
key role of new firms in developing EU high-tech industries.
BOX III
Does the EU have an investment problem?
Over the last decade there have been several EU initiatives to encourage (private) investment, the low level of
which is perceived to be a key obstacle to growth. For example, increasing investment is the purpose of the
InvestEU programme and its predecessor, the Juncker Plan. However, the total capital expenditure (capex)
of the top-2,500 companies in the Scoreboard is similar in the EU and US, as shown in Appendix Figure 13.
Actually, the overall investment-to-GDP ratio is at 23%, slightly higher in the EU than in the US (21%).
Once again, what is different is the composition of investment, with the share of high-tech industries being much
greater in the US than in Europe. Importantly, in 2022 the US software sector spent €104 billion on capex, mostly
on the computing power essential for the AI revolution, while EU companies spent only €1 billion. Such a 1-to-
100 ratio will make it very difficult for EU companies to catch up, or even to carve out significant niches in the
market for generative AI, which requires huge computing power to train large foundational models.
Aghion et al. (2024) propose a number of policies to support the development of AI in France and Europe. But
the scale of US companies’ spending is so large that even generous government support is unlikely to provide
EU companies with the required computing power.
18
II.
CAN THE EUROPEAN
INNOVATION COUNCIL
HELP EUROPE ESCAPE
THE MID-TECH TRAP?
19
A successful EU innovation policy should break the path dependency

discussed in the previous section and promote the creation of new
companies in the high-tech industry. EU resources should support
projects that are initially too risky to be financed by private funds, but offer the
potential of attracting business R&D once they reach a sufficiently mature stage
to scale them up.
In the US, spurring breakthrough innovation is a key objective of the Advanced

Research Projects Agencies – or ‘ARPA model’ of innovation. There are several
ARPA-style programmes in the US, each associated with a respective government
agency. Examples include the Defense Advanced Research Projects Agency
(DARPA) associated with the Department of Defense, ARPA-E with the Department
of Energy and ARPA-H with the Department of Health.19 Below we refer to “ARPAs”
as the set of agencies conforming to the ARPA model.
A key characteristic of ARPAs is their mission-oriented approach to innovation.20

The EIC devotes roughly half of its budget to predefined thematic calls, or ‘missions’.
Some of the literature suggests that mission-oriented R&D funding can be an
effective tool to crowd in private R&D (Pallante et al., 2021). On the other hand,
predefined thematic calls might prevent promising projects in other fields from
being financed; relatedly, they might also reflect the preferences of the European
Commission rather than the best technological prospects (see Box IV).
This section compares the ARPA model with the EU funding scheme that is
explicitly devoted to breakthrough innovation and was officially inspired by ARPAs:
the European Innovation Council (EIC).21 22
THE ARPA MODEL AND THE EIC

Similar to ARPAs, the details of EIC calls are set by programme managers in line with
the overarching objectives of the European Commission. Programme managers
refine the broad goals set in the work programme and include them in thematic
portfolios.
Still, despite efforts to emulate the salient features of the ARPA model, the EU
approach falls short in several respects.
19 There are other ARPAs, such as ARPA-I, related to the Department of Transportation, and IARPA, an arm of the Office of
the Director of National Intelligence. These agencies create the advanced technologies that are then used in the relative
“downstream” sectors.
20 Mission-oriented policies refer to a set of public-sector interventions aimed not only at promoting innovation but also at
directing technical change towards the achievement of well-defined technological or societal goals (Mazzucato, 2018).
21 The flagship EU programme for research and innovation is Horizon Europe. Later, in Section 3, we discuss in detail its
composition.
22 The EIC underwent a pilot phase between 2018 and 2020, and became fully operational only in 2021. See
https://eic.ec.europa.eu/about-european-innovation-council_en.
20
One key aspect, the type of projects financed by the EIC, can best be explained
using a technology readiness level (TRL) indicator, described in Figure 5. ARPAs
typically focus on developing ‘proof-of-concept’ (Azoulay et al., 2019) or projects
up to TRLs 3-4 at most. Once projects reach a sufficient maturity, usually taken to
be the demonstration stage (TRL 5 or above), they ‘graduate’ and leave ARPAs with
the expectation that private capital will flow and scale them up. Instead, the EIC
focuses on projects with TRLs above 5.23
BOX IV
Mission-oriented agencies vs. bottom-up project selection
It is customary to oppose a mission-oriented agency with predefined thematic goals, and a more open and
bottom-up approach. Of course, each support policy entails a bit of both. Even a bottom-up approach requires
allocating the total budget across programmes: the ERC for example has to allocate funds between mathematics
and economics, say, and there are multiple ways of proceeding to that purpose. Conversely, a mission-oriented
agency has discretion as to the strategy to be employed to achieve the stated goals. Indeed, this discretion
is a strength: the fact that the US government did not settle on one approach to a COVID-19 vaccine but only
defined a target was a wise policy (at the time it was far from certain that mRNA would be the most efficient
approach).
A mission-oriented agency facilitates accountability and adequacy to the needs; it makes most sense when
there is a clear view of what is needed, as in the case of DARPA (although even there, the DARPA inventions,
perhaps unexpectedly, went way beyond defense). On the other hand, the choice of thematic goals may miss
key research opportunities; and if political, it may reflect the officials’ preferences more than the scientific value.
Overall, choosing between the two approaches and their variants is a delicate balancing act. It is our feeling,
though, that the European pendulum has swung too far in the direction of a top-down approach.
23 Public support for projects whose TRL is below 5 can also be found at the national Level. For instance, in both France
and Germany there exist several programs/agencies aimed at supporting innovations that are still far from commercial
applications, i.e. low TRL projects (examples include in France the CEA, CNES and France 2030, in Germany the recently
created Federal Agency for Disruptive Innovation, Sprin-D). As our focus is on the efforts at the EU level, we did not
investigate how these national efforts could be better coordinated among themselves and with the EU level.
21
Figure 5.
Technology Readiness Levels (TRL)
9 ACTUAL SYSTEM PROVEN

IN OPERATIONAL ENVIRONMENT
8 SYSTEM COMPLETE AND QUALIFIED
Accelerator
SYSTEM PROTOTYPE DEMONSTRATION
IN OPERATIONAL ENVIRONMENT
6 TECHNOLOGY DEMONSTRATED
IN RELEVANT ENVIRONMENT
5 TECHNOLOGY VALIDATED
Transition
IN RELEVANT ENVIRONMENT
4 TECHNOLOGY VALIDATED IN LAB
Pathfinder & ARPA

3 EXPERIMENTAL PROOF OF CONCEPT
2 TECHNOLOGY CONCEPT FORMULATED
1 BASIC PRINCIPLES OBSERVED
Sources: authors’ representation based on official sources.
22
HELP EUROPE ESCAPE THE MID-TECH TRAP?
Azoulay et al. (2019) position ARPA-funded projects on the initial flat part of the
innovation S-curve, relating research effort and technical progress (Foster, 1986).24
On the initial part of the curve, a high degree of effort results in very limited
performance gains, and delayed payoffs limit incentives to pursue the project. This
is where public-sector support is most needed, because it addresses a clear market
failure (see Box V).
By comparison, projects with higher TRLs correspond to the final, flat part of the
S-curve and so are closer to their maturity. Uncertainty about the commercial
viability of the product or process involved is considerably lower. In this part of the
curve, subsidies can only be justified if they address capital market imperfections.
BOX V
Prerequisites for an effective industrial policy
1. Identification of the market failure so as to design the proper policy;

2. Use of independent high-level experts to select projects and recipients of public-sector funds;
3. Attention to the supply side (talent and infrastructure) and not just to the demand side (will there be the
talented people who are going to make this happen?);
4. Adoption of a competitively neutral policy;
5. A humble approach of not prejudging the solution, but instead setting objectives (as illustrated by the
Covid-19 vaccines: the vaccines could be based on several approaches, including the innovative but
relatively untested mRNA technology);
6. Ex-post evaluation and dissemination of policy effectiveness, and the inclusion of a ‘sunset clause’ in the
programme, forcing its closure in the event of a negative assessment;
7. Involvement of the private sector in risk taking, so as to avoid white elephants;
8. A process that is as expedient and hassle-free as possible.
For more detail, see chapter 13 of Tirole (2017).
24 The metric of technological progress depends on the technology considered, such as kilowatt-hour of electricity generated,
or computational speed.
23
GOVERNANCE OF EU INNOVATION POLICY (EIC)

The EIC is not a self-standing body.25 The EIC’s core bodies are managed by an EU
agency – the European Innovation Council and SMEs Executive Agency (EISMEA).
This agency ‘implements’ and manages other EU programmes aimed at SME
support. Thus, the management of the EIC programme was given to an existing
body whose task is to support SMEs (it used to be called the Executive Agency
for Small and Medium-sized Enterprises, EASME). The fact that the EIC was simply
added to the tasks of an agency for the promotion of SMEs already shows that,
behind the creation of the EIC, the mission of encouraging innovation is at best
competing with that of supporting SMEs.
The EIC Board and its president advise the European Commission on the broad
strategy of the EIC and on its work programme – a document specifying the various
funding opportunities. More specifically, the EIC Board advises the European
Commission on the thematic portfolios and the profiles of EIC programme
managers, both key elements of the EIC as discussed below.
The EIC Board and its president are appointed by the European Commission and
have an advisory role, thus leaving to the European Commission the bulk of the
decision power. This compromises the independence of the EIC and sets it apart
from the European Research Council (ERC), an EU-funded programme devoted to
basic research that has acquired considerable prestige over the years.26
The governance structure of the EIC is summarised in Figure 6.
Figure 6.
Governance of the EIC
COMMISSION
DG RESEARCH AND INNOVATION
EU
BUDGET
9
8
EISMEA 100% EIB 7
Share of
EIC Fund
6
Directorate I Directorate E 5
EIC
EI
BOARD Share of company X

4
CF
EIE Accelerator
Share of company Y
3
UND
Single Market Programme

Share of company Z
Transition
(SME Instrument)
Venture investment
Accelerator-funded 2
EIC
(B projects
)
1
Pathfinder AS RG
ED
IN LUXEMBOU
25 In fact, while referred to as the ‘Council’, in reality no such council exists.

26 The ERC is discussed in more detail in Section 3.
24
EIC Board members are not required to be leading scientists, and many have a
business background. This echoes the EIC inclination towards bringing projects to
market rather than developing new technologies.
Only 4 of its 21 members are professors and fewer than half have a degree in science
or engineering (see Box 6). Business ‘insiders’ might be better positioned to identify
key global challenges and technology needs, but they appear unfit to specify the
details of the work programme and identify the right people to manage scientific
projects. Indeed, research suggests that organizations with talented people are
better at screening new employees, and therefore a temporary advantage can be
self-sustaining (Board et al., 2017).27
Of the 10 EIC Board members with a PhD, only 7 have a profile in the Web of
Science, with an average of close to 2,700 citations and an h-index equal to 18.
One can compare these numbers with the average citations and h-index of the
ERC Scientific Council, respectively 18,000 and 58. The EIC Board thus has fewer
scientists, and they are less prominent than those of the ERC.
More details on the comparison between the governance of EIC and ERC can be
found in Appendix Table 2.
BOX VI
Can scientists be in charge of European funds?
The reluctance to implement a non-political distribution of funds seems to stem from a strong risk-aversion, a
concern that a non-political governance might generate a backlash in case of failure. A few remarks about this:
■ The chosen alternative is, except for the ERC, a political process, whereby the European Commission has
the final say on scientific choices. A most visible implication of political control is, in most of the European
innovation policy, the requirement of large consortia, under the pretext of connecting multiple parts of Europe
(as if professional scientists and managers did not know which potential partners would offer synergies!).
There may of course be less visible, but no less important drawbacks, such as the choice of inferior programs
or projects under (a) limited knowledge, or (b) the pressure of industrial lobbies, or else (c) electioneering.
■ The ERC has existed since 2007 and, besides receiving acclaim from the scientific community, has been
managed with integrity. In the US, DARPA, the National Science Foundation and the National Institutes of
Health, agencies all managed by scientists, have been distributing much larger amounts of money for decades
without substantial complaints. The presupposition that one must be a civil servant to behave with integrity
seems groundless. And so is the idea that independent agencies are “out of control”. As a matter of fact, it
is precisely an integrity reasoning that took the judiciary system, central banking, and competition policy out
of political reach.
■ Of course, any agency, whether independent or under political control, must be monitored. The selection
process of its leadership ex ante, and accountability ex post are central to obtaining a good performance.
27 This has implications for projects’ selection too, as we discuss further below.
25
BUDGET
The EIC has a total budget of €1.4 billion per annum and oversees three key funding
schemes: Pathfinder, Transition and Accelerator (see Figure 6 above). Only the first
two of these programmes finance the types of low-TRL projects that are typical of
the ARPA model and make use of highly-skilled programme managers. This implies
that at best €470 million of the EIC budget is managed similarly to ARPAs. Given
that the EIC should ideally cover all fields, this seems very small if compared with
the combined annual budget of US ARPAs, which exceeds $7 billion.28
The other half of the EIC budget is allocated to the Accelerator programme, which
finances projects close to commercialisation, i.e. projects with TRLs above 5.
Therefore, the support from the Accelerator programme for the scaling of almost-
mature technologies is unlikely to generate breakthrough innovation. Its purpose
is more to remedy a perceived capital market imperfection. The question is then
whether this use of EIC funds is the best way to address firms’ limited access to
capital and loans, if any.
WHO RECEIVES SUPPORT FROM EIC PROGRAMMES?

In light of the technological path dependency afflicting the EU innovation
ecosystem, new firms have a key role to play in boosting the development of
high-tech industries. While the Accelerator programme’s official aim is to support
startups, i.e. firms no older than 4 years, much of the funding is directed towards
projects by mature mid-tech companies.
First, the median age of Accelerator beneficiaries is 7 years, as revealed by a
sample of EU-funded projects matched with the dataset of Crunchbase, a widely
used database. Less than 20% of Accelerator beneficiaries are startups (Figure 7).29
Appendix Figure 18 shows that only 20% have revenue below €1 million but the
vast majority have fewer than 50 employees.30 This suggests that the EIC mostly
targets mature SMEs.
Figure 7.
Age of Horizon Europe companies by program
Age of Horizon Europe companies by programme
Pathfinder (TRL 1−4) Transition (TRL 3−6) Accelerator (TRL 5−9)
80 80 80
% of companies
% of companies
% of companies
60 60 60
40 40 40
20 20 20
0 0−4 5−7 8−10 11+ 0 0−4 5−7 8−10 11+ 0 0−4 5−7 8−10 11+
Years Years Years
Sources: authors’ calculations based on CORDIS and Crunchbase.
28 Considering only three ARPA-style agencies in 2024, DARPA has an annual budget of more than $4 billion, ARPA-E $0.650
billion, and ARPA-H $2.5 billion.
29 Pathfinder and Transition participants have a higher median age of 12-14 years, but that does not constitute an issue since
their missions are not to create new companies.
30 Because Crunchbase is a database of companies, we have data only for the private companies that participate in the
projects, but not for research institutions. Revenue and employment figures are estimates and need to be interpreted with
caution. Age, however, is highly accurate.
26
Our evidence is consistent with the EIC’s declared allocation of 70% of its budget
to scale up SMEs. This is not the typical task of ARPAs. Available sources suggest
that DARPA spends less than 2.5% of its budget ($100 million per year) on SBIR
and STTR, the two US programmes devoted to SME support.31 Other ARPAs
can participate in these programmes, but they are unlikely to spend more than
DARPA.32
It is an open question whether an innovation model so focused on SMEs will be

successful (Box VII).33 But it seems clear that the EIC aims more at remedying capital
market imperfections than encouraging breakthrough innovation.
The second, broader issue is that the EIC seems to be underfinancing the sectors that
could help the EU escape the middle technology trap. Specifically, Crunchbase lists
the industries associated with companies whose projects are financed by the EU.34
The detailed results are presented in Appendix Figure 19, which shows the number
of financed projects by industry. The overall result is that the industries usually
classified as high-tech, such as software and AI, appear relatively infrequently in
the data. Biotech appears more often than software-related industries, but lies far
behind the general categories of ‘science and engineering’ and ‘healthcare’, which
are not necessarily high-tech although they have some advanced technological
content.
31 Each federal agency with an extramural R&D budget of $1 billion or more is required to allocate a portion of its R&D
funding to conduct a multi-phase R&D grant programme for small businesses. The total EIC budget is not far from the $1.7
billion spent by the whole US Department of Defense in 2019 in favour of SMEs.
32 In FY2019, US agencies awarded $3.3 billion in SBIR funding. The Department of Defense and the Department of Health
and Human Services accounted for more than three-fourths of SBIR funding in FY2019.
33 For instance, Akcigit and Kerr (2018) estimate and simulate a structural model in which small firms are more likely
to innovate. Yet, Akcigit et al. (2022) find that in a world of imperfect information, higher productivity firms have a
comparative advantage at innovation and should be subsidised more than small firms. Santoleri et al. (2022) exploits
a regression discontinuity design to provide evidence that R&D grants to small and young firms with relatively mature
technology boost patenting and investment in intangibles; they do not find a significant impact of grants on early-stage
projects. Bronzini and Ianchini (2014) do find a positive impact of R&D grants to small firms on innovation, although these
did not result in crowding in additional private investment.
34 See the documentation from Crunchbase here.
27
BOX VII
Administering R&D grants: timing and selection problems
Researchers have evaluated the impact of R&D grants, in particular the Small Business Innovation Research (SBIR)
programme in the US. This programme has awarded grants (the government takes no stake in the project), on
a competitive basis, to small high-tech firms since 1982. Applications are ranked by officers, and this rank is
known only to the administration and not to the market, which observes only whether a grant application was
successful. Howell (2017) compares the outcomes of projects that almost received a grant to those that just
passed the bar (economists call this a Regression Discontinuity approach), that is projects that are assessed
as very similar in merit. She does so for the two phases of the programme. The “phase 1 competition” awards
small grants ($150,000). Phase 1 winners can 9 months later apply for a much bigger “Phase 2” grant ($1 million).
Sabrina Howell summarises her research in the following terms: “The Phase 1 award has powerful effects. First,
it increases a firm’s subsequent cite-weighted patents by at least 30 percent. Second, a Phase 1 grant increases
a firm’s chance of receiving venture capital (VC) investment from 10 percent to 19 percent, and also increases the
amount of money raised and the number of deals. Within two years of the grant, the effects on cite-weighted
patents and VC are just over one-half of their long-term effect. Third, a Phase 1 grant almost doubles the
probability of positive revenue and, conditional on positive revenue, increases it by 30 percent. Finally, Phase 1
increases the probability of survival and successful exit (IPO or acquisition).”
In contrast, Phase 2 grants have little such effects, so the author concludes that it would be more efficient to
reallocate that money to Phase 1. A few other observations are interesting. First, conditional on winning a Phase
1 grant, ranks are relatively uninformative about outcomes. This suggests unfortunately that officials struggle
to analyse the quality of proposals, a fact consistent with the view that only a few experts have the ability to
assess a project’s potential. Second, the associated limited certification brought about by winning a Phase 1
grant raises the question of what explains the success of Phase 1; these small grants (to be compared with the
$9 million average VC round in her data) are found to relax financial constraints of would-be innovators, allowing
them to develop a proof-of-concept. Third, a likely explanation for the poor performance of Phase 2 grants is
that officials face an adversely selected sample (see Lerner, 2000). Entrepreneurs prefer private money for a
variety of reasons (Gans and Stern, 2003) and so those who receive favorable information in Phase 1 do not
apply for a Phase 2 grant – or else cannot apply as they already have non-negligible private funding.
28
ROLE OF PROGRAMME MANAGERS

The key role of ARPA project managers is to identify projects with high potential for
breakthrough innovation. To accomplish this task in a broad set of scientific fields,
ARPAs rely on hundreds of highly-qualified programme managers (or directors),
who manage a portfolio with a limited number of projects and enjoy a high degree
of discretion in project selection and management. These programme managers
have an entrepreneurial mindset/background and a strong incentive to decide
whether to put more means on functioning projects or stopping/abandoning them
despite sunk costs.
Meanwhile, the EIC has only 9 programme managers, each managing projects
worth around €200 million annually.35 What is more constraining than the financial
amount each EIC programme manager has to oversee is the sheer number of
projects and the wide range each manager has to cover. The small number of
programme managers severely limits the diversity of available expertise.
Moreover, less than half of the €470 million devoted to breakthrough innovation
is managed by ARPA-style programme managers. The majority of EIC funding
(roughly 55%) is awarded through open calls with no predefined thematic priorities
(‘EIC Open’) and without the involvement of programme managers. 36 By contrast,
ARPAs publicise funding opportunities primarily by soliciting proposals tied to
programme-specific areas of research and development, set and managed by
programme managers, with unsolicited proposals being allowed but generally
discouraged. 37
ELIGIBILITY REQUIREMENTS
Unlike ARPAs, but similar to other Horizon Europe funding schemes, two of the
three EIC programmes require projects to be collaborative, meaning that applicants
must form consortia of independent legal entities, with at least one established in
a Member State, and at least two others in different Member States or Associated
Countries.38
Almost all Pathfinder projects are run jointly by companies and research institutions.
In Transition, 13% of projects are run solely by companies, and 6% solely by research
institutions alone. Accelerator exclusively finances single companies and does not
require collaboration. The average size of consortia is six members in Pathfinder
and three in Transition. Accelerator does not have collaborative requirements.
These statistics must be seen in conjunction with the average size of the grants
disbursed: €3.2 million for Pathfinder, €2.3 million for Transition, and €2.3 million for
35 Programme managers manage only Pathfinder and Transition projects, whose annual 2022-2023 budget was €0.47
billion. Roughly 40% of this budget (Challenges calls) is managed by the 9 programme managers.
36 Programme managers manage only Pathfinder and Transition projects, whose annual 2022-2023 budget was €0.47
billion. Roughly 40% of this budget (Challenges calls) is managed by the 9 programme managers.
37 Recent evidence from Howell et al. (2021), based on the US, suggests that projects selected through open calls involve
younger and smaller firms, and that they are successful according to different metrics of innovation activity. However,
these firms are less likely to win further funding competitions.
38 This applies to Pathfinder Open, but not necessarily to Pathfinder Challenges. Collaboration applies only to part of
Transition and it does not apply to Accelerator, which are targeted at individual legal entities (startups).
29
Accelerator. This implies that each consortia member in the first two programmes
receive on average just €0.5-0.75 million in grants. It is unlikely that these very small
sums can generate breakthrough innovations.
The application success rate of EIC projects is quite low. The average success rate
for all of Horizon Europe is 16%, but for the EIC it is less than 10% (with some
variation across subprogrammes).39 It would be a mistake, though, to conclude
that, for a given budget, a low success rate jeopardizes innovation. The success rate
may reflect the cost of writing proposals (a low cost inducing more applications).
Another determinant of the success rate is the quality of screening. As Adda and
Ottaviani (2023) show, the number of applications for a grant call increase with
the imprecision of the grading; in a nutshell, the more random the allocation of
grants, the more weak projects have a chance of being selected, and the more
applications are received, increasing the rejection rate. So, it may be the case that
a high rejection rate is a symptom of an insufficient quality of screening rather than
of a size of grants that exceeds the optimum one given the fixed budget for the
programme. Thus, a low application cost or an accurate screening, while desirable,
may well decrease the success rate.
Furthermore, increasing the acceptance rate given a fixed budget would lead to
sprinkling European money, with limited ability for Horizon Europe to have a real
impact on the best projects. So, just increasing the acceptance rate for a fixed
budget is unlikely to be a good idea.
The selection mechanism might be improved by being articulated in two phases. The
first phase could be a streamlined version of the current screening process under
the supervision of experts who would have the power to ‘desk reject’ proposals
that are anyway unlikely to be approved. One reason for the success of the ERC
is that its selection panels involve top scientists. But these individuals should not
devote their attention to unpromising applications; instead, their valuable time
should be used to select the best projects. It might thus be useful to associate
several specialised scientific boards with the EIC, composed of top scientists and
engineers. Their task would then be to make the final decision among the best
applications that have survived the first screening process. In this way, they would
not spend their time on projects whose rejection is a no-brainer, but rather on
comparing marginal projects, i.e. projects for which the decision is complex.40
39 The data are available here.

40 A more fundamental issue is that almost by definition, new fields of science will have only a limited number of experts
working in them. This restricts the pool of potential funding awardees, creating a conflict of interest for the programme
managers, who are likely to personally know them. International cooperation with non-EU scientists might help solve the
problem.
30
FUNDING DELAYS
EIC funding of successful proposals is expected to take place in 8 months. In ARPAs,
it can be as little as 3 months. This may be due to several factors. For instance,
EIC officials evaluate proposals following complex guidelines requiring special
briefings. Awardees are subject to a Commission-wide grant agreement contract
(241 pages long) and additional provisions specific to the EIC programmes.41 By
contrast, proposals to ARPAs are required to follow the ‘Heilmeier Catechism’, a
concise list of questions speeding up the evaluation process, without compromising
clarity and informativeness.
A related issue is that ARPAs have the authority (typically delegated by their
respective governmental agency) to enter into, and flexibly administer, contracts
and grant agreements. By contrast, EIC funding has to follow standardised EU
procedures that tend to be highly bureaucratic.42 There has been progress in the
speed of decisions and disbursement, but unless procedures and contracts are
radically reformed, little further progress seems feasible.
41 Different EIC stakeholders have complained about intellectual property issues with the current model.
42 Most notably, ARPAs use ‘Other Transactions’, contracts designed to avoid burdensome regulations, complicated
accounting rules, and extractive intellectual property terms that characterise federal contracts.
31
III.
ANATOMY OF HORIZON
EUROPE
32
III. ANATOMY OF HORIZON EUROPE
H orizon Europe has a budget of about €11 billion per annum. Increasing
the EU’s budget for innovation would be highly desirable but might be
difficult for political reasons; therefore, our aim is to propose budget-
neutral reforms. To that end, this section identifies which parts of Horizon Europe
could be repurposed to finance breakthrough innovation.
Figure 8 presents the structure of Horizon Europe, which is organised around three
‘pillars’. The figure provides the breakdown of the average 2021-2022 R&D grants
paid by the EU through its various research and innovation programmes.43
Figure 8.
Horizon Europe by its pillar components and relative subprogrammes
Accelerator: 0,41 bn EIT: 0,30 bn
(TLR 5-9) EIE: 0,06 bn
Accelerator: 0,41 bn EIT: 0,30 bn
Transition: 0,11 bn (TLR 5-9) EIE: 0,06 bn
(TLR 3-6)
Transition: 0,11 bn
Pathfinder: 0,35 bn (TLR 3-6)
(TLR 1-4)
Pathfinder: 0,35 bn
(TLR 1-4)
Pillar I
Pillar I
Pillar II
Pillar II
Pillar I Pillar II Pillar III

Excellent Science Global challenges Innovative Europe
Pillar I collab. projectsPillar II collab. projectsPillar III
€3.25 billion
Excellent Science €7.33 billion
Global challenges €0.96 billion Innovative Europe
28% of total funding collab. projects
64% of total funding 8% of total fundingcollab. projects
€3.25 billion
European Research
28% of total funding €7.33 billion €0.96 billion
Council (ERC) 1. Cluster (Health)
64% of total funding 8% of total funding
European Research 2. Cluster (Culture & Creativity)
Marie Skłodowska-Curie 3. Cluster (Civil 1. Cluster (Health)
security)
Actions (MSCA) Council (ERC) 4. Cluster (Digital &2.Space)
Cluster (Culture & Creativity)
Marie Skłodowska-Curie
Research Infrastructure 5. Cluster (Climate,3.Energy
Cluster&(Civil security)
Mobility)
Actions (MSCA) 4. Cluster (Digital & Space)
6. Cluster (Bioecononmy & Environment)
Research Infrastructure 5. Cluster (Climate, Energy & Mobility)
6. Cluster (Bioecononmy & Environment)
Source: CORDIS.
43 The main source of data is the Community Research and Development Information Service (CORDIS), the European
Commission’s primary public repository on all EU-supported R&D activities. It should be noticed that the sums allocated
from one year to another can vary to some extent. Also, they do not include venture capital investment, which is discussed
further below.
33
PILLAR I
The first pillar funds basic and applied research; it includes the well-known ERC.
Due to data availability, the ERC is the most widely studied among EU funding
programmes (e.g. Nagar et al., 2023; Ghirelli et al., 2023). The first pillar has a yearly
budget of roughly €3 billion and accounts for the vast majority of basic research,
around 30% of Horizon Europe.
The ERC accounts for about two thirds of the budget of Pillar I. It is characterised
by its own government structure, which differs from other Horizon Europe
programmes. Specifically, the ERC Scientific Council is composed of independent
and internationally reputed scholars. Its president is appointed by the European
Commission, which selects one among three names proposed by an independent
commission, the Scientific Council having a veto right on these names. Unlike for
other Horizon Europe programmes, the ERC work programme is set by the Scientific
Council itself, without interference from European Commission officials. 44
Likewise, the scientific panels of the ERC are made up of renowned scientists, who
pick the laureates without any intervention from the European Commission (other
than checking the conformity of applications and grant awards with the general
rules). The award process establishes shortlists on which much of the skilled
attention focuses, as it should.
It is therefore no surprise that, like its US counterparts the National Science

Foundation (NSF) and the National Institutes of Health (NIH), the ERC functions
with integrity and funds disruptive science. For example, Ugur Sahin, the BioNTech
founder, received in 2018 a grant from the ERC to explore the mRNA technique in
order to develop mRNA vaccines for various types of cancer. The scientist behind
the AstraZeneca vaccine had also received an ERC grant.45
However, the ERC has a much smaller budget (about €2 billion) than its transatlantic
counterparts. In 2023, the NSF had a budget of $9.5 billion, with $7.6 billion
dedicated to research. The same year, the NIH had a budget of around $47 billion.
It should be noticed that these sums do not include the budget of ARPA-H ($2.5
billion).
44 Anecdotal evidence of past conflicts between the Scientific Council of the ERC and the Commission suggests that the
independence and academic excellence of the ERC cannot be taken for granted and must be safeguarded. In 2020,
there was a clash between the newly nominated president and the other 20-members of the ERC scientific Council, who
unanimously asked him to resign after only 3 months. The issue was his failure to represent properly the interests of the
ERC and the priority he gave to personal activities, (strangely) authorised by the European Commission. More generally,
there has been a political backlash against the independence of central banks, competition authorities and the judiciary,
despite the fact that these institutions have served the general interest very well whenever competent and impartial
leaderships were selected.
45 For more on the need for fundamental research and on European attitudes towards it, see ‘The Future of European Science’
by the former head of the ERC, Prof. Jean-Pierre Bourguignon.
34
PILLAR II
The second pillar is the largest in terms of budget – over €7 billion annually or 64%
of the Horizon Europe budget. The official title of Pillar II is ‘Global Challenges and
European Industrial Competitiveness’. It thus has a thematic focus. This pillar has
two salient features.
The first is that it adopts a top-down approach by eliciting applications on specific

topics decided by the European Commission46. It is organised around six main
thematic clusters. These are: Cluster 1, Health; Cluster 2, Culture, Creativity and
Inclusive Society; Cluster 3, Civil Security for Society; Cluster 4, Digital, Industry and
Space; Cluster 5, Climate, Energy and Mobility; and Cluster 6, Food, Bioeconomy,
Natural Resources, Agriculture and Environment. These themes reflect the
judgement of the Commission of what constitute the main global challenges.
The second feature is that it funds collaborative projects by imposing consortia

requirements, designed to strengthen EU country representativeness – a formal
blending of EU cohesion policy and innovation policy. Figure 9 shows the
distribution of projects by number of consortium members in each pillar. Pillar
II features many projects with a very large number of participants. It is an open
question whether the cross-country collaboration has a positive net impact.47 The
impact is probably case-specific. One thing is certain: top-down collaboration
requirements do not work. Teams collaborate on a proposal to grab their share
of money and the announced collaboration is rarely operational. Bottom-up
collaborations are a different ballgame. So, our recommendation is of course not
to rule out collaborations, but rather to not make them mandatory. In this way, only
value-creating collaborations will emerge: if the teams deliver more through an
alliance, they will form the latter so as to increase their chances of success.48
46 Conclave Europe (2023) argues that ‘R&D initiatives must be based on bottom-up projects, while top-down support
should focus on a reduced number of strategic areas with enough continuity’.
47 On the one hand, this might be desirable given that there is literature suggesting that collaboration – research consortia
in particular – generates positive outcomes, especially if involving basic research (e.g. Branstetter and Sakakibara, 2002).
However, imposing collaboration across multiple entities in different countries might hamper innovation outcomes. This
possibility is supported by a large body of literature studying the negative impact of coordination and communication
frictions (e.g. Backer and Murphy, 1992), with some contributions arguing that small teams are better at developing
radical innovation (e.g. Wu et al., 2019).
48 In this we agree with Conclave Europe (2023) that ‘Europe must eliminate the geo-return rule, which ties contract
allocations to national contributions, prioritizing competitiveness and smart investment over political motives’. This rule
used to plague the funding of fundamental research prior to the creation of the ERC, and still constrains some important
agencies such as the European Space Agency.
35
Figure 9.
Number of consortium members by pillar
Horizon Europe consortia size
Pillar I - Excellent Science Pillar II - Industrial Competitiveness
4000 250
Number of projects
Number of projects
200
3000
150
2000
100
1000
50
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Consortia size Consortia size
Pillar III - Innovative Europe

Number of projects
600
400
200
0
0 10 20 30 40 50 60 70 80
Consortia size
Source: authors’ calculations based on CORDIS.
Appendix Figure 20 shows total and per-project funding by thematic cluster,

distinguishing between basic and applied research, and product development.
The expected TRL of the funded projects varies depending on the call, and
a similar share of funding is devoted to research and development. Other than
the collaborative aspect, however, Pillar II differs from the EIC in its mission of
boosting “industrial competitiveness”, rather than “supporting the development of
disruptive and market-creating innovations”, like the EIC.
The average Pillar II grant per project is around €6 million. Bearing in mind that the
number of consortium members can easily be around 10-20, like for the EIC, one is
left to wonder how impactful such fragmented funding can be.
Finally, the governance of Pillar II is handicapped by its political process. Not

having programme directors independently selecting and managing projects risks
undermining their scientific value in favor of officials’ preferences.
36
PILLAR III
The objective of the third pillar is funding the development of ground-breaking
technologies. It is composed of the EIC (the largest programme discussed in the
previous section), the European Institute of Innovation & Technology (EIT) and the
European Innovation Ecosystem (EIE). We argue that the EIT and the EIE do not
have a clear mission and it is hard to understand what they do. Given that they
absorb substantial resources, we review them below. We also describe the EIC
Fund, the venture capital arm of the EIC.
European Institute of Innovation & Technology

The EIT, headquartered in Budapest, was created in 2008 and its mission has
changed over time. At present it aims at encouraging the formation of ‘networks’.
Its mission is to connect education, research and business, not to finance innovation.
Yet, its declared budget is about €300 million per year, close to the budget of
Pathfinder. Official data from the European Commission show that it financed eight
projects, on average around €50 million, more than ten times the average grant
from the EIC. The eight projects correspond to eight Knowledge and Innovation
Communities (KICs) – whose categories are largely similar but somewhat more
granular than the thematic clusters of Pillar II.
The EIT governing board selects the KICs and evaluates them every 3 years with
the support of external experts.49 The KICs typically receive financing for between
7 and 15 years, which implies that tens of million in grants are paid each year to the
same entity, which in turn independently decides how to allocate the grant to the
beneficiaries of its activities.50 Appendix Figure 21 shows the cumulative amount in
grants paid to the KICs, some reaching over €700 million over their term.
KICs are supposed to create ‘meeting hubs’ around Europe for workshops,
networking and incubator activities, but it is not clear how they operate or allocate
the funding. Each KIC has its own governance and operational approach, implying
reduced accountability and fragmentation of these funds. Moreover, it is not clear
how the grants paid by the EIT differ from those awarded by the EIC or those in
Pillar II, whether the eligibility requirements and contractual forms used by KICs
differ from the rest of Horizon Europe, or how the EIT differs from other training-
focused programmes such as Erasmus+.51
Finally, and more broadly, it is not clear why the EIT should be better placed than
scientists and businesses to identify potential partners and synergies. For these
reasons, it is natural to question the EU value added of the EIT and ask whether its
budget could be more efficiently allocated elsewhere.
49 See Appendix Table 2 for a comparison of the EIT’s governance with that of the EIC and ERC.
50 Financing data are available here. Some additional info is here.
51 This has been recently argued also by Denmark’s representatives. See this article here, this one here and another here.
37
European Innovation Ecosystem

The smallest component of Pillar III is the EIE, a funding programme aimed at
‘creating more connected and efficient innovation ecosystems’ across the EU. The
overall budget for the period from 2021 to 2027 is relatively small – roughly €60
million per year. Given that Pathfinder provides individual grants of up to €4 million,
this implies that the funds of EIE would enable it to finance at least 15 potential
breakthrough projects.
A key goal of the EIE is to overcome spatial issues of specialisation and integration
in global value chains in less developed regions. However, it is not clear how the
programme would achieve such a goal or how the funding it provides to SMEs
differs from other existing programmes, which in this case also casts doubt on
whether the programme serves any specific purpose.
EIC Accelerator and EIC Fund

The EIC Accelerator has the largest declared budget of the EIC, accounting for half
to three-quarters of the total EIC budget depending on the year. It is implemented
through an agency, the EISMEA. which collects and evaluates proposals by startups
and SMEs through public calls, split roughly equally between ‘EIC Open’ and ‘EIC
Challenges’.
Following selection, and conditional on the rubberstamping of the decision by

the European Commission, EISMEA can disburse a grant (up to €2.5 million). But
while the Pathfinder and Transition programmes offer support exclusively through
grants, selected Accelerator companies can receive much more financing (up to
€15 million) in the form of equity.
Indeed, more than half the budget allocated to Accelerator takes the form of
equity investment, around €400 million depending on the year.52 The equity
investment is delegated to the EIC Fund, the venture arm of the EIC. It is a capital
fund subject to private law, with the European Commission as the unique investor
and shareholder. 53 This implies that all risk, profits and losses are borne by Member
States.
The EIC Fund decides on financing operations and does the monitoring, milestone
disbursements, reporting and exit.54 The external Fund Manager, AlterDomus
Management Company S.A., is responsible for making investment decisions
following the selection process designed by EISMEA and due diligence by the
EIB.55
52 See Annex I here.

53 Official EU sources state: ‘In the future it is envisaged that the European Investment Bank (EIB) will take over this role
under a mandate agreement.’
54 The EIC Fund Board is vested with the broadest powers of the EIC Fund. The Board oversees the investment strategy and
makes the final decision on any single operation. The EIC Fund Manager is an external authority that manages the EIC
Fund. It makes all the investment and divestment decisions on the selected companies. With the support of the EIB, it
manages the EIC portfolio of invested companies.
55 The EIC Fund Manager receives a commission fee of up to a 10%. Once the EIC Fund has approved the investment, the EIB,
in its role of an investment adviser, guides the work of the lawyers for each specific transaction. This leads to an investment
agreement, which is signed by the EIC Fund and the company.
38
Despite the EIC’s focus on funding ‘unicorns’, 55 of the top-100 startups in the
world are from the US, while only 8 are EU companies.56 Moreover, in a sample
of 5,859 companies that received Horizon Europe funding, from Crunchbase, we
find that less than 10% of funding rounds involved additional investors; for EIC-
funded projects, it was less than 5%.57 Some of these are private funds. This
evidence suggests that EU-backed venture capital investment mobilises little
private investment, possibly due to adverse selection. That is consistent with the
literature, which casts doubt on the effectiveness of public-sector venture capital
in boosting innovation.58 In a nutshell, it is not obvious that a statist body, such as
the Executive Agency for Small and Medium-sized Enterprises within the Research
and Technical Development framework, can emulate the best private equity funds.
These concerns are supported by data from the EU Innovation Scoreboard

(Grassano et al., 2022), which show that: (i) EU-based corporations provide about
22% of global corporate venture capital, but EU-based startups receive only 9%
of the global total; and (ii) over 80% of venture capital investment by EU-based
companies targets US-based companies (Appendix Figure 22). This seems an
indication of a lack of investment opportunities, casting doubt on the claim that
developing its venture capital market would solve the EU’s innovation problem.
Finally, and more broadly, it is unclear whether an EU venture capital market can
be as developed as it is in the US.59 Reasons include differences in fiscal regimes,
corporate and labour laws, which in the EU are often inappropriate for the world
of startups and innovation.60 Cultural differences can also play a role. In the EU,
universities – at which researchers, by lack of incentives or because of ideological
reasons – are loath to participate in applied innovation. In the US, the general mindset
emphasises advice, handholding, monitoring and relatively early termination –
values that are not necessarily shared in the EU. And entrepreneurship is very rarely
taught in the EU at the undergraduate level. Another often-mentioned hindrance
is the absence of a single home market in a number of European tech markets
(defence, AI, energy, healthcare... ), complicating the emergence of sufficient scale
technology champions.
56 See the EIC Impact Report 2022. The full list from Crunchbase can be found here.
57 According to Crunchbase, 30% of EU-funded companies never benefited from additional investors. However, we cannot
observe whether the remaining 70% got funding from private sources, or whether the investments followed or arrived
before the EU contribution.
58 For instance, Breschi et al. (2021) find that government-backed venture capital underperforms its private counterpart,
and that mixed public-private investment fails to attract additional private investment in later funding rounds. For these
reasons, it might be better to use Accelerator funding solely for R&D grants, which are generally seen as an efficient policy
instrument to boost innovation (Teichgraeber and Van Reenen, 2022).
59 See Coatanlem (2024).
60 Risky strategies mean a high probability of failure and a need to keep moving on, making slow and expensive layoff
processes particularly cumbersome (see Bartelsman et al., 2016). The corporate law literature points to substantial
differences between the flexible US legislative system – largely deferring contractual choices to private parties – and the
EU, where regulatory rigidities might prevent the development of an efficient venture capital market (see Enriques et al.,
2024).
39
Innovation Fund
The Innovation Fund is an EU fund for climate policy, independent from Horizon
Europe, which invests in innovative Green-tech projects with the goal of reducing
European emissions. It is entirely financed by the Emissions Trading System (EU
ETS) and hence its overall funding depends on the carbon price. Given a carbon
price of €70/tCO2, it might amount to annual funding of about €4 billion.
One might argue that, for a given climate objective, the higher the ETS carbon
price, and therefore the stronger the European abatement ambition, the less need
there is for green innovation support, all the more that a high carbon price already
allows green innovators to monetize their technologies. While a high carbon price
and high funding of green R&D are excellent news given the perilous climate
situation, it is by no means clear that the financing mode as such is desirable. For
certain it exposes green R&D funding to a populist backlash leading to a weak ETS
price. In any case, even after the recent fall to about €60/tCO2, the result of higher
ETS prices has been that a substantial amount of funding will be available to foster
research on green energy solutions. It is thus important that this fund be effectively
used.
The Innovation Fund is managed by the European Commission through delegation

to the European Climate, Infrastructure and Environment Executive Agency
(CINEA), led by four heads of department, who are responsible for more than
seven other programmes, including Cluster 5 of Pillar II in Horizon Europe.
The Innovation Fund is actually not about innovation, but rather finances large-
scale demonstration projects (with TRLs above 5). A large portion of these projects
is related to hydrogen, which exposes the fund to the risk involved in a scarcely
diversified portfolio.
The fund mostly awards grants through calls for proposals, and in some part through
competitive bidding (auctions). Specifically, projects need to be sufficiently mature
in terms of planning, business model and financial and legal structure, which sets
the Innovation Fund apart from the kind of breakthrough innovation the EU needs.61
61 The financed projects might use advanced technologies that are produced in high-tech industries, but they are not aimed
at developing high-tech itself.
40
IV.
A BUDGET-NEUTRAL
POLICY PROPOSAL
TO BOOST EUROPEAN
DEEP-TECH
41
T his section formulates a concrete policy proposal to boost the EU’s

capability for breakthrough innovation.
Reallocation towards disruptive innovation

Panel (a) of Figure 10 depicts the status quo by illustrating the declared 2024
budget of key components of the EU ecosystem for innovation.
Our proposal is articulated in five steps:
1. Reform the governance of the EIC, hiring more independent and highly
qualified programme managers and giving them greater discretion over
project selection and management. Streamline management practices and
simplify the application process. Design a procedure that allows prominent
and busy scientists to cast the decisive vote on the best proposals.
2. Scale down progressively the European Institute of Technology (EIT) and the
European Innovation Ecosystem (EIE). The current projects (KICs) of the EIT
are anyway expected to become financially independent in 7 to 15 years.
Over time, this could free up to €0.51 billion per annum.
3. Replace the financing of equity stakes by the Accelerator budget with other
sources whose mission is investment, rather than innovation. For example, the
EIC could be merged with the European Investment Fund (EIF, which does
not rely on EU budgetary resources) or the proposed Sovereignty Fund. This
would free up €0.41 billion per annum.
4. Use the resulting €0.92 billion for grants to increase the budget for ARPA-
style programmes, through Pathfinder and Transition.
5. Pool a share of resources from Pillar II clusters, as well as part of the Innovation
Fund, to finance Pathfinder-type programmes (or create two thematic agencies
on energy and health, with the same reformed ARPA-style governance of the
EIC).
Panel (b) of Figure 10 depicts Horizon Europe after the reform. The dark colouring
of the new EIC represents an ARPA-style governance, which extends to two new
agencies through the dark bands. These are Pathfinder/Transition-E and Pathfinder/
Transition-H.
42
TO BOOST EUROPEAN DEEP-TECH
Accelerator
(VC) (Grants):
nd
0,27 bn
EIC Fu
EIT:
0,41 bn 0,42 bn EIE:
Figure 10.
0,09 bn
Before and after our policy proposal (in € billion)
Innovation Fund: Transition:
4,00 bn 0,09 bn
Accelerator
Pathfinder: (VC) (Grants):
0,26 bn nd Pillar I:
0,27 bn
EIC Fu
EIT: 3,47 bn
0,41 bn 0,42 bn EIE:
0,09 bn
Innovation Fund: Transition:

4,00 bn 0,09 bn
Pathfinder:
0,26 bn Pillar I:
3,47 bn
Pillar II (6 Clusters):
4,11 bn
Pillar II (6 Clusters):
4,11 bn
Invest
Accelerator: VC
me
Transition:
0,27 bn
nt
0,34 bn
Pathfinder: 0,41 bn
0,93 bn
“Pathfinder/
Innovation Fund: Transition-H“
4,00 bn (part of 0,52 bn
Cluster 1: Health)
Invest Pillar I:
Accelerator: VC 3,47 bn
me
Transition:
0,27 bn
nt
0,34 bn
Pathfinder: 0,41 bn
0,93 bn
“Pathfinder/
Innovation Fund: Transition-H“
4,00 bn (part of 0,52 bn
Cluster 1: Health)
Pillar I:
3,47 bn
“Pathfinder/
Transition-E“
(part of 1,11 bn Cluster 5:
Climate, Energy & Mobility
+ part of 4,00 bn from
Innovation Fund) Pillar II (other Clusters):
2,48 bn
Source: authors’ representation. “Pathfinder/

Transition-E“
(part of 1,11 bn Cluster 5:
Climate, Energy & Mobility
+ part of 4,00 bn from
Innovation Fund) Pillar II (other Clusters):
2,48 bn
43
Pathfinder/Transition-E would would focus on energy, climate and transportation

matters, and pool resources from Cluster 5 of Pillar II and the Innovation Fund.
The Innovation Fund has a much larger budget than the EIC, so devoting only a
small portion, say 10-20%, of its budget to breakthrough innovation would free up
resources equivalent to what is currently available to the EIC.
Pathfinder/Transition-H would deal with health issues, and pool resources from
Cluster 1 of Pillar II.62 Resources from the original, ‘general’ Pathfinder and Transition
programmes could also be used to finance the new agencies.
More broadly, we argue that through the EIC, the ARPA-style governance model
could be extended to other components of the EU innovation landscape. A natural
candidate is the European Defence Fund, which represents a long-overdue effort
to foster cooperation in defence-related R&D. Currently, the fund devotes only
4-8% of its budget to disruptive innovation.
The amounts on the table for promoting disruptive innovation are very small in the
EU (see Box 8). It is therefore important to find a solution for bringing more money
to this end. Yet, il is also interesting to note that we could get “more bang for the
buck”. Our proposed reform is therefore EU budget-neutral (except for the minor
cost of hiring programme managers).
About €400 million would be needed to finance the equity stakes that currently
use up to 60% of the EIC Accelerator programme. Buying equity stakes in startups
represents a financial operation that does not need to be funded by budgetary
resources. Alternative funding could potentially come from an EU sovereign fund,
or the existing European Investment Fund – part of the European Investment Bank
group – whose remit could be enlarged to take over the equity stakes from the
Accelerator programme in its portfolio.
MONITORING, EVALUATION AND DATA SHARING

In line with its current focus on remedying capital market imperfections, the key
performance indicators (KPIs) set by the EIC Board are mostly related to investment
performance, rather than innovation.63 This should change in favour of more
appropriate indicators.
Given the diversity of projects funded, it is unlikely that a one-fits-all approach

to monitoring and evaluation can be successfully applied. For this reason, each
individual project should be evaluated on the basis of milestones established by
the programme manager before starting the programme, and possibly adjusted
during its course.64
Nonetheless, there are two general principles that can be borrowed from the
transatlantic agencies. First, ARPA-funded projects are meant to have the potential
of being transformational. Box 9 presents a version of the “Heilmeier” questions
used by ARPAs to assess whether a project fulfils this criterion.
62 As shown in Appendix Figure 18, most of the funding of Pillar II already goes to basic and applied research.
63 See the annex of the EIC impact report 2023.
64 Examples of evaluation reports used by DARPA programme managers can be found here.
44
The second criterion is that a project should fill technological white space, namely
address a perceived gap or opportunity in the technological landscape. Importantly,
the latter should be evaluated in terms of other EU funding schemes and private
companies to avoid duplication and redundancies. To that extent, the EIC should
carefully monitor the emergence of new science and technology, and then issue
calls for projects on specific themes it deems promising. However, such a scientific
and technological watch requires the involvement of top scientists, which again
raises the issue of their role in the overall funding scheme.65
The qualitative approach outlined above should be complemented by an analysis of

key metrics, such as: i) scientific publications and their forward citations; ii) patents
and their forward citations; iii) follow-on funding, and iv) new firm foundation.
Clearly, these metrics are not always a good measure of success. For instance,
publications might be more suitable to evaluate funding to academic institutions;
(quality-weighted) patents might be better measures for small companies than
large ones, which might use results internally, and not all projects might result in
startup formation. Moreover, a sufficiently long amount of time might be needed
to properly evaluate the results of a project. Nevertheless, making available a
dashboard with such measures can help forming a holistic view on the progress of
a funded project.
Importantly, project-specific metrics should be compared against other similar

projects – possibly from other EU financing schemes. This would allow to evaluate
the effectiveness of ARPA-style programme management relative to more
traditional programmes.
Case studies can complement the evaluation. One possible way, inspired by
ARPA-E, is to form a portfolio of projects representing a mix of different recipients
(e.g. academia vs private companies) and financing mode (open calls vs thematic
calls). Beneficiaries can be asked about the status of their research before receiving
funding, so to be able to form a “pre-post” type of analysis.
Appendix Figure 23 shows an example based on ARPA-E independent evaluation

from the National Academies of Sciences (2017). In that example, Harvard University
received an ARPA-E award of $2.750 million in 2013. A year and a half into the
project, it raised $3 million in venture capital and in October 2014, the start-up
company “SLIPS Technologies, Inc.” was founded.
Data Sharing and Transparency

Finally, data on EU funded projects, as well as rejected applications, should be
openly available to researchers interested in conducting evaluation exercises. The
European Commission has already undertaken important steps in this direction by
publishing the CORDIS dataset. However, the information provided is not always
consistent across datasets and more granular information could be provided
without undermining privacy concerns.
65 In the United States, the “Fast Track Action Subcommittee on Critical and Emerging Technologies” is tasked with
identifying critical and emerging technologies. See the report here.
45
BOX VIII
EU vs national funding to innovation
A closer look at the data shows that the EIC represents an important part of the overall (national and EU-level)
funding of breakthrough innovations. This box also argues that an EU-wide selection mechanism has important
advantages over national initiatives.
Public spending on R&D in the European Union is dominated by Member States. The EU level (mostly through
its Horizon Europe programme with an annual budget of about €11 billion) accounts for roughly 1/10 of the
combined public R&D expenditure by Member states. However, Horizon Europe – and the EIC in particular – are
much more important in the area of funding breakthrough research.
The budgets of some of the most important national agencies with the official goal of funding transformative
innovation appear quite modest. We focus on the three biggest Member States. Examples include in Germany
SPRIN-D and the Agency for Innovation and Cyber Security (Cyberagentur), with respective budgets of €150
million and €70 million per year. In France, the Defense Innovation Agency (AID) has a yearly budget of around
€1 billion, but little of this seems to be spent on DARPA-style projects. In Italy, ENEA has a yearly budget of
roughly €150 millions, which is spent mostly on energy efficiency. The Franco-German Joint European Disruptive
Initiative (JEDI) acts as a network, but does not have its own funding sources as shown by its balance sheet.
The combined total available from national sources for low-TRL projects is thus of a similar order of magnitude
as the EIC. The EIC, especially Pathfinder and Transition are thus an important part of the overall funding for
breakthrough research in Europe.
There are a number of advantages with funding breakthrough research at the European, rather than national
level: First, it expands the pool and diversity of applicants and evaluators. Second, the fact that ambitious R&D is
bound to fail with non-negligible probability means that it is hard to conduct advanced research without risking
a mediatic backlash if only few projects can be funded and several of them fail. The law of large numbers at the
EU level makes this occurrence less likely. Finally, and most importantly, a European-level funding increases the
physical, cultural, and social distance between the reviewers and those who originate the projects. This reduces
the danger of cronyism and political influence in the selection process.
46
BOX IX
Is the project potentially transformational?
■ Does the programme/project hold potential to create “forks” in the technology roadmap that could change
the conventional understanding of what is possible, practical, and profitable in applications with large
potential impact?
■ Does it reach for impacts that may be very far off but that may eventually result in faster learning curves and
potentially disruptive entitlements?
■ Does it bridge gaps in technological development that might otherwise prevent a particular research pathway
from being advanced?
■ Does it challenge a community to tackle barriers seen as too risky for rate of return–based investment decision
metrics?
■ Does it empower a community to challenge conventional wisdom regarding feasibility?
■ Does it shake up “group think” and political or industrial establishments?
■ Does it set aggressive targets that, if attained, could result in high impact?
■ Does it create a stakeholder or technical constituency?
■ Does it revisit a failed approach in the context of related supporting advances or changes in economic or
other external factors?
■ Does it fuse disparate technologies in novel ways?
47
V.
SUMMARY
AND CONCLUSIONS
48
V. SUMMARY AND CONCLUSIONS
T he first part of the report documents the reasons for the transatlantic
differences in business R&D. We argue that a key reason is that the
European industry is skewed towards mid-tech sectors, with a middle
level of R&D-intensity. This specialisation in mid-tech has persisted for decades,
even increasing over the recent past, whereas the weight of different sectors has
profoundly changed in the US. Possibly due to path dependency, the EU is thus
failing to catch up in high-tech industries and with the AI and biotech revolutions.
Moving away from the status quo requires public-sector intervention to encourage
research in new technologies and to promote the development of EU high-tech
industries.
The second part of the report examines the main body set up by the EU to boost
breakthrough innovation: the European Innovation Council (EIC). We argue that in
order for the EIC to be more effective and compete with its benchmark models
on the other side of the Atlantic, it needs, first of all, a substantial restructuring
of its governance. Specifically, the EIC needs less political decision-making, more
highly qualified scientists and engineers on the EIC Board, and more programme
managers who are experts in the fields of projects they manage. Programme
managers should also be given more decision-making power. Top scientists and
engineers could become involved in a redesigned, two-stage evaluation process.
Moreover, a large part of the EIC’s budget currently goes to high-TRL activities
and therefore does not support breakthrough research, but rather the upscaling of
technologies by SMEs. In principle, the EIC was meant to support startups (i.e. firms
no older than 4 years), but few of them are beneficiaries. This applies in particular
to the equity investments by the Accelerator programme, which tend to support
mature enterprises instead of breakthrough innovation.
The third part of the report illustrates the plethora of funding instruments in Horizon
Europe, the flagship EU programme for R&D, and two formally independent
initiatives, the European Innovation Fund and the European Defence Fund. We
argue that while there are a number of funding sources nominally devoted to
innovation, de facto most of this funding is earmarked for projects that are close to
being commercially viable. These resources, including those currently allocated to
the European Institute of Technology and European Innovation Ecosystem, could
have a greater impact on innovation if they were transferred to the (reformed) EIC,
or at least reassigned to the task of supporting ground-breaking innovation and
managed accordingly.
Finally, the last part of the report formulates our policy proposal for a budget-
neutral but radical restructuring of the EU ecosystem for innovation. If successful,
the reform would enable the EU to escape its middle technology trap by supporting
innovation, thus creating the basis for private investment in new sectors with higher
growth potential. The aim is not to channel more resources into specific areas, but
rather to create more space for innovation and economic dynamism across the
board.
49
REFERENCES
Acemoglu, Daron. ‘Distorted innovation: does the market get the direction of
technology right?’ AEA Papers and Proceedings, American Economic Association
113 (2023).
Adda, Jérôme, and Marco Ottaviani. “Grantmaking, grading on a curve, and the
paradox of relative evaluation in nonmarkets.” The Quarterly Journal of Economics
(2023): qjad046.
Aghion et al. ‘AI: Notre Ambition pour la France’ (March 2024). Commission de
l’intelligence artificielle, presided by Philippe Aghion and Anne Bouverot.
Aghion, Philippe, Céline Antonin, and Simon Bunel. The power of creative
destruction: Economic upheaval and the wealth of nations. Harvard University Press
(2021).
Aghion, Philippe, et al. ‘Carbon taxes, path dependency, and directed technical
change: Evidence from the auto industry.’ Journal of Political Economy 124.1 (2016):
1-51.
Akcigit, Ufuk, and William R. Kerr. ‘Growth through heterogeneous innovations.’

Journal of Political Economy 126.4 (2018): 1374-1443.
Akcigit, Ufuk, Douglas Hanley, and Stefanie Stantcheva. ‘Optimal taxation and R&D
policies.’ Econometrica 90.2 (2022): 645-684.
Akcigit, Ufuk. ‘Economic Renaissance: Unleashing Business Dynamism for Economic

Growth.’ (2024).
Amoroso, S., Aristodemou, L., Criscuolo, C., Dechezleprêtre, A., Dernis, H.,
Grassano, N., Moussiegt, L., Napolitano, L., Nawa, D. Squicciarini, M., and Tübke,
A. World Corporate Top R&D investors: Paving the way for climate neutrality. A
joint JRC and OECD report. EUR 30884 EN, Publications Office of the European
Union, Luxembourg (2021), doi:10.2760/49552, JRC126788.
Azoulay, P., Fuchs, E., Goldstein, A. P., & Kearney, M. ‘Funding breakthrough
research: Promises and challenges of the “ARPA Model”.’ Innovation policy and the
economy 19.1 (2019): 69-96.
Bartelsman, Eric J., Pieter A. Gautier, and Joris De Wind. ‘Employment protection,
technology choice, and worker allocation.’ International Economic Review 57.3
(2016): 787-826.
50
Becker, Gary S., and Kevin M. Murphy. “The division of labor, coordination costs,
and knowledge.” The Quarterly journal of economics 107.4 (1992): 1137-1160.
Beebe, Barton, and Jeanne C. Fromer. ‘Fake trademark specimens.’ Columbia Law
Review 120.7 (2020): 217-249.
Biondi, Filippo, et al. Declining business dynamism in Europe: The role of shocks,
market power, and technology. No. 2023-011. Jena Economic Research Papers
(2023).
Board, Simon, Moritz Meyer-ter-Vehn, and Tomasz Sadzik. Recruiting talent. UCLA
working paper, 2017.
Branstetter, Lee G., and Mariko Sakakibara. “When do research consortia work well
and why? Evidence from Japanese panel data.” American Economic Review 92.1
(2002): 143-159.
Breschi, Stefano, Nick Johnstone, and Carlo Menon. ‘Are start-ups funded by public
venture capital different? New cross-country evidence from micro-data.’’ Industrial
and Corporate Change 30.6 (2021): 1615-1632.
Bronzini, R. and Iachini, E. ‘Are incentives for R&D effective? Evidence from a
regression discontinuity approach’. American Economic Journal: Economic Policy
6.4 (2014): 100-134.
Cincera, Michele, and Reinhilde Veugelers. “Differences in the rates of return to

R&D for European and US young leading R&D firms.” Research policy 43.8 (2014):
1413-1421.
Coatanlem, Yann. ‘Why Europe Lags Behind in Tech.’ Financial Times (February 27,
2024).
Conclave Europe (2023) “Europe 2040: Tomorrow Is Today”, EuropaNova.
Decker, Ryan A., et al. ‘Changing business dynamism and productivity: Shocks
versus responsiveness.’ American Economic Review 110.12 (2020): 3952-3990.
Enriques, Luca, Casimiro A. Nigro and Tobias H. Tröger. ‘Venture Capital Contracting
in Continental Europe: Bargaining in the Shadow of Corporate Law Constraints.’
Working paper (2024).
Foster, Richard N. ‘Working the S-curve: assessing technological threats.’ Research

Management 29.4 (1986): 17-20.
51
Gans, Joshua and Stern, Scott (2003) “The Product Market and the Market for
‘Ideas’: Commercialization Strategies for Technology Entrepreneurs.” Research
Policy 32 (2): 333–50
Ghirelli, Corinna, et al. “The long-term causal effects of winning an ERC grant.”
(2023).
Grassano, N., Hernandez Guevara, H., Fako, P., Nindl, E., Georgakaki, A., Ince,
E., Napolitano, L., Rentocchini, F. and Tubke, A. ‘The 2022 EU Industrial R&D
Investment Scoreboard.’ Publications Office of the European Union, Luxembourg
(2022). doi:10.2760/485748, JRC132035.
Howell, Sabrina T. ‘Financing Innovation: Evidence from R&D Grants.’ American

Economic Review 107.4 (2017): 1136-64.
Howell, Sabrina T., et al. ‘Opening up Military Innovation: Causal Effects of Reforms
to US Defense Research.’ No. w28700. National Bureau of Economic Research
(2021).
Lerner, Josh (2000) “The Government as Venture Capitalist: The Long-Run Effects
of the SBIR Program.” Journal of Private Equity 3 (2): 55–78.
Marino, Marianna, et al. ‘‘Additionality or crowding-out? An overall evaluation of

public R&D subsidy on private R&D expenditure.’’ Research Policy 45.9 (2016):
1715-1730.
Mazzucato, Mariana. “Mission-oriented innovation policies: challenges and

opportunities.” Industrial and corporate change 27.5 (2018): 803-815.
Nagar, Jay Prakash, Stefano Breschi, and Andrea Fosfuri. “From Academia to
Invention: Decoding the European Paradox Through ERC Science.” Available at
SSRN 4635463 (2023).
National Academies of Sciences, Engineering, and Medicine. 2017. An Assessment

of ARPA-E. Washington, DC: The National Academies Press. https://doi.
org/10.17226/24778.
Pallante, Gianluca, Emanuele Russo, and Andrea Roventini. ‘Does mission-oriented

funding stimulate private R&D? Evidence from military R&D for US states.’ No.
2020/32. LEM Working Paper Series (2021).
52
Santoleri, Pietro, et al. ‘The causal effects of R&D grants: Evidence from a regression
discontinuity’. Review of Economics and Statistics (2022): 1-42.
Sutton, J. ‘Market structure: theory and evidence.’ Handbook of industrial

organization. 3 (2007): 2301-2368.
Szücs, Florian. ‘Do research subsidies crowd out private R&D of large firms?
Evidence from European Framework Programmes.’ Research Policy 49.3 (2020):
103923.
Teichgraeber, Andreas, and John Van Reenen. ‘A policy toolkit to increase research
and innovation in the European Union.’ (2022).
Tirole, Jean. Economics for the Common Good. Princeton University Press (2017).
United States Patents and Trademark Office (USPTO). Trademarks and patents in
China: The impact of non-market factors on filing trends and IP systems (January
2021).
Veugelers, Reinhilde. Research and innovation policies and productivity growth.

No. 08/2021. Bruegel Working Paper, 2021.
Wu, Lingfei, Dashun Wang, and James A. Evans. “Large teams develop and small
teams disrupt science and technology.” Nature 566.7744 (2019): 378-382.
53
APPENDIX
Figure 11.
EU R&D funding and per-capita GDP
EU R&D funding
0,25
BE
0,20 CY
EU R&D funds as % of GDP
0,15
EL
EE
NL
0,10
SI
FI
PT ES
MT AT DK
0,05 LV SE
IT FR DE
BG HR HU LTCZ
SK RO PL
0,00
50 70 90 110 130 150
GDP per capita at PPP as % of EU average
Source: authors’ calculations based on Eurostat https://ec.europa.eu/eurostat/statistics-explained/index.php?title=R%26D_

expenditure&oldid=627002.
Figure 12.
(a) Electrical, semiconductors & audi-visual technologies (b) Transport and special machines
Number, international patent applications (PCT) Number, international patent applications (PCT)
20.000 USA Japan EU27 7.500 USA Japan EU27
15.000
5.000
10.000
2.500
5.000
0 0
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
2023
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
2023
(c) Computer and digital technologies (d) Pharmaceuticals, Medical- and Biotechnologies
Number, international patent applications (PCT) Number, international patent applications (PCT)
USA Japan EU27 17.500 USA Japan EU27

16.000
14.000 15.000
12.000 12.500
10.000 10.000
8.000
7.500
6.000
5.000
4.000
2.000 2.500
0 0
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
2023
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
2023
Source: WIPO.
54
Figure 13.
Capital expenditure by technology level 2022 (Top 2,500 companies)
Capital Expenditure by Tech−level 2022 (Top 2500 companies)
400
300
CapEx (in billion €)
200
100
0
EU US China Japan ROW
Other High−tech Software & Computer Services Pharmaceuticals & Biotechnology

Other Mid−tech Automobiles & Parts Other
Net sales by technology level (billion €)

Figure 14. Net sales by technology level (billion €)
Net sales by technology level (billion €)
EU US
EU 4000 High-tec US Mid-tech Other
4000 High-tec Mid-tech Other
3000 3000
3000 3000
2000 2000
2000 2000
1000 1000
1000 1000
0 0
0 2003 2008 2013 2018 2023 0 2003 2008 2013 2018 2023
2003 2008 Year
2013 2018 2023 2003 2008 Year
2013 2018 2023
Year Year
Japan China
Japan China
High-tec Mid-tech Other 4000 High-tec Mid-tech Other
4000
High-tec Mid-tech Other 4000 High-tec Mid-tech Other
4000
3000
3000
3000
3000
2000 2000
2000 2000
1000 1000
1000 1000
0 0
0 2003 2008 2013 2018 2023 0 2003 2008 2013 2018 2023
Year
2003 2008
2013 2018 2023 2003 2008 Year
2013 2018 2023
Year Year
55
Figure 15.
Profit margins by technology
Profits level (billion €)
by High−/Mid−tech (billion ?)
EU US
600 600
400 400
200 High−tech High−tech

200
Other Other
0 0
2003 2008 2013 2018 2023 2003 2008 2013 2018 2023
Year Year
Japan China
600 600
400 400
200 High−tech 200 High−tech

Other Other
0 0
2003 2008 2013 2018 2023 2003 2008 2013 2018 2023
Year Year
Figure 16.
Profits margins by technology
Profits margin level (%
by technlogy of sales,
level (% of 2003−2022)
sales, 2003−2022)
15
10
0
EU US China Japan
High−tech Mid−tech Other
Note: sector profit margin is the ratio of profits to sales. Average for the years 2003-2022.
56
Figure 17.
Country share of total international BERD
Software & Computer Services

2003 2012 2022
7% 10% 6%
12% 18%
21%
72%
78% 77%
1 EU 1 EU 1 EU
2 Other Regions 2 Other Regions 2 Other Regions
3 US 3 US 3 US
Automobiles & Parts

2003 2012 2022
17% 19%
25%
42% 45% 45%

20%
30%
31% 1% 8% 17%
1 EU 1 EU 1 EU
2 Other Regions 2 Other Regions 2 Other Regions
3 Japan 3 Japan 3 Japan
4 USA 4 USA 4 USA
Note: BERD based on sample of global 500 largest companies.

Source: Industrial R&D Investment Scoreboard (2004, 2013, 2022).
57
Figure 18.
Revenue of Horizon Europeofcompanies
Revenue by programme
Horizon Europe companies by program
Pathfinder (TRL 1-4) Transition (TRL 3-6) Accelerator (TRL 5-9)
80 Revenue of Horizon Europe companies by program80 80
% of companies
% of companies
% of companies
Pathfinder (TRL 1-4) Transition (TRL 3-6) Accelerator (TRL 5-9)
60 60 60
80 80 80
% of companies
% of companies
% of companies
40 40 40
60 60 60
20 20 20
40 40 40
0 0 0
20 20 20
Under $1M $1M-$10M $10M-$100M $100M+ Under $1M $1M-$10M $10M-$100M $100M+ Under $1M $1M-$10M $10M-$100M $100M+
Revenue (US$) Revenue (US$) Revenue (US$)

0
Under $1M $1M-$10M $10M-$100M $100M+
Employees
0 of Horizon Europe companies by program
Under $1M $1M-$10M $10M-$100M $100M+
0
Under $1M $1M-$10M $10M-$100M $100M+
Pathfinder (TRL 1-4)
Revenue (US$) Transition (TRL 3-6)
Revenue (US$) Accelerator (TRL 5-9)
Revenue (US$)
Employees of Horizon Europe companies by programme
50 50 50
Employees of Horizon Europe companies by program
% of companies
% of companies
% of companies
40 Pathfinder (TRL 1-4) 40 Transition (TRL 3-6) 40 Accelerator (TRL 5-9)
50 50 50
30 30 30
% of companies
% of companies
% of companies
40 40 40
20 20 20
30 30 30
10 10 10
20 20 20
0 0 0
1-11 11-50 51-100 101-250 251-500 500+ 1-11 11-50 51-100 101-250 251-500 500+ 1-11 11-50 51-100 101-250 251-500 500+
10 10 10
Employees Employees Employees
0 0 0
1-11 11-50 51-100 101-250 251-500 500+ 1-11 11-50 51-100 101-250 251-500 500+ 1-11 11-50 51-100 101-250 251-500 500+
Employees Employees Employees

Source: Crunchbase.
Figure 19.
400
300
Number of projects
200
100
0
n
e
ty
g
gy
n
er
s
l
re
T
gy
g
io
AI
na
ar
ar
at
ig
ic
rin
ili
IC
in
th
er
at
Ca
on
dw
ftw
lo
es
io
b
er
tu
O
En
rt
na
no
ss
D
tr
ne
lth
ac
ar
po
So
ec
ai
fe
h
H
uf
gi
ea
ec
ns
st
o
El
an
En
Pr
Su
H
ot
a
er
Tr
M
Bi
d
um
an
ns
e
nc
Co
ie
Sc
Source: Crunchbase.
58
Figure 20.
EU funding by activity type (million €)
1. Health 2. Culture 3. Security

15
10
0
Basic & Applied Development Basic & Applied Development Basic & Applied Development
Research Research Research
4. Digital 5. Climate 6. Food

15
10
0
Basic & Applied Development Basic & Applied Development Basic & Applied Development
Research Research Research
Total (000s) Per-project

10
Source: authors’ calculations based on CORDIS.
Figure 21.
Cumulative Funding of EIT-KCIs
Cumulative funding of EIT 15
10
800
700
600
500
EUR million
400
Note:
within a sector t
300
200
100
0
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Climate Digital InnoEnergy Raw Materials
Health Food Manufacturing Urban Mobility
Source: own calculations based on https://eit.europa.eu/global-challenges/about-our-communities.
59
Figure 22.
Corporate venture capital investment by headquarter
region of the Scoreboard parent company
EU27_mother EU27_startup
21.96 % 8.96 %
US_mother
US_startup
42.67 %
81.24 %
JP_mother
7.30 %
CN_mother
1.37 % JP_startup
0.01 %
RoW_mother RoW_startup
26.70 % 10.12 %
Source: Grassano et al. (2022), page 30.
Figure 23.
Case study diagram
SLIPS Funding Timeline

6
ARPA-E Funding Quantity
Cumulative Funding
and Award Date

(millions of Dollars)
5
Project Period
4
3
Company
Founded
2
0
2013 2014 2015 2016
Source: National Academies of Science (2017).
Governance of European research/innovation bodies

The table below provides a synoptic overview of the governance characteristics of
key EU bodies tasked with promoting research and innovation. These bodies have
very different governance structures.
The main message that emerges from this table is that the governance of the ERC
is much more independent from the Commission than the EIC or the EIT. Moreover,
the academic qualifications of the scientific members of the respective boards vary
enormously.
60
Table 2.
Overview of main governance elements of EU research/innovation bodies
European Research Council European Innovation Council European Institute of Innovation

and Technology
President’s qualifications Senior and internationally A high-profile public figure Governing Board shall elect its
respected scientist linked to the world of innovation Chairperson from among its
members
Appointment of Appointed by the Commission, Appointed by the Commission Appointed by the Commission
the president which selects one among
three names proposed by the
Scientific Council66
Board members’ The Scientific Council is Appointed by the Commission Appointed by the Commission
appointment appointed by the Commission
under the recommendation of
an Independent Identification
Committee
Qualifications required Independent scientists, Entrepreneurs, corporate Proven experience in the fields
engineers and scholars of the leaders, investors, public of higher education, research,
highest repute administration experts and innovation or business
researchers, including academic
experts on innovation
Scientific/technical All 22 members are professors 5 out of 21 members are 2 out of 14 members are
qualifications of the Board/ and in the Web of Science, with professors, 11 have a PhD, professors, 8 have a PhD, of
Scientific Council an average h-index of 58 and of whom 7 are in the Web whom 5 are in the Web of
average citations of over 18,000. of Science with an average Science with an average h-index
h-index of 18 and average of 25 and average citations of
citations of 2,000. 8,000.
Term 4 years, renewable once 2 years, renewable twice 4 years, extendable by the
Commission once by 2 years on a
proposal of the Governing Board
Independence Independent of extraneous Not required Not required

interests
Tasks/competences (a) overall strategy for the ERC Advise the Commission Provide strategic guidance and
(b) the work programme for the Upon request address select, evaluate, and support
implementation of the ERC’s recommendations to the the Knowledge and Innovation
activities Commission Communities
(c) the methods and procedures
for peer review and proposal
evaluation
Work programme Established by the ERC Scientific Commission decision, following Strategic Innovation Agenda set
Council the advice of the EIC Board by EU regulation for the 7-year
period
66 The Chair is appointed by the European Commission to a position as Special Advisor to the Commission, after submission of the dossiers of the three finalists in the
selection made by an independent committee to the Scientific Council, which can only issue a veto.
Note: The legal framework of EIT has been modified since its creation in 2008, in particular in 2021.
See https://eur-lex.europa.eu/EN/legal-content/summary/european-institute-of-innovation-and-technology-eit.html
Sources. Own evaluation on the basis of Publications Office (europa.eu) for the EIC and ERC;
for the EIT, see https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:L:2021:189:FULL
61

Middle Technology Trap

Uploaded by

Copyright:

Available Formats

Middle Technology Trap

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Middle Technology Trap

Uploaded by

Copyright:

Available Formats

Economics for the

A Report by the European

We propose an ARPA-style model of governance and a budget-neutral shift of resources to

We wish to thank, without implicating them in our conclusions, Philippe

@2024 econpol@cesifo, IEP@BU Institute for European Policymaking@Bocconi University, and

I. INNOVATION SPENDING COMPARED:

II. CAN THE EUROPEAN INNOVATION COUNCIL

III. ANATOMY OF HORIZON EUROPE 26

IV. A BUDGET-NEUTRAL POLICY PROPOSAL

V. SUMMARY AND CONCLUSIONS 39

THE MIDDLE TECHNOLOGY TRAP

BERD by Tech−level 2022 (Top 2500 companies)

Other High−tech Software & Computer Services Pharmaceuticals & Biotechnology

Source: Industrial R&D Investment Scoreboard (2023).

R&D Intensity by technlogy level (% of sales)

Source: Industrial R&D Investment Scoreboard.

Clearly, specialising in mid-tech industries limits growth opportunities. While there

PATENTS AS A MEASURE OF INNOVATION OUTPUT

In 2022, around 270,000 Patent Cooperation Treaty (PCT) applications were

Figure 4 shows that each region tends to specialise in a different technological

Furthermore, the specialisation of the EU in mobility-related technologies has

2003 2012 2022

US Ford (auto) Microsoft (software) Alphabet (software)

EU Mercedes-Benz (auto) VW (auto) VW (auto)

JPN Toyota (auto) Toyota (auto) Toyota (auto)

Source: Industrial R&D Investment Scoreboard (2004, 2013 and 2023).

A successful EU innovation policy should break the path dependency

In the US, spurring breakthrough innovation is a key objective of the Advanced

A key characteristic of ARPAs is their mission-oriented approach to innovation.20

THE ARPA MODEL AND THE EIC

9 ACTUAL SYSTEM PROVEN

8 SYSTEM COMPLETE AND QUALIFIED

4 TECHNOLOGY VALIDATED IN LAB

Pathfinder & ARPA

2 TECHNOLOGY CONCEPT FORMULATED

1 BASIC PRINCIPLES OBSERVED

Sources: authors’ representation based on official sources.

1. Identification of the market failure so as to design the proper policy;

GOVERNANCE OF EU INNOVATION POLICY (EIC)

The governance structure of the EIC is summarised in Figure 6.

BOARD Share of company X

Single Market Programme

25 In fact, while referred to as the ‘Council’, in reality no such council exists.

WHO RECEIVES SUPPORT FROM EIC PROGRAMMES?

Sources: authors’ calculations based on CORDIS and Crunchbase.

It is an open question whether an innovation model so focused on SMEs will be

ROLE OF PROGRAMME MANAGERS

39 The data are available here.

Pillar I Pillar II Pillar III

It is therefore no surprise that, like its US counterparts the National Science

The first is that it adopts a top-down approach by eliciting applications on specific

The second feature is that it funds collaborative projects by imposing consortia

Pillar III - Innovative Europe

Source: authors’ calculations based on CORDIS.

Appendix Figure 20 shows total and per-project funding by thematic cluster,

Finally, the governance of Pillar II is handicapped by its political process. Not

European Institute of Innovation & Technology

European Innovation Ecosystem

EIC Accelerator and EIC Fund

Following selection, and conditional on the rubberstamping of the decision by