Content uploaded by Daniele Archibugi
Author content
All content in this area was uploaded by Daniele Archibugi on Apr 30, 2019
Content may be subject to copyright.
CIMR Research Working Paper Series
Working Paper No. 5
The Impact of the Economic Crisis on Innovation:
Evidence from Europe
by
Daniele Archibugi
Italian National Research Council
Via Palestro, 32- 00185 Rome, Italy
Tel. +39-06 492724241
Email: daniele.archibugi@cnr.it
Birkbeck University of London
Department of Management and Centre for
Innovation Management Research
Malet Street, Bloomsbury, London WC1E 7HX
Tel. +44(0)20 7631 6741
Email: d.archibugi@bbk.ac.uk
Andrea Filippetti
Italian National Research Council
Institute for the Study of Regionalism, Federalism and Self-Government
Via dei Taurini, 19 - 00185 Rome, Italy
Tel: +39 0649937704
Email: andrea.filippetti@cnr.it
Marion Frenz
Birkbeck University of London
Department of Management and Centre for Innovation Management Research
Malet Street, Bloomsbury, London WC1E 7HX, UK
Tel. +44(0)20 7631 6829
Email: m.frenz@bbk.ac.uk
April 2012
ISSN 2052-062X
2
The Impact of the Economic Crisis on Innovation:
Evidence from Europe
Daniele Archibugi*§, Andrea Filippetti°, and Marion Frenz§
*Italian National Research Council – CNR – IRPPS
° Italian National Research Council – CNR - ISSIRFA
§Birkbeck College – University of London
Abstract
Economic crises cause companies to reduce their investment, including investment in
innovation where returns are uncertain and long-term. This has been confirmed by the
2008 financial crisis, which has substantially reduced the willingness of firms to invest
in innovation. However, the reduction in investment has not been uniform across
companies and a few even increased their innovation expenditures. Through the analysis
of a fresh European Survey, this paper compares drivers of innovation investment
before, during and following on from the crisis, applying the Schumpeterian hypotheses
of creative destruction and technological accumulation. Before the crisis, incumbent
enterprises are more likely to expand their innovation investment, while after the crisis a
few, small enterprises and new entrants are ready to by
expanding their innovative related expenditures.
Keywords: Economic crisis, innovation investment, Innobarometer
JEL: O12, O30, O33, O52
3
1. The effect of an economic shock on long-term investment
Major economic shocks, such as the 2008 financial crisis, make business opportunities
less certain, and, in turn, companies become less willing to invest in long-term activities
where returns are risky. Most companies react to a short- or medium-term adverse
macroeconomic environment by downsizing expenditures, including expenditures on
investment and innovation. However, economic crises also provide an opportunity for
companies, industries and entire nations to restructure productive facilities and to
explore new opportunities. Smart companies do perceive that an economic crisis will
not last forever and that a recovery will sooner or later arrive. A new economic cycle,
however, is also likely to bring structural changes in the composition of output and
demand. In order to reap benefits from opportunities in changing economic
environments, successful companies need to be prepared to provide new and improved
goods and services.
As already predicted by Schumpeter and the Schumpeterian economics, while an
economic crisis has an adverse impact on most of the economic agents, in the long-run
it will not generate losers only. On the one hand, a few economic agents may emerge as
winners and we assume that they will be found among those companies that understand
earlier than others that the composition of output and relative prices to emerge from the
crisis will be very different from the past. On the other hand, losers are more likely to be
found among those firms that react not just by reducing employment and productive
capacity in general, but also downsizing their investment in innovation. Which are the
key characteristics of the companies belonging to the two categories?
The 2008 economic crisis offers a unique opportunity to test two models of
innovation originating from Schumpeter and the Schumpeterian economics and that can
be labelled creative destruction and technological accumulation. In turn, these models
may help us to identify what will be the typology of companies that will lead the
recovery. Our paper is an attempt to test the interplay between the forces of creative
destruction and accumulation in innovation before, during and after the financial crisis
that started in the Fall of 2008. In fact, there was in Europe a substantial drop of
innovative investment (Filippetti and Archibugi, 2011), and this leads to wonder what
are the best strategies that should be taken at the country level (see Sharif, 2012).
Our analysis is made possible thanks to a recent wave of the Innobarometer
Survey designed and collected by the European Commission in 2009 (European
Commission, 2009). Each year the Innobarometer introduces a different topic and the
2009 survey emphasises innovation related expenditure, including the effects on it of
the economic downturn. Enterprises from the 27 EU member states, plus Norway and
Switzerland responded to the survey.
The paper is structured as follows. Section 2 discusses the state of the art against
which the paper is set. Section 3 develops the conceptual framework by providing a
sketch of the two ideal type models of creative accumulation and creative destruction.
Section 4 introduces the dataset and methodology. Section 5 presents the results that are
discusses in the last Section.
4
2. Innovation generated through technological accumulation and economic
creative destruction
The young Schumpeter (1911) looked at innovation as an event that could revolutionize
economic life by bringing into the fore new entrepreneurs, new companies and new
industries. The mature Schumpeter (1942), on the contrary, observed and described the
activities of large oligopolistic corporations, able to perform R&D and innovation as a
routine by building on their previous competences. The relative importance of these two
processes has been further investigating in the Schumpeterian tradition (see Nelson and
Winter, 1982; Breschi et al., 2000; Malerba and Orsenigo, 1995; Patel and Pavitt, 1994).
Creative destruction is described as a result of a regime characterized by low
cumulativeness and high technological opportunities, leading to an environment with
greater dynamism in terms of technological ease of entry and exit, as well as a major
role played by entrepreneurs and fierce competition. Creative accumulation is
associated with a technological regime that is characterized by high cumulativeness and
low technological opportunities, bringing about more stable environments in which the
bulk of innovation is carried out by large and established firms incrementally, leading to
a market structure with high entry barriers and oligopolistic competition.
There are arguments supporting the relevance of cumulativeness and of
reinforcing patterns of technological development and innovation, and arguments
lending support to a
several studies suggest that learning processes that underlie innovation activities are
both local and cumulative resulting in path-dependency (e.g. Pavitt et al., 1989;
Antonelli, 1997; Pavitt, 2005). In addition, empirical evidence indicates that there is a
degree of persistence in innovation and among innovators (Cefis and Orsenigo, 2001).
Concerning the latter, it has often been stressed that there are periods of turbulence
associated with a change in the leading sectors and/or the emergence of new sectors,
which brings about a decline of technological and profit opportunities in established
industries (Perez, 2002). This, in turn, might lead to a change in the knowledge and
technological base for innovation and could substantially affect the hierarchy of
innovators (Devez et al., 2005). Other research has stressed the fact that firm-specific
organizational routines and capabilities can bring about inertia and hamper the capacity
of established firms to keep up with major discontinuities (Henderson and Clark, 1990;
Leonard-Barton, 1992; Levinthal and March, 1993).
(1977) and his followers on the grounds of the fact that the population of incumbent,
large firms has remained stable over the last decades. This thesis has been challenged by
Simonetti (1996), Louca and Mendonca (1999), and by Freeman and Louca (2001), who
claim that a stream of new firms has joined incumbent firms during periods of radical
discontinuities. This can also be contingent to the specific knowledge base and technical
skills attached to different industries. For example, while Klepper and Simons (2000)
show that firms established in making radios were successful in developing colour TVs,
Holbrock et al. (2000) illustrate that this pattern is not mirrored in the evolution of the
semiconductor industry.
In this paper the emphasis is not on specific industries or technologies, but rather
innovative strategies. As a result, we expect to find an array of different innovation
drivers both before and in response to the crisis. These are examined in view of the
5
changes at the macro level, as we aim to understand whether the crisis has led to some
variation/discontinuity at the aggregate level as a result of a different composition
among innovating firms.
3. An attempt to identify the core characteristics of creative destruction and
technological accumulation
To guide the analysis we elaborate on the ideal type models of creative destruction and
creative accumulation as two possible aggregate outcomes of micro behaviours.
Creative destruction describes a dynamic environment in which new firms emerge as
the most significant innovators as a result of a major discontinuity such as an economic
downturn. Creative accumulation is underpinned by a more stable pattern of innovation
which emphasizes cumulativeness and persistency of innovative activities in response to
the crisis. We make here an attempt to identify these two patterns in relation to firm
behaviour rather than to the evolution of technological regimes. In this sense, our
approach is complementary to the research pioneered by Malerba and Orsenigo (1995)
to identify Schumpeterian patterns of innovation with reference to various technological
fields. A sketch of the differences between the models of creative destruction and
creative accumulation is given in Table 1 where four categories are singled out: i)
characteristics of the innovating firm, ii) type of knowledge source dominant in the
innovation process, iii) type of innovations, and iv) characteristics of the market.
Table 1 Innovative firms characteristics in the context of the ideal type creative
accumulation and creative destruction models
Categories
Creative accumulation
Creative destruction
Characteristics of the
innovating firms
Innovations are driven by
large, incumbent firms that
seek new solutions through
formal research exploiting
their pre-existing capability.
Small firms, new entrants are
key drivers in the innovation
process. They use innovations
and exploit economic
turbulences to acquire market
share from incumbent firms or
to open new markets.
Type of knowledge
sources
High relevance of past
innovations and accumulated
knowledge. Importance of
formal R&D, in-house, but
also jointly performed, or
externally acquired.
Higher relevance of
collaborative arrangements
leaning towards the applied
knowledge base (other firms).
Exploration of new markets
and technological
opportunities.
6
Type of innovations
The innovation process is
dominated by a large number
of incremental innovations.
Organizational routines drive
the generation of innovations.
The emphasis is on path-
breaking innovations often
able to create new industries.
New organizational forms
contribute to generating
innovations.
Characteristics of the
market
Barriers to entry are high due
to relative importance of
appropriation and
cumulativeness of knowledge
and high costs of innovation.
Dominance of oligopolistic
markets. Technological
advancement based on path-
dependent and cumulative
technological trajectories.
Low barriers to entry into the
newly emerging industries. A
high rate of entry and exit
leads to low levels of
concentration and high
competition. Discontinuous
technologies are available that
generate growing markets and
new opportunities.
In the empirical part of the paper some of these factors, those more directly associated
to our data, will be used to test if the two ideal type models can be related to the patterns
of innovation investment of firms.
3.1 Characteristics of the innovating firms
The creative accumulation model assumes that incumbent firms explore systematically
technological opportunities. For them, to innovate is a routine, and it is one of the core
things that the top management supervises. They have to upgrade periodically their
products, often because they operate in concentrated oligopolistic industries. A stream
of incremental innovation does not only guarantee that costs and prices are kept
competitive, but also that products are differentiated and improved compared to those of
the competition. This provides the possibility to accumulate knowledge and often not
just in the areas of their core products. When new technological opportunities are
identified, these companies may also be quick in entering into new fields and industries,
thanks to their wide, accumulated knowledge (Laperche et al., 2011). However, when
firms diversify, they tend to do so along some kind of technological relatedness, defined
as coherence (Piscitello, 2004; Teece et al., 1994). Pavitt makes this
the increasingly specialized and professional nature of the knowledge on which they are
based, manufacturing firms are path-
convert a traditional textile firm into one making semicond
By contrast, the creative destruction model emphasizes the role played by
individual inventors and entrepreneurs. This model reflects a more uncertain landscape
of early stages of new technologies. By anticipating or even creating technological
opportunities, these far-sighted individuals manage to generate new firms and often new
industries that substantially change the economic landscape. These individuals can be
independent, e.g. setting up or owning their own business, but they can also be
dependent and employed by (sometimes large) organisation.
These individuals do not find the most conducive environment in existing
organizations since learned and accumulated routine activities, organizational settings,
and decision processes somehow discourage an entrepreneurial stance. Moreover, the
larger the company, the greater might be a resistance to change by the company as a
7
whole. Thus, patterns linked to creative destruction are associated at the firm level with
innovation driven by smaller size, and new entry into markets alongside established
firms, as entrepreneurial activities might be greater due to lower inertia, greater
flexibility and responsiveness to changes in demand conditions and technological
discontinuities. This type of innovative behaviour could be found in spin-offs from
established companies, universities or simply new businesses.
3.2 Type of knowledge sources
In creative accumulation routine-based research is more important as a key source in the
innovation process than sudden insights. This favours the large firm that; i) has the
capacity and the resources to set-up and maintain internal R&D laboratories, ii) can use
interactions with others, and iii) has well-established internal functions (including
design, production, and marketing). High-tech companies are also able to plug into the
knowledge base of other companies, public institutions and countries. They are in the
position to reduce the risks and costs associated with exploring new technological
opportunities through strategic technological agreements, they have qualified personnel
able to interact periodically with universities and public research centres, they can also
establish intra-firm but international research networks through subsidiaries in other
countries (Laperche et al., 2011). All these factors allow them to build on and add to
their already existing competences.
Creative destruction on the contrary will be based on internal sources that in
economic activity, as it has happened for companies in emerging fields such as
biotechnology and software. This will also be combined to the concentric exploration of
new opportunities, to specific ventures with companies operating in other industries, or
generating symbiotic contacts with university departments (see Breschi et al., 2000). In
the case of small or newly established firms, the development of new products, services
or processes is likely to favour external collaborations and strategic alliances over and
above than in the case for large corporations. Such set-ups help to overcome possible
resource, finance and capability constraints within new and comparatively small firms.
3.3 Type of innovations
Creative destruction is linked to patterns of path-breaking innovations and radically new
solutions that are incompatible with traditional solutions. Several scholars have argued
that in this case innovations are more likely to be introduced by new firms, as existing
firms can face problems in terms of a lack of the adequate new skills and competences
(Tushman and Anderson, 1986; Henderson and Clark, 1990; Leonard-Barton 1992),
organizational adaptation (Levinthal and March, 1993), and difficulties in changing
context (Christensen and Rosenbloom, 1995; Christensen, 1997).
Creative accumulation is linked with frequent, but more incremental innovation
patterns. Accumulation or cumulativeness suggests that firms innovation activities are
driven by past innovation activities. Current technologies build on past experience of
production and innovation specific to the firm. Malerba and Orsenigo (1995) and
Breschi et al. (2000) suggest that cumulativeness of technological change is high when;
i) the firm is established and can build on a history of innovation success, ii) there is a
tradition of research carried out inside the firm.
Pavitt and his colleagues suggested that incumbents might have the resilience to
survive and to adapt to major changes (Pavitt et al., 1989, Patel and Pavitt, 1994).
8
Methé et al. (1996) present empirical evidence showing that established firms often are
sources of major innovations, for example in telecommunications and medical
instruments. In a similar vein, Iansiti and Levien (2004) suggest that, despite the many
industry were survived by the overwhelming majority of firms. Studying a sample of
large French firms, Laperche et al. (2011) also show how they have quickly modified
their innovative strategies to face the post-crisis context.
3.4 Characteristics of the market
In a Schumpeterian model, firms compete to become oligopolistic in their market. This
allows them to gain extra profits through the appropriation of returns from their
innovations. In a dynamic context, the oligopolistic structure is seen as a necessary evil
to foster dynamic efficiency led by the continuous introduction of innovations
(following Schumpeter, 1942; see Galbraith, 1952; Sylos Labini, 1962; for a review
Scherer, 1992; Kamien and Schwartz, 1982, Cohen, 1995). Creative destruction has
been associated with a market structure characterized by high dynamism and
competition, as well as high rate of change in the hierarchy of innovators. On the
contrary, creative accumulation patters are linked to oligopolistic market structure with
high entry barriers and high degree of stability of innovators.
Nelson and Winter (1982) suggest that the market structure in a specific
industry, the degree of concentration and rate of entry, are influenced by the degree to
which technological opportunities arise and the ease with which innovations can be
protected from imitation (i.e. the appropriability conditions). High technological
opportunity together with low appropriability causes lower concentration in an industry
and vice versa. These arguments are picked-up and empirically tested by Breschi et al.
(2000) and Malerba and Orsenigo (1995, 1997) in their work on technological regimes
and their role in the evolution of industrial structures, hierarchy of innovators and
innovation activities. The following section operationalizes the concepts discussed in
this section and summarized in Table 1.
4. Data and methodology
4.1 The data
The empirical part of the paper analyses the Innobarometer Survey 2009 that is
designed and collected by the European Commission (European Commission, 2009). In
each of the 27 EU Member States, plus Norway and Switzerland, 200 enterprises with
main activities in innovation intensive industry sectors and with 20 or more employees
were sampled.
1
5,238 telephone interviews were completed between the 1st and 9th of
1
In the smallest EU countries, Cyprus, Malta, and Luxembourg, the sample consisted of 70 enterprises
and in non-EU countries, Switzerland and Norway, the sample size was 100. The industry sectors
included are: aerospace, defence, construction equipment, apparel, automotive, building fixtures,
equipment, business services, chemical products, communications equipment, construction materials,
distribution services, energy, entertainment, financial services, fishing products, footwear, furniture,
heavy construction services, heavy machinery, hospitality and tourism, information technology, jewellery
and precious metals, leather products, lighting and electrical equipment, lumber and wood manufacturers,
medical devices, metal manufacturing, oil and gas products and services, paper, (bio)pharmaceuticals,
plastics, power generation & transmission, processed food, publishing and printing, sport and child goods,
textiles, transportation and logistics, utility.
9
April 2009. The sample is a random sample, stratified by country, enterprise size (5 size
bands) and industry (2-digit industry codes).
2
Since 2001 Innobarometer is conducted on a yearly basis. Each year the survey
highlights a different issue/theme, which is picked up on in additional and specific
questionnaire items over and above a core set of variables. The focus of the current,
2009 survey is on innovation related expenditures and the effects of the economic
downturn on innovation related expenditures. It is this section of the questionnaire from
which our key variables are developed. In the remainder of this section we introduce our
dependent and independent variables and discuss the methodology.
4.2 The dependent variables
Our dependent variables measure change in innovation related investment as it is
reported by the firms themselves and with reference to different time periods (before,
during and following on from the crisis). Innovation related investment are captured in a
wide sense, incorporating not only expenditures on in-house R&D but also technology
embodied in the purchase of machinery, equipment and software, licensed-in
technology (patents or other know-how), training of staff in support of innovation, and
expenditures on design of products, process and services. This broad definition (in line
with the definition adopted in the Community Innovation Surveys) has advantages over
a narrow definition, such as investment in R&D. R&D expenditures will not be able to
capture short-term responses to the financial crisis on the grounds that R&D projects are
typically commitments made for several years. Moreover, R&D is also concentrated in a
few firms and sectors. In contrast, the wider definition of innovation related investments
used in this paper that includes other innovation related expenditures over and above
R&D, is better suited to capture short-term adjustments due to changes in the economic
environment. Firms are quicker in cutting training for innovation, design budgets or
purchases of software, than they are in adjusting R&D projects.
questions.
compared to 2006 has the total amount spent by your firm
on all innovation activities in 2008 increased, decreased or stayed approximately the
same
in the last six months
3
has your company taken one of the
following actions as a direct result of the economic downturn; increased total amount of
innovation expenditures, decreased […] or maintained […]?
compared to 2008, do you
expect your company to increase, decrease or maintain the total amount of its
innovation expenditure in 2009?”.
The observations feeding into the empirical analysis are all those firms that were
innovation active and, thus, firms that stated they increase, decrease or maintain their
innovation investment in the three periods respectively. The weakness of our dependent
variables change in innovation related investment is that the scales are categorical
rather than continuous (e.g. three choices as opposed to the total amount spent on
2
A detailed description of the survey, including the sampling and data collection methods, can be found
in a methodological report by the European Commission (2009).
3
The interviews were conducted between 1 and 9 April 2009, and, thus, the question relates to the period
starting October 2008 ending with March 2009.
10
innovation); but the strength is that they provide a unique possibility to distinguish
between three different time periods around the crisis.
Table 2 provides the descriptive statistics for the three dependent variables,
including the number (frequency) and percent of enterprises that increased, maintained
we
T2 and
4
Table 2 Investment in innovation related activities before, during and following on
from the beginning of the crisis
Dependent variable:
change in innovation
related investment
Before the crisis
During the crisis
Following on from
the beginning of
the crisis
(T1)
(T2)
(T3)
Frequency
Percent
Frequency
Percent
Frequency
Percent
Increase
1,985
38
453
9
659
13
Decrease
472
9
1,231
24
1,560
30
Maintain
2,207
42
2,961
57
2,452
47
Innovation active firms
4,664
89
4,645
90
4,671
90
No innovation activities
328
6
457
9
343
7
Missing observations
242
5
132
3
220
4
Number of observations
5,234
100
5,234
100
5,234
100
T1 refers to the change in innovation related investment in the calendar year 2008 compared to 2006; T2
refers to the change in innovation related investment in the six months period October 2008 to March
2009; T3 refers to the expected change in innovation related investment in 2009 compared with 2008.
Table 2 reveals two patterns. Firstly, 38% of enterprises reported that they increased
innovation related investment in 2008 compared with their investment in 2006 (see
enterprises reported increased investment. Thus, there is a strong drop in the number of
firms that increased innovation related investment during the crisis and following on
from the crisis. This pattern is mirrored in a shift from few firms to many firms
reporting decreased investment over the three time periods. In T1 only 9% of firms
decreased their innovation related expenditures, but in the midst of the financial crisis
in T2 24% decreased investment and 30% planned to decrease investment in 2009
compared to investment levels in 2008. This might at the aggregate level point towards
destruction. Secondly, a large share of firms (about half of all firms) reported that they
maintained innovation related investment irrespectively of the crisis leaning towards an
accumulation hypothesis.
4
The Innobarometer survey reports a lower number of non-innovation active firms compared with
similar datasets, and specifically the Community Innovation Surveys. The following factors might
contribute: (a) a difference in the industrial composition viewed in Innobarometer
EC (2009), and (b) Innobarometer includes
firms with 20 or more employees while the Community Innovation Survey includes enterprises with 10
and more employees.
11
In Table 3 we report the cross-tabulations and Chi2 statistics between the
dependent variables producing three cross-tabulations: before the crisis (T1) with during
the crisis (T2); before the crisis (T1) with following on from the crisis (T3); and during
the crisis (T2) with following on from the crisis (T3). We present the cross-tabulations
to gain insight into the level continuity/discontinuity in innovation investment decisions.
For example, are the firms that increased investment during the crisis also among the
firms that increased investment before the crisis?
Table 3 Innovation investment before, during and following on from the crisis. Cross-
tabulations of the dependent variables
During the crisis (T2)
Increase
Decrease
Maintain
Total
Before
the crisis
(T1)
Increase
Frequencies
332
445
1,124
1,901
Column percentages
76
38
40
43
Decrease
Frequencies
18
255
167
440
Column percentages
4
22
6
10
Maintain
Frequencies
88
469
1,538
2,095
Column percentages
20
40
54
47
Total
Frequencies
438
1,169
2,829
4,436
Column percentages
100
100
100
100
Chi2(4)=463; p<0.01
Following on from the crisis (T3)
Increase
Decrease
Maintain
Total
Before
the crisis
(T1)
Increase
Frequencies
358
631
907
1,896
Column percentages
58
43
39
43
Decrease
Frequencies
62
225
158
445
Column percentages
10
15
7
10
Maintain
Frequencies
200
625
1,270
2,095
Column percentages
32
42
54
47
Total
Frequencies
620
1,481
2,335
4,436
Column percentages
100
100
100
100
Chi2(4)=168; p<0.01
12
Following on from the crisis (T3)
Increase
Decrease
Maintain
Total
During
the crisis
(T2)
Increase
Frequencies
192
73
159
424
Column percentages
32
5
7
10
Decrease
Frequencies
61
812
256
1,129
Column percentages
10
57
11
26
Maintain
Frequencies
350
544
1,832
2,726
Column percentages
58
38
82
64
Total
Frequencies
603
1,429
2,247
4,279
Column percentages
100
100
100
100
Chi2(4)=1,400; p<0.01
Source: As for Table 2.
In the cross-tabulations we report frequencies and column percentages below the
frequencies. In the first column total of the top cross-table we report that 438 firms
increased investment during the crisis (T2), and, in the first cell of the first cross-
tabulation, we report that, out of these 438 firms, 332 also increased investment before
the crisis (T1). This is the same as stating that 76% of firms that increased investment
during the crisis are firms that already increased investment before the crisis. These
76% or 332 firms indicate some consistency of investment patterns and may already
point towards, despite of the crisis, a confirmation of the importance of technological
accumulation.
But, out of the 438 firms that increased investment during the crisis (and 620
that increased investment following on from the crisis, see the middle cross-
tabulations), 24% (and 42%) decreased or maintained investment before the crisis. And,
it is among these firms that we could see a shift in firm characteristics and market
conditions associated with increased innovation investment before, during and
following on from the crisis.
From the information presented in Table 3 we also know that there is greater
stability in the investment choices of firms between the two periods during (T2) and
following on from (T3) the crisis, also resulting in the higher measure of association
(Chi2(4) = 1,400; p<0.01), compared with before the crisis (T1 and T2, T1 and T3).
To fully address our research question of who the firms are that increase
investment (top row of Table 2) in the midst of the crisis (a) the most dynamic ones
that compete largely on continuous upgrading or (b) new players that could be newly
established firms or firms less relevant in aggregate innovation we use a set of
measures capturing firm and market characteristics to which we now turn, and that we
use to predict innovation related investment across T1, T2 and T3 in the results section
of the paper.
4.3 The independent variables
Table 4 contains an overview of the independent variables arranged by the categories
introduced in Table 1. These categories are; i) characteristics of the innovating firms, ii)
type of knowledge sources, iii) type of innovations and iv) market characteristics.
13
Table 4 Characteristics of the innovating firms, type of knowledge sources, type of
innovations and characteristics of the market. Overview of the independent variables
Characteristics of the innovating firms
Newly established
The enterprise was established after 1 January 2001
Small enterprise
There are four dummies that we use to measure the size of
the enterprise. Small enterprises here have 20-49 employees
Medium enterprise
The variable selects all enterprises with 50 to 249 employees
Large enterprises
The variables selects all enterprises with more than 250
employees
Low innovation
intensity
The enterprise invests less than 5% of turnover in innovation
related activities in 2008
High innovation
intensity
The enterprise invests at least 5% of turnover in innovation
related activities
Type of knowledge sources
In-house R&D
The enterprise had expenditures on in-house R&D since
2006
Bought-in R&D
The enterprise had expenditures on R&D performed for the
company by other enterprises or by research organisations
since 2006
Link with other
firms
The enterprise developed strategic relationships in support of
innovation with customers, suppliers or other companies
since 2006
Link with the
knowledge base
The enterprise developed strategic relationships in support of
innovation with research institutes and educational
institutions since 2006
International
collaboration
The enterprise started or increased cooperation with local
partners in other countries in support of innovation since
2006
Investment in
companies abroad
The enterprise invested in companies located in other
countries in support of innovation since 2006
Type of innovations
Enterprise
competes on
innovations
The enterprise sees the main competitive advantage in new
products, services and processes
Enterprise
competes on
improvements
The enterprise sees the main competitive advantage in the
modification of existing products, services and processes
Enterprise
competes on new
business models
The enterprise sees the main competitive advantage in the
developments of new business models or ways to market
products and services
Enterprise
competes on cost
The enterprise sees the main competitive advantage in
reducing costs of existing products
14
Characteristics of the market
IPRs
The enterprise applied for a patent or registered a design since
2006
Technological
opportunities
New technologies emerged in the market since 2006
Market
opportunities
New opportunities to enter into new markets or expand sales in
existing markets emerged since 2006
International
market
The enterprise operates in international markets
The first column in Table 4 gives the variable names of the independent variables and
the second column the variable description. All our independent variables are dummy
variables coded 1 if a characteristic is met and zero otherwise. We rely on dummies
because of a lack of more detailed information. In the first category entitled
variable is coded 1 if a firm was established after 1 January 2001 and 0 if it was
established earlier. This variable is used as a proxy to identify new entrants. The second
set of variables is made of three dummies that we use to proxy firm size. Small firms
(20 to 49 employees) are used as the base comparison group in the regressions. The
final variable proxies the innovation intensity of firms or the stock/level of investment
in innovation related activities with reference to the calendar year 2008. High
innovation intensity is measured as a share of turnover at least 5% is spent on
innovation related activities.
5
Low innovation intensity (i.e. below 5% of turnover) is
the base group.
that captures if the enterprise engaged in in-house R&D, second, if it engaged in
extramural R&D. The remaining four variables relate to linkages or joint knowledge
sources; specifically, collaboration on innovation with other businesses, collaboration
on innovation with educational and other research institutions, collaborations with
partners located abroad, and investment in companies located abroad. All variables are
coded 1 for yes answers and zero for no answers.
are
proxies for the strategic orientation of the firms with respect to their innovations:
whether or not firms compete based on their innovations, based on improvements to
existing products, based on a new business model, or based on cost savings. Competing
on innovation might lean more closely to activities at the frontier and might be seen as
more closely related to path-breaking developments vis-à-vis the remaining categories.
While improvements lean towards incremental innovations, new business models might
be indicative of a new service. Competing on cost might favour the upgrading of
processes. There is, of course, much blurring and overlap across such categories when
Under
first one captures the use of IPRs, specifically whether or not the firm applied for a
patent or registered a design. The next two variables are used to capture the
technological opportunities and market opportunities as assessed by the responding
5
The dataset has a fourth category innovation related expenditure above 50% of turnover but less than
1% of firms fell into this group and this is why we merged it with the next smaller band.
15
firms. 1 indicates that the firm perceived that there were opportunities (technological or
market) and zero suggests a lack of opportunities. The final variable takes values of 1 if
the enterprise operates in international markets and zero otherwise. Table 5 provides
and overview of the descriptive statistic for all independent variables.
Table 5 Descriptive statistics of the independent variables
Independent variables
Number of
observations
Mean
Standard
deviation
Characteristics of the innovating firms
Newly established
4,664
0.08
0.28
Small enterprise (base group)
4,664
0.40
0.49
Medium enterprise
4,664
0.32
0.47
Large enterprise
4,664
0.28
0.45
Low innovation intensity (base group)
4,298
0.68
0.47
High innovation intensity
4,298
0.32
0.47
Type of knowledge sources
In-house R&D
4,635
0.48
0.50
Bought-in R&D
4,631
0.32
0.47
Link with other firms
4,627
0.67
0.47
Links with the knowledge base
4,628
0.38
0.49
International collaboration
4,602
0.29
0.45
Investment in companies abroad
4,620
0.11
0.31
Type of innovations
Enterprise competes on innovations
4,558
0.24
0.43
Enterprise competes on improvements
4,558
0.23
0.42
Enterprise competes on business models
4,558
0.16
0.37
Enterprise competes on cost (base group)
4,558
0.34
0.47
Characteristics of the market
IPRs
4,613
0.15
0.36
Technological opportunities
4,594
0.40
0.49
Market opportunities
4,596
0.58
0.49
International market
4,588
0.50
0.50
Source: As for Table 2.
Most of the dependent variables are observed for 4,664 firms (out of 5,234 observations
in the initial database) in T1 (and 4,645 and 4,671 in T2 and 3 respectively) and Table 5
presents descriptive statistics for the independent variables based on these 4,664
observations. With respect to some of the independent variables we have missing
observations where respondents stated that they did not know the answer. Specifically,
4,298 respondents provided a valid response with respect to their innovation intensity
and so on. Because of missing values (and missing values not occurring systematically
by appearing within the same observations) we have a final dataset of 3,959
observations in T1 (3,886 T2 and 3,890 T3) that is used in the regressions. This dataset
is the largest possible dataset that contains observations for all dependent and
independent variables.
Because these are all dummy variables, this column is the share of enterprises that
16
engage in a specific activity, e.g. 0.08 or 8% of firms were newly established, 40% were
small, 50% of firms reported that they operated in international markets.
4.4 Methodology
We use regressions to analyse the relationships between our dependent and independent
variables. Table 6 provides the zero order correlations between the dependent and
independent variables, reporting polychoric correlations for the categorical dependent
variables and tetrachoric correlations between the binary independent variables.
Table 6 Correlations between the dependent and independent variables
Dependent variables
1
2
3
Investment in innovation related activity
1
Investment before the crisis
1.00
2
During the crisis
0.28
1.00
3
Following on from the crisis
0.21
0.44
1.00
Independent variables
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Characteristics of the innovating firms
1
Newly established
1.00
2
Small enterprise (base group)
0.09
1.00
3
Medium enterprise
0.02
-1.00
1.00
4
Large enterprise
-0.13
-1.00
-1.00
1.00
5
Low innovation intensity (base)
-0.03
0.05
-0.02
-0.05
1.00
6
High innovation intensity
0.03
-0.05
0.02
0.05
-1.00
1.00
Type of knowledge sources
7
In-house R&D
-0.03
-0.29
0.03
0.31
-0.28
0.28
1.00
8
Bought-in R&D
-0.02
-0.31
0.01
0.33
-0.15
0.15
0.63
1.00
9
Link with other firms
0.08
-0.15
-0.01
0.19
-0.28
0.28
0.45
0.37
1.00
10
Links with the knowledge base
0.02
-0.25
0.01
0.27
-0.25
0.25
0.53
0.51
0.58
1.00
11
International collaboration
-0.06
-0.19
-0.02
0.23
-0.25
0.25
0.41
0.36
0.47
0.37
1.00
12
Investment in companies abroad
-0.06
-0.25
-0.09
0.34
-0.16
0.16
0.38
0.35
0.39
0.29
0.65
1.00
Type of innovations
13
Enterprise competes on innovations
-0.01
-0.02
0.01
0.02
-0.20
0.20
0.21
0.18
0.18
0.17
0.13
0.13
1.00
14
Competes on improvements
0.05
-0.06
0.03
0.04
0.03
-0.03
0.04
0.00
0.09
0.05
0.04
-0.07
-1.00
1.00
15
Competes on business models
-0.04
0.01
-0.04
0.03
-0.05
0.05
0.04
0.06
0.13
0.08
0.06
0.12
-1.00
-1.00
1.00
16
Competes on cost (base group)
-0.02
0.02
0.00
-0.03
0.14
-0.14
-0.17
-0.15
-0.20
-0.20
-0.14
-0.11
-1.00
-1.00
-1.00
1.00
Characteristics of the market
17
IPRs
-0.05
-0.24
-0.06
0.31
-0.26
0.26
0.53
0.44
0.37
0.39
0.38
0.36
0.19
0.05
0.00
-0.18
1.00
18
Technological opportunities
0.00
-0.18
0.00
0.21
-0.31
0.31
0.39
0.32
0.48
0.43
0.30
0.28
0.18
0.07
0.08
-0.19
0.31
1.00
19
Market opportunities
0.03
-0.16
0.02
0.18
-0.27
0.27
0.35
0.28
0.48
0.31
0.41
0.29
0.18
0.04
0.13
-0.16
0.33
0.50
1.00
20
International market
-0.02
-0.23
0.01
0.26
-0.17
0.17
0.35
0.26
0.25
0.22
0.54
0.53
0.11
0.02
0.01
-0.05
0.36
0.22
0.37
Polychoric correlations between the dependent variables, and tetrachoric correlations between the independent variables, are reported. The variables Compete on
innovations, improvements, business models and cost are mutually exclusive and thus yield a tetrachoric correlation of -1. Source: As for Table 2.
The correlations reveal, in line with our expectations and our Table 3, that there is a
respect to T2 and T3). Among the independent variables, the highest overlap exists
between in-house R&D and bought-in R&D (r=0.63; p<0.01). Previous studies have
shown that internal and bought-in R&D activities are complementing strategies, rather
than substitutes (Cassiman and Veuglers, 2006). A high overlap also exists between
p<0.01 and r=0.53; p<0.01 respectively), suggesting that these variables taken together
might be indicative of an international orientation of firms.
6
The variables in the
this is why the
tetrachoric correlations return a value of -1. Competing on cost is our base comparison
group in the regressions.
It is a limitation of our dependent variables that we do not have continuous data
and so we cannot use the classic linear model. The dependent variables are categorical
variables that take the following categories: 1 = decrease in innovation related
investment; 2 = innovation investment maintained; 3 = increase in innovation related
investment.
We report the results from two estimation models: a logistic regression model
and a multinomial logistic regression model. The logistic regression predicting
increased innovation investment compared to both the remaining outcomes taken
together (decreased and maintained) is presented because the interpretation of the
coefficients is easier; however, the model ignores that the firm is presented with three
choices to increase, decrease or maintain investment. The latter is picked up by the
multinomial logistic regression. The logistic model is:
where xj is the row vector of the values of the independent variables. The
multinomial logistic that picks up the three choices is:
where pij is the probability that the jth observation is equal to the ith outcome. 1 is
assumed to be the base outcome, k is the number of categories (in our case 3), bm is the
coefficient for the outcome m (in our case either 2 or 3), and as before xj is the row
vector of the values of the independent variables. Based on one multinomial logistic
regression, three sets of coefficients are reported: the first set of coefficients compares
6
In order to address an issue of multicollinearity between these variables, we have computed all
upon request from the authors.
19
the choice to increase investment with maintained investment; the second set compares
increase with decrease in investment; and the third set compares the effects of the
independent variables on maintaining investment compared with decreasing investment.
We now turn to the presentation of the empirical results in the next section.
5. Results
Two models are presented in this section. The first logistic regression reports
coefficients that are indicative of the probability to increase innovation investment if the
independent variables all dummies take a value of 1, i.e. the characteristic such as
20
Table 7 Factors explaining the choice to increase innovation investment compared to
maintaining or decreasing investment (combined) over time
Dependent variable: increase in innovation
related investment
Before the
crisis
During the
crisis
Following on
from the
crisis
Estimation method: logistic
(T1)
(T2)
(T3)
Characteristics of the innovating firms
Newly established
-0.19
-0.12
0.27*
(0.13)
(0.20)
(0.16)
Medium enterprise
0.13
-0.13
0.10
(0.08)
(0.13)
(0.11)
Large enterprise
0.12
-0.64***
-0.15
(0.09)
(0.16)
(0.13)
High innovation intensity
0.97***
0.20*
0.01
(0.08)
(0.12)
(0.10)
Type of knowledge sources
In-house R&D
0.33***
0.21
0.20*
(0.08)
(0.14)
(0.12)
Bought-in R&D
0.26***
-0.08
-0.07
(0.09)
(0.13)
(0.11)
Link with other firms
0.36***
0.33**
0.23*
(0.08)
(0.15)
(0.12)
Links with the knowledge base
0.07
0.15
0.15
(0.08)
(0.13)
(0.11)
International collaboration
0.30***
0.38***
0.35***
(0.09)
(0.13)
(0.11)
Investment in companies abroad
-0.02
-0.05
-0.33**
(0.13)
(0.19)
(0.17)
Type of innovations
Enterprise competes on innovations
0.29***
0.36**
0.58***
(0.10)
(0.15)
(0.13)
Enterprise competes on improvements
0.24**
0.22
0.61***
(0.10)
(0.16)
(0.13)
Enterprise competes on business models
0.14
0.15
0.52***
(0.11)
(0.17)
(0.15)
Characteristics of the market
IPRs
0.27**
0.32**
0.16
(0.11)
(0.15)
(0.13)
Technological opportunities
0.20***
0.04
0.07
(0.08)
(0.12)
(0.11)
Market opportunities
0.16**
0.40***
0.17
(0.08)
(0.13)
(0.11)
International market
-0.16*
-0.02
0.00
(0.08)
(0.13)
(0.11)
Industry dummies
Included
Included
Included
Country dummies
Included
Included
Included
Number of observations
3,959
3,886
3,890
Wald Chi2 (64)
524***
150***
179***
Pseudo R2
0.11
0.07
0.06
*** p<0.01, ** p<0.05, * p<0.10. Robust standard errors are reported in brackets under the logistic
regression coefficients. Source: As for Table 2.
21
Before the crisis (column T1 in Table 7), and with respect to the characteristics of the
innovating firms, the coefficients suggest that firms are more likely to increase
innovation investment if they exhibit high innovation intensity (our proxy for stock of
investment). The coefficient b=0.97 (p<0.01) is the largest coefficient in the column T1.
Size and age are not significantly associated with increased investment, but the positive
sign of the coefficients is in line with technological accumulation patterns (as per Table
investment, meaning that small firms (our base group) are statistically more likely to
increase investment compared with the group of large firms. The coefficient b=-0.64
(p<0.01) is the most influential coefficient in the column T2. Following on from the
crisis (T3) new entrants are more likely to increase investment (b=0.27; p<0.10). Both
patterns, small firms in T2 and new entrants in T3, lean towards the creative destruction
hypothesis (as per Table 1).
In relation to type of knowledge sources, our second category of independent
-
- the crisis supporting accumulation of technology before the
-
-in
either T2 or T3 and the sign of the coefficients are negative.
proxy for access to applied knowledge that we thought less closely linked to
variables do not suggest a change in pattern from before the crisis to during the crisis.
Finally, firms that invested in companies abroad appear less likely to increase
innovation investment following on from the crisis (no effect before then in columns T1
and T2). This variable, albeit restricted to the time period starting 2006, might capture if
a firm was part of a larger, multinational company. Interpreted that way, the finding is
closer to a destruction hypothesis. From our theoretical point of departure, the drop in
significance of in-house and bought-in R&D during and following on from the crisis
lends some support for the destruction hypothesis. But the findings in this category are
less clear with respect to applied and generic knowledge sources as the coefficients are
consistent across our three time periods.
Our proxies for types of innovations reveal that throughout the three periods,
firms that increase investment in innovation are less likely to compete on cost, than they
are to compete on innovations (confirming similar results previously reported by
Bogliacino and Pianta, 2010). Firms competing on cost are also less likely to increase
investment compared with firms that compete on improvements before and following
on from the crisis, but not during the crisis. The size of the coefficients increases over
the three time periods, which indicates that firms that compete on costs are increasingly
less likely to increase innovation related investment, specifically in T3 where the
coefficients (compete on innovation, improvements and business model contrasted with
competing on costs) have the strongest impact in the regression model. The sole
significance of competing on innovation during the crisis, coupled with the increase in
perhaps less indicative of accumulation as it
is of destruction in T2 and T3.
22
With respect to the characteristics of the market, our final category of
independent variables, the coefficients in Table 7 for IPRs are positive and significant
both before and during the crisis (but not following on from the crisis T3). The
er, are positively and significantly associated with increased
hypothesis both before and during the crisis.
In Table 8, a pattern consistent with that in Table 7, but with greater detail with
respect to the differences in the choices to maintain investment and decreasing
investment is reported. Table 8 (a-c) contains one regression model for T1, T2 and T3
respectively, but three sets of coefficients are reported: (a) the first set of coefficients
contrasts increase in innovation investment against maintaining of investment; (b)
contrasts increase in innovation investment against decrease in investment; and (c)
maintaining in investment against decrease in investment.
23
Table 8.a Factors explaining the discrete choices to increase, maintain, or decrease
innovation related investment over time
Dependent variable: increase in innovation
investment (base group: maintain)
Before the
crisis
During the
crisis
Following
on from the
crisis
Estimation method: multinomial logistic
(T1)
(T2)
(T3)
Characteristics of the innovating firms
Newly established
-0.19
-0.14
0.22
(0.15)
(0.50)
(0.19)
Medium enterprise
0.13
-0.18
0.06
(0.15)
(0.17)
(0.60)
Large enterprise
0.06
-0.67***
-0.21
(0.56)
(0.00)
(0.11)
High innovation intensity
0.99***
0.30**
0.15
(0.00)
(0.02)
(0.16)
Type of knowledge sources
In-house R&D
0.39***
0.23
0.18
(0.00)
(0.10)
(0.14)
Bought-in R&D
0.23***
-0.09
-0.06
(0.01)
(0.53)
(0.62)
Link with other firms
0.42***
0.37**
0.28**
(0.00)
(0.01)
(0.02)
Links with the knowledge base
0.05
0.17
0.11
(0.55)
(0.19)
(0.36)
International collaboration
0.33***
0.41***
0.36***
(0.00)
(0.00)
(0.00)
Investment in companies abroad
-0.00
-0.04
-0.27
(0.98)
(0.83)
(0.13)
Type of innovations
Enterprise competes on innovations
0.25**
0.22
0.39***
(0.01)
(0.16)
(0.00)
Enterprise competes on improvements
0.21**
0.07
0.47***
(0.04)
(0.64)
(0.00)
Enterprise competes on business models
0.14
0.08
0.43***
(0.19)
(0.65)
(0.00)
Characteristics of the market
IPRs
0.32***
0.34**
0.11
(0.00)
(0.03)
(0.43)
Technological opportunities
0.18**
0.07
0.10
(0.03)
(0.57)
(0.35)
Market opportunities
0.13
0.39***
0.16
(0.11)
(0.00)
(0.16)
International market
-0.15*
0.02
0.06
(0.09)
(0.86)
(0.61)
Industry dummies
Included
Included
Included
Country dummies
Included
Included
Included
Number of observations
3,959
3,886
3,890
Wald Chi2 (64)
652***
431***
419***
Pseudo R2
0.10
0.07
0.06
*** p<0.01, ** p<0.05, * p<0.10. Robust standard errors are reported in brackets under the multinomial
logistic regression coefficients. Source: As for Table 2.
24
Table 8.b Factors explaining the choice to increase, maintain or decrease innovation
investment over time
Dependent variable: increase in innovation
investment (base group: decrease)
Before the
crisis
During the
crisis
Following
on from the
crisis
Estimation method: multinomial logistic
(T1)
(T2)
(T3)
Characteristics of the innovating firms
Newly established
-0.16
-0.09
0.35**
(0.43)
(0.68)
(0.05)
Medium enterprise
0.16
-0.01
0.16
(0.23)
(0.95)
(0.20)
Large enterprise
0.40**
-0.54***
-0.04
(0.01)
(0.00)
(0.79)
High innovation intensity
0.91***
-0.02
-0.22*
(0.00)
(0.86)
(0.06)
Type of knowledge sources
In-house R&D
0.04
0.15
0.25*
(0.79)
(0.33)
(0.05)
Bought-in R&D
0.34**
-0.07
-0.09
(0.02)
(0.66)
(0.45)
Link with other firms
0.10
0.23
0.14
(0.45)
(0.15)
(0.29)
Links with the knowledge base
0.13
0.10
0.21*
(0.35)
(0.51)
(0.09)
International collaboration
0.21
0.32**
0.33***
(0.14)
(0.04)
(0.01)
Investment in companies abroad
-0.11
-0.06
-0.43**
(0.58)
(0.77)
(0.02)
Type of innovations
Enterprise competes on innovations
0.45***
0.71***
0.89***
(0.00)
(0.00)
(0.00)
Enterprise competes on improvements
0.36**
0.55***
0.83***
(0.02)
(0.00)
(0.00)
Enterprise competes on business models
0.11
0.29
0.63***
(0.51)
(0.13)
(0.00)
Characteristics of the market
IPRs
0.05
0.28*
0.26*
(0.76)
(0.10)
(0.08)
Technological opportunities
0.31**
-0.04
-0.00
(0.02)
(0.79)
(1.00)
Market opportunities
0.27**
0.45***
0.20
(0.04)
(0.00)
(0.10)
International market
-0.22*
-0.15
-0.10
(0.09)
(0.30)
(0.41)
Industry dummies
Included
Included
Included
Country dummies
Included
Included
Included
Number of observations
3,959
3,886
3,890
Wald Chi2 (64)
652***
431***
419***
Pseudo R2
0.10
0.07
0.06
*** p<0.01, ** p<0.05, * p<0.10. Robust standard errors are reported in brackets under the multinomial
logistic regression coefficients. Source: As for Table 2.
25
Table 8.c Factors explaining the choice to increase, maintain or decrease innovation
investment over time
Dependent variable: maintained innovation
investment (base group: decrease)
Before the
crisis
During the
crisis
Following
on from the
crisis
Estimation method: multinomial logistic
(T1)
(T2)
(T3)
Characteristics of the innovating firms
Newly established
0.03
0.05
0.13
(0.88)
(0.74)
(0.32)
Medium enterprise
0.03
0.17*
0.10
(0.80)
(0.07)
(0.26)
Large enterprise
0.34**
0.13
0.18*
(0.02)
(0.21)
(0.07)
High innovation intensity
-0.08
-0.32***
-0.37***
(0.55)
(0.00)
(0.00)
Type of knowledge sources
In-house R&D
-0.36***
-0.08
0.07
(0.01)
(0.40)
(0.42)
Bought-in R&D
0.11
0.02
-0.04
(0.44)
(0.84)
(0.70)
Link with other firms
-0.31**
-0.13
-0.14
(0.02)
(0.16)
(0.11)
Links with the knowledge base
0.08
-0.08
0.11
(0.56)
(0.42)
(0.23)
International collaboration
-0.12
-0.09
-0.03
(0.40)
(0.40)
(0.78)
Investment in companies abroad
-0.11
-0.02
-0.16
(0.59)
(0.88)
(0.23)
Type of innovations
Enterprise competes on innovations
0.20
0.50***
0.50***
(0.20)
(0.00)
(0.00)
Enterprise competes on improvements
0.15
0.48***
0.36***
(0.33)
(0.00)
(0.00)
Enterprise competes on business models
-0.03
0.21*
0.19*
(0.83)
(0.07)
(0.08)
Characteristics of the market
IPRs
-0.27
-0.05
0.15
(0.13)
(0.66)
(0.20)
Technological opportunities
0.12
-0.11
-0.10
(0.33)
(0.23)
(0.22)
Market opportunities
0.14
0.06
0.04
(0.26)
(0.53)
(0.62)
International market
-0.07
-0.17*
-0.16*
(0.58)
(0.06)
(0.06)
Industry dummies
Included
Included
Included
Country dummies
Included
Included
Included
Number of observations
3,959
3,886
3,890
Wald Chi2 (64)
652***
431***
419***
Pseudo R2
0.10
0.07
0.06
*** p<0.01, ** p<0.05, * p<0.10. Robust standard errors are reported in brackets under the multinomial
logistic regression coefficients. Source: As for Table 2.
26
One caveat that Table 8 reveals, and that cannot be seen in Table 7, is that firms that
maintain investment as opposed to both increase (Table 8.a) and decrease (Table 8.c),
report lower innovation intensity during the crisis. Thus, reacting to the crisis by either
increasing or decreasing innovation related investment are the two choices made by the
more innovative firms.
Another caveat taken from Tables 8.a-c is related to large firms. Before the
crisis, large firms are more likely to increase investment (as opposed to decrease
investment Table 8.b) and are more likely to maintain investment (as opposed to
decrease investment Table 8.c). In contrast, during the crisis large firms are less likely
to increase investment as opposed to both the alternative choices to maintain or
decrease investment (Tables 8.a and b). This, in line with the findings reported in Table
7, suggests that the role of small firms in innovation during the crisis is greater (a) than
before the crisis and (b) compared with large firms during the crisis, supporting the
destruction hypothesis.
Finally, comparing the choices increase and decrease in investment in the time
period following on from the crisis, Table 8.b reports (as Table 7 before) newly
established firms as more likely to increase investment. Among the remaining
coefficients of the same set of coefficients, Table 8.b also reports that firms with low
innovation intensity (stock) increase investment in T3. But, among the same set of
-
significant, providing a mixed picture with some characteristics closer to creative
-
while we might have expected the patterns between T2 and T3 to be highly similar but
different from T1, increased investment is not necessarily done by firms with the exact
same characteristics and environments across T2 and T3, and some of the patterns
dominant (significant coefficients) in T1 re-emerge in T3.
6. Discussion
The aim of this paper is to investigate whether the current economic downturn is
significantly affecting the composition of innovating firms. During major recessions,
the economic landscape is characterized by huge uncertainties about the direction of
technological change, demand conditions, and new market opportunities. The first
significant result at the aggregate level is that the crisis has substantially reduced the
number of firms willing to increase their innovation investment, from 38% to 9%. No
investment. But the anatomy of these 9% of firms that are still expanding their
innovation investment can provide some insights to check if the gales of destruction are
also bringing something creative.
We used two well-established, ideal type models creative destruction and
creative accumulation to frame our results (as summarized in Table 1). For the
purpose of developing the framework, we assumed a more clear-cut division according
to which in regular times the model of creative accumulation prevails, while in times of
crisis the model of creative destruction affirms itself. We are well aware that such a
clear-cut division between the two models does not exist. We recognize that both
patterns of innovation co-exist, and are likely to be also technology and industry
27
specific (as tested empirically by Malerba and Orsenigo, 1995). However, our data
suggest that during the recession
destruction, while before the recession there is an overall landscape of creative
accumulation.
More specifically, Innobarometer allowed us to test two hypotheses: a) that in
periods of economic expansion firms that are already innovating are the most important
drivers of increased innovation investment, supporting the technological accumulation
hypothesis; and b) that economic crises generate turbulence, and that newcomers are
eager to spend more to innovate, confirming the creative destruction hypothesis.
The empirical results support our arguments. The identikit of the innovators has
in fact changed considerably. Before the economic downturn, firms expanding their
innovations are: i) well-established; ii) engaged in formal research activities both
internally and bought-in; iii) exploit strong appropriability conditions; and iv) involved
in collaboration with suppliers and customers. During the economic downturn the few
are: i) smaller than before; ii) collaborating with other businesses; ii) exploring new
market opportunities; iii) using methods of technological appropriation; and iv) less
likely to compete on costs. Last but certainly not least, it also seems that younger firms
are more likely to increase innovation investment after the crisis. While before the crisis
technological opportunities have a positive impact on investment, during and after the
crisis this is no longer true. On the contrary, in response to the crisis firms are more
likely to explore innovative solutions by looking at opportunities in new markets.
This witnesses an important change in the drivers of innovation as a result of the
economic downturn. Since innovation is less based on local searching and cumulative
processes, and less based on R&D activities within large firms, we conclude that the
relative importance of behaviours is changing from creative accumulation to creative
destruction in the snap shot of the business cycle that the Innobarometer makes it
-à-vis
face. During the crisis both formal R&D and technological opportunities stop to play
might be interpreted as the result of a decline of technological opportunities in
established sectors which is typical during recessions characterized by technological
discontinuities (Perez, 2002). Also, contrary to the previous period, innovation is driven
by fresh opportunities in new markets.
new cluster of innovations that will generate the recovery (as indicated by Linstone and
Devezas, 2012), but at least provide some useful information to trace the identikit of the
post-crisis innovating firm.
behaviour lean towards accumulative patterns of innovation. During economic
upswings firms have access to greater financial resources and thus might be seen more
likely to explore radical and risky solutions. Similarly, it can be conceivably maintained
that during a depression large established firms are better equipped to manage a
situation of fall in demand and lack of financial supply in the market. However, we
show that this is not the case. The number of firms declaring to increase their innovation
expenditure has dropped dramatically as results of the crisis. It seems that what matter
28
are not large size and internal R&D, but flexibility, collaborative arrangements and
exploration of new markets.
Prospects for future research. Future work should focus on accessing data which
allows for estimates based on longer time periods, the inclusion of more countries and
more precise indicators on innovation intensity and the direction of technological
change. In particular, we suspect that the crisis is reinforcing the shift from the
manufacturing to the service industries, as indicated by in-depth country case studies
(see Kim, 2011). We can wonder if this is a general rule or is something associated to
the current phase of capitalist development, where the manufacturing sector, the core
generator of technological innovations, is progressively accounting for lower shares of
income and employment while, on the contrary, the service sector is gaining shares and
is more likely to compete through non-technological innovations and by finding new
markets. We can speculate that, if the economic recession is reinforcing the shift from
manufacturing to services, it would not be a surprise that the firms increasing their
innovation investment are more likely to be driven by searching new business lines and
business models than by technological opportunities. In order to corroborate this
hypothesis a definition of innovation able to capture the process of change in both
manufacturing and services is needed. For many years, the Schumpeterian economics
has concentrated on the technological dimension of innovation, which is typical of the
manufacturing industries, and has somehow denied the non-technological dimension,
which is more common when innovating in services. Times are ready to use a wider
understanding of innovation, similar to what was pioneered by Schumpeter himself a
century ago in the first edition of the Theory of Economic Development. The definition
provided by Innobarometer and used in this paper has the advantage to be more
inclusive than others
Limitations of the study. The analysis presented here is limited by the data and
the statistical models. First, the results are confined to Europe, and exclude the US as
well as emerging countries. Second, the data offer information on three time periods for
the dependent variables (but not for the independent variables), which allows comparing
innovation related investment patterns before, during and following on from the crisis.
Time series data would be able to provide much better information on the effects of the
crisis, and the next surveys will certainly shed light on this. Third, data do not allow
singling out the dynamic at the industry level. Finally, some variables are not totally
satisfactory. True, the Innobarometer survey offers a unique opportunity to shed light
onto the impact of the recent economic downturn on innovation, but we are well aware
of the limitations of having carried out such a clear-cut classification
Policy implications. In terms of policy analysis, it should be seen what the
restricted number of firms increasing the innovation investment will generate. Public
incentives to promote innovation can either be directed towards supporting the already
existing R&D infrastructures or towards fostering new entrants. Identifying the
characteristics of the innovators during the turmoil, as we have tried to do here, can
shed some light on how policy instruments interact with technological accumulation and
creative destruction. In which group of firms will the Bill Gates and Steve Jobs, Larry
Page and Sergey Brin of the next generation be found? And are we sure that European
governments, more and more concerned with the knowledge based economy, are doing
their best to foster creative innovators, even if this will imply the destruction of slow
growing wood?
29
References
Antonelli -
International Journal of Industrial Organization 15 (6), 643-675.
using revised Pavitt classesResearch Policy 39 (6), 799-809.
Economic Journal 110 (463), 388-410.
ity in the innovation
Management Science 52
(1), 6882.
-
country and cross-Research Policy 30 (7), 1139-1158.
Chandler, A. (1977), The Visible Hand: The Managerial Revolution in American
Business. Belkap: Cambridge, Mass.
Christensen, C. M. (1997), The Innovator's Dilemma. HarperBusiness Edition: New
York.
Christensen, C. M. and
Research
Policy 24 (2) 233-257.
(ed.),
Handbook of the Economics of Innovation and Technological Change. Basil Blackwell:
Oxford.
e 72, 913-935.
European Commission (2009), Innobarometer 2009. Brussels, DG Enterprise and
Industry.
Filippetti, A. and Archibugi, D. (2011
Research Policy 40 (2), 179-192.
Freeman, C. and Louca, F. (2001), As Times Goes By: From the Industrial Revolutions
to the Information Revolution. Oxford University Press: Oxford.
Galbraith, K. (1952), American Capitalism: The Concept of Countervailing Power.
Houghton-Mifflin: Boston.
30
Administrative Science Quarterly 35( ) 9-30.
Holbrook, D., Cohen,
and consequences of firms differences in the early history of the semiconductor
Strategic Management Journal 21 (10-11) 1017-1041.
Iansiti, M. and Levien, R. (2004), The Keystone Advantage. Harvard Business School
Press Boston: Massachusetts.
Kamien, M.I and Schwartz, N.L. (1982), Market Structure and Innovation. Cambridge
University Press: Cambridge.
r-
Change 79, 1345-1355.
producers and competitive ramifications in the U.S. televisio
Strategic Management Journal 21 997-1016.
and Social Change 78, 1319-1331.
Leonard-
Strategic Management Journal 13 (1), 111-126.
Strategic
Management Journal 14 (S2), 95-112.
-416.
gest US manufacturing
-ISEG, UTL, Lisbon.
Cambridge Journal of Economics 19 (1), 47-65.
Methé, D., Swaminathan, A.
Industrial and Corporate
Change 5 (4), 1181-1203
Nelson, R. and Winter, S. (1982), An Evolutionary Theory of Economic Change.
Harvard University Press: Cambridge, MA.
31
diversification and organization in UK companies, 1945-Management Science 35
(1) 81-99.
Patel, P. and Pavitt
Industrial and
Corporate Change 3 (3) 759-787.
nd Nelson, R.
(eds.) The Oxford Handbook of Innovation. Oxford University Press: Oxford.
Perez, C. (2002), Technological Revolutions and Financial Capital: The Dynamics of
Bubbles and Golden Ages. Edward Elgar: Cheltenham.
Industrial and Corporate Change 13 (5), 757-787.
Journal of Economic
Literature 30 (3), 1416-1433.
Schumpeter, J. A. (1911 (1934)), The Theory of Economic Development. Harvard
University Press: Cambridge.
Schumpeter, J. A. (1942), Capitalism, Socialism and Democracy. Harper: New York.
Technological Forecasting and Social Change 79, 595-604.
Fortune List, 1963- Open Discussion Papers in Economics 13, The Open
University, Faculty of Social Sciences, Department of Economics.
Sylos Labini, P. (1962), Oligopoly and Technical Progress. Harvard University Press:
Boston.
coherence: theory and evidence Journal of Economic Behaviour & Organization 23
1-30.
Administrative Science Quarterly 31 (3), 439-465.
