Sunteți pe pagina 1din 29

Paper to be presented at the DRUID Summer Conference on

"Industrial Dynamics of the New and Old Economy - who is embracing whom?"
Copenhagen/Elsinore 6-8 June 2002

Theme A : Technical Change, Corporate Dynamics and Innovation


Session 3-2 : Technological Trajectories

COMPLEMENTARITIES BETWEEN OBSTACLES TO INNOVATION :


EMPIRICAL STUDY ON A FRENCH DATA SET

Fabrice Galia and Digo Legros

ERMES - UMR 7017 CNRS


Universit Panthon-Assas Paris II - 12, place du Panthon
75231 PARIS Cedex 05

Tel.: 01 44 41 89 90 - Fax.: 01 40 51 81 30

E-mail : galia@u-paris2.fr and dlegros@u-paris2.fr

May 7th 2002

Abstract : The economics of innovation traditionally focuses on the determinants of research-


development (R&D) and innovation. Other approaches studying obstacles to innovation seem
to have been marginalized. We concentrate on this alternative approach. First, we describe
and detail the obstacles of the 1997 French Technological Innovation survey (CIS2). We
study econometrically what types of firms experience what kinds of obstacles. Second, we
identify complementarities between obstacles, namely three groups characterized
respectively by (1) lack of information and rigidities, (2) risks and costs, and (3) problems of
outlets on markets. Complementarities found between them confirm that internal human
capital and its management are key for any systemic innovation policy.

Keywords : Innovation, Complementarity, Factor analysis, Generalized residuals.

JEL - code(s). C25, D21, O32.


Complementarities between Obstacles to Innovation : Empirical
Study on a French Data Set
Fabrice Galia and Digo Legros y
ERMES - Universit Panthon-Assas Paris II

May 2002

Abstract : The economics of innovation traditionally focuses on the determinants of research-


development (R&D) and innovation. Other approaches studying obstacles to innovation seem to
have been marginalized. We concentrate on this alternative approach. First, we describe and detail
the obstacles of the 1997 French Technological Innovation survey (CIS2). We study econometrically
what types of rms experience what kinds of obstacles. Second, we identify complementarities
between obstacles, namely three groups characterized respectively by (1) lack of information and
rigidities, (2) risks and costs, and (3) problems of outlets on markets. Complementarities found
between them conrm that internal human capital and its management are key for any systemic
innovation policy.

Keywords : Innovation, Complementarity, Factor analysis, Generalized residuals.

JEL Classication : C25, D21, O32.


We would like to thanks D. Encaoua, G. Eliasson, S. Lhuillery, P. Mohnen, the participants of the Onzimes
Journes du SESAME, September 2001, Lille, the participants of Seminar ERMES, January 2002, Paris, and the
participants of Workshop Innovation, Paris I, May 2002 for their comments. All remaining errors or omissions are
ours.
y
ERMES - CNRS UMR 7017 - Universit Panthon-Assas Paris II - 12, place du Panthon - 75270 PARIS Cedex
05 - Tl. : +33 (0)1 44 41 89 90 - Fax. : +33 (0)1 40 51 81 30 - E-mail : galia@u-paris2.fr and dlegros@u-paris2.fr
1. Introduction

Innovation is the key factor of competitiveness in rms and nations. Much works has been
devoted to explain the determinants of innovation and research and development (R&D). The size,
the type of the rm, the technological opportunities, the degree of competition and the capacity
of appropriation of the benet of the innovation are the most examined factors (see Cohen et
al. (1989), Cohen (1995), Encaoua et al. (2000), Kleinknecht and Mohnen1 (2001), Crampes and
Encaoua (2001) and Crpon et al. (2000) for France). Other approaches to the study of obstacles
to R&D and to innovation seem to have been marginalized. To our knowledge, only one study on
the subject (Mohnen and Rosa (2000)) was carried out on data of Canadian service rms.
R&D is an investment activity subject to consistently uncertainty, of the beyond calculation.
Will success be here ? How much will have to be invested in order to develop a new product or
process ? Moreover, the costs of research are (sunk costs), irreversible2 (Tirole (1993)) and its
output, i.e. the knowledge is not easily transferable3 being specic in the majority of cases. Finally,
innovation involves a signicant xed cost. The risk does not disappear once the new product or
process is ready. Indeed, the consumers will make a good reception with the newborn ? If so, up to
what point ? Will the request be such sucient to cover the expenditure of R&D. Will commercial
success be experience ? In other words, a successful innovation is an innovation only if commercial
success in the market is realized (Barlet et al. (1998)).
The importance of the costs of innovation is reinforced by the lack of nancing private as well
as external. Because of the imperfections of the capital markets, an insucient amount of own
capital stocks can cause a degradation -damaging, erosion- of the conditions of nancing. These
phenomena moreover are pronounced in the case of investment in R&D (see Belin (2001), Duguet
(2001) and Lhomme (2001)). The importance of budgets and the randomness of R&D explain
why it is generally self-nancing or nanced by equity (three-quarters of the rm). Moreover,
the nancial problems do appear during the process of innovation, which we can break up into
ve phases : exploratory research, R&D strictly dened, preparation and commercial launching,
industrialization and manufacturing and nally marketing.
However, when the rm judges that the incurred risks and costs are to large, it can short them
with partners. Partnerships take various forms (i.e. R&D states or public organization subsidies
(Guillaume (1998), CGP (1999), Belin (2001), Duguet (2001), Lhomme (2001)) and cooperation
agreements (Aghion and Tirole (1995)). However, an excessive perceived economic risk involves
diculties of nding the nancing and partners necessary to the various projects of innovation.
The information collected by the future rm and its partners on the outlets (in term of sales), but
also on the technique and technology to be set up, makes it possible to evaluate the success and
the feasibility of the project (Monjon and Waelbroeck (2000)). The success of a project of R&D
thus depends largely on the reactions of the market and the technological capacities of the rm.
The innovation requires not only general knowledge but also particular and specic aptitudes in
each industry or the receiven competence (Eliasson (1990)) to commercialize a technical innovation.
1
They identify clearly (p. 17) : factor conditions (human resources, basic research infrastructure, information
infrastructure, and the supply of risk capital), supporting environment (competition, innovation incentives, presence
of clusters, and local suppliers), and demand conditions (sophisticated customers, anticipated needs).
2
Indeed, when a rm invests wrongly in a machine tool, it always has the possibility of recovering part of the cost
by the resale of the machine what is not the case of the expenditure of R&D.
3
The diversity of the activities of the rm plays an important role on this level. According to the literature, a
greater diversity of the rms activities would make it possible to use knowledge more easily resulting from the phase
of R&D.

2
The process of innovation also presupposes increased capabilities to solve problems which do not
fail to intervene. These various capacities are developing not only by learning by using or learning
by doing but also by the acquisition of skilled workers. For these reasons, the rm tends to increase
their human capital stock either by a sta training (general or specic), or by an increase in the
number of skilled workers (particularly the senior executive, the technicians, and the engineers).
This strategy remains valid during the adoption and the adaptation of new technologies resulting
from other rm (Gassmann and VonZedwitz (1998), Galia (2000))
Engineers may be essential to innovation, but are not the only actor of importance. Then is a
world of dierence between a technological innovation or a new technology and an industrial scale
production (see competence bloc theory, Eliasson (2000, pp. 12-17). The rm eager to conclude
its project will have also to engage specialists in marketing and sales engineers. The structuring,
organizing of research and the managing of engineers to achieve commercial objectives in order to
avoid dissipated eorts (Tarondeau (1994), Vinck (1991), Gaillard (1997), Gassmann and VonZed-
witz (1998), Galia (2000)) ; the marketing and commercial services will be charged to promote and
sell the new product(s) or process(es).
For these reasons, the lack of skilled workers within the rm, the low availability on the labour
market of highly qualied workers constitute not only an obstacle to innovation but also to the use
of new advanced technologies. Baldwin and Johnson (1995) showed, in a Canadian empirical study,
that the innovating rm employs a higher proportion of professional personnel and technical sta
of production than the average rm (see also Baldwin and Da Pont (1996), Baldwin and Peters
(2001)).
In order to develop the capacity of innovation, the management of the rm must be suciently
exible. Indeed, regarding research and development, the hierarchical formalism, the excessive reg-
ulation (Guillaume (1998)), the lack or the absence of communication between the research and
production units, the system of the repetitive tasks, fragmented, the deresponsabilisation of the
employees can degrade innovation capacities. In other words, division of the labour constitutes a
barrier to innovating activity and to individual creativity.
The rm can develop through varying its exibility in two ways : quantitatively or qualitatively.
The rst way is the number of employees, the number of worked hours or the remuneration. The
second one, relates to competences of the workers related to the skills level, the experience, the
polyvalency/versatility, the motivation, the autonomy level in organization. They are also to make
it possible for each member of the rm to express his new ideas in order to facilitate the appearance
of innovation such as the process innovations, which are generally the fact of production personnel.
Firm strategy also includes the imitation of competitor innovations, the absorption of new
technologies implies that organizational exibility is also necessary. Imitation also requires high
qualications to carry out reverse-engineering (at least the study of the new product), the use and
the maintenance of new machines, quality control (via quality circles), etc.
Greenan (1996) shows that there is a strong interdependence between innovation, organization
of the production and competences of the employees. According to it, the qualications within the
rm inuence its technological and organizational choices (see also Greenan and Mairesse (1999),
and Caroli et al. (2001)). A reorganization of the productive system thus passes through qualied
workers4 .
Lastly, public authorities, consumers associations and industrial lobbying groups can establish
standards or legislate and regulate the economic activity of production. A standard is a measure
4
The qualication of the managers of the rm also intervenes in organisational rigidities.

3
making it possible to evaluate the viability and the quality of a particular product. Standards might
benet the consumer, but can have negative eects on competition (creation of barriers preventing
the arrival of new rms or new products on the market).
Thus, to carry a project of innovation, the rm faces many obstacles of nancing, feasibility,
problems of information, lack of qualied or skilled personnel, resistance to innovation, rigidities.
In this article, we study these various obstacles and their possible interrelationships and com-
plementarities using the French Community Innovation Survey 1997 LInnovation Technologique
dans lIndustrie (CIS2) over the period 1994 to 1996. Initially, we carry out a detailed descriptive
analysis of the obstacles to innovation. 68% of the rm with projects of innovation delayed their
projects, and four out of ten gave up due to encountered diculties. The withdrawal of projects
comes primarily from the economic risks incurred by the rm, from the innovation costs too high
and the absence of appropriate source of nancing. The causes of delay are identical to those of
the withdrawal except the lack of qualied personnel plays a more signicant role. A Data Analy-
sis then makes it possible to set up three groups of obstacles characterized respectively by a lack
of information and rigidities, by purely economic aspects and nally by the problems of outlets
on the markets. Second, we study the possible complementarities between the obstacles using an
econometric approach in two stages. This procedure brings us in the rst stage, to regress each
obstacle to innovation using a probit model. In a second stage, we carry out an analysis of the cor-
relations of the generalized residuals resulting from the rst stage in order to establish the possible
complementarities.

2. Data

The data used for our study come from two French data sets. The rst one is the French
Community Innovation Survey 1997, LInnovation Technologique dans lIndustrie (CIS2) over
the period 1994 to 1996 carried out by SESSI 5 , results from a questionnaire sent to a representative
sample of more than 5000 industrial rms above 20 employees. This investigation belongs to the new
Community Survey6 , on technological innovation (for a detailed description, see Franois and Favre
(1998a)). The rms answer primarily on the nature of the technological innovations, the supervision
of these innovations (in terms of projects of innovation), the internal and external sources (of R&D),
the objectives of the technological innovation, the principal sources of information to innovate, the
cooperation to innovate and nally the obstacles to the projects of innovation. This last part of
the questionnaire being the objective of this paper, it will be examined in details in the following
section. The response rate is close to 85%, for more than 95% in sales turnover. The second one is
the Annual Survey on Firms 19937 (in French, Enqute Annuelle dEntreprises (EAE)), informs
us about the general characteristics of the rm like the average number of employee in order to
constitute classes of size and the line of business.
We use the answers on the nature of innovations8 (products or process technologically new or
improved) of the rm. We know also the rms implication in projects of innovation (of products
or processes technologically new or improved) which are still under development or marketing. The
rm also informs us on the failures in terms of projects of innovation. Firms having answered by
5
The SESSI is the industrial statistics department of the French Ministry of Economics, Finances and Industry.
6
Survey CIS2 is based on the Oslo Manual drawn up by the O.C.D.E., and revised in 1996.
7
The year 1993 makes it possible to avoid problems of simultaneity with the endogenous variable.
8
See Franois and Favre (1998a, p. 159).

4
the armative at least one of four preceding questions9 are qualied innovating by the SESSI.
However, only the rms that had projects of innovations in the past are likely to answer questions
relating to the obstacles to innovation. The last question is put in the following way : <Between
1994 and 1996, which obstacles with the projects of innovation your rm met ?>. The merger of
these data sets provides a sample of 1773 rms having had projects of innovations during the period
1994-199610 .
The distribution of rms according to the type of innovations appears in the table of sample
descriptive statistics in appendix. Among the industrial rms that had projects of innovation in
the past, meadows of nine out of ten innovated technologically according to the SESSI over the
three years period (1994 to 1996). The introduction on the market of new technological products
relates to 81.4% of the rms, while 69.8% state to innovate in processes. The majority of them
(61.7%) combined at the same time product and process innovations. Competitiveness is reected
by the prevalence of the product innovation. The apparent goal is to directly take market shares
to the competitor. The rms seek to attract customers sensitive to the characteristics and the
new performances of the products, and not by focusing their eorts on an innovation aiming at
reducing the production costs and the prices. Engagement in the process of innovation seems
constant since 95.9% of the rm continue, after 1996, of the projects that yet did not lead or did
not know dedication on the market. The rm remains in a dynamics of creation of innovations that
will probably lead in the future years. However, one quarter of rms underwent failures with the
projects of innovations and only a little hundred knew only failures.
The statistics of activities related to innovation are in the appendix table. The fact of carrying
out internal R&D within the rm relates to eight rm out of ten. The formalization of the projects
of innovation is thus carried out generally in-house and using specics structures gathered within
the department of R&D. However, a rm on three sub-contracts R&D, is included with another
rm of the group. We can distinguish two behaviours. The rst relates to the rms estimating
having a too weak capacity of R&D them thus preferring sub-contracting the totality of their
projects at specialized laboratories. Second is characteristic of the rm having research laboratories
operating like true autonomous entities and not marking clearly part of the rm (see Tarondeau
(1994)). The signature of cooperation agreements by half of the rm established the argument
in favour of the personnel and method, know-how, knowledge, exchange of information between
partners (undertaken or organization of research) for innovation. The sta training related directly
to the technological innovation represents an important share of the strategy of R&D since 30% of
the rm use it.
The distribution of rms by size takes again the standard classication used by the SESSI in
seven classes of size. Moreover, we use three categories of size : with the small rm (SF) from 20
to 99 employees, the medium-sized rm (MF) from 100 to 499 employees and the large rm (LF)
of 500 employees and more. This classication thus widens the traditional contrast between small
and large rms. Among the rm with projects of innovations (see table in appendix), one on three
are large rm. We observe the same proportion for the medium-sized rm. The small rms with
projects account for 38.7% of our sample.
9
<Between 1994 and 1996, 1) Introduction on the market of new (or improved) technological products for the
rm ; 2) Introduction on the market of new (or improved) technological processes for the rm ; 3) Firm having had
projects of new or improved products or processes which are still under development or marketing ; 4) Firm having
had projects of new or improved products or processes which were failures>.
10
Projects of innovations still under development or marketing, or which were failures. Let us note that the 1773
rms having had projects of innovations between 1994 and 1996 belong obligatorily to the category of the innovating
rms according to the SESSI.

5
The sectorial eects (line of business) are taken into account in terms of technological intensity.
The technological intensity of rms used by the SESSI is a typology carried out by OECD in 1994.
It is about a gathering of the industrial sectors according to intensity11 (direct and indirect) of R&D
in the production after weighting on the 10 principal Member States. Four groups are thus formed
(see table in appendix) : sectors of high technological intensity (HT), of medium high technology
(MHT), of medium low technology (MLT) and of low technology (LT). A quarter of the rms
belongs to the group of low technology (LT), the same proportion remains for the group of medium
high technology (MHT). The group of medium low technology (MLT) accounts for 38.2% of the
population with projects of innovation. Lastly, high technology (HT) represents one rm out of ten.
Competition being regarded as sharper in the growth industries such as pharmaceutical industry,
aeronautic industry and space engineering, data processing and electronic components, innovation
seems to be a condition of survival and competitiveness for the rm on the market.

3. Obstacles to innovation : descriptive analysis

The ten obstacles used by our study are included in table 1. We have a distinction between
the delayed and given up projects. Various diculties appear at the time of project of innovation
since 68% of the rm were brought to delay certain projects and four out of ten gave up some
projects. Firms prove to be more subjected to the economic constraints than technological and
organizational. ones The withdrawal of project is mainly due to excessive economic risk (OBS1)
incurred by the rm (such as the disappearance or the loss of control) and to innovation costs too
high (OBS2), lack of appropriate source of nance (OBS3) which can be attached to these obstacles.
The lack of customer responsiveness to new products and processes (OBS9) is also a reason with
the withdrawal of project. By comparison, the other obstacles are all of low importance. However,
let us note the important place of the failure of cooperation (OBS10) involving the withdrawal
and a fortiori the end of the research project in collaboration with other rm. The impact of these
obstacles is taken into account by the fact that 26.3% of the rm state to have had failures between
1994 and 1996.
For the delayed projects, having vocation to emerge in ne on an innovation (of product or
process), the frequency of the obstacles appears more signicant. We nd the two principal ob-
stacles : the economic risk (OBS1) and the costs of innovation (OBS2), but the lack of skilled
personnel (OBS5) also seem a signicant brake. The lack of customer responsiveness (OBS9), the
lack of information systems on markets and technologies (OBS6 and OBS7) are shown abruptly
like determining obstacles. For the innovation, the whole of dimensions of the process (markets,
nancing, human resources, organization) must then be taken into account and managed by the
rm.
11
After measurement of the intensities by sectors, 4 groups are made up : High Technology (intensity>8,5%),
Medium High Technology (2,6%<intensity<4,5%), Medium Low Technology (1%<intensity<2,6%) and Low Tech-
nology (intensity<1%). To obtain the connection between the groups of technological intensity and NAF level 114
French nomenclature, refer to Franois (1998, pp. 154-155).

6
Table 1 : Descriptive statistics : Obstacles to innovation (1773 rms between 1994-1996)
Delayed Given up
Obstacles Description
pro ject pro ject

OBS1 Excessive perceived economic risk 21.26% 18.73%


OBS2 Innovation costs too high 19.85% 14.66%
OBS3 Lack of appropriate source of nance 13.76% 5.75%
OBS4 Resistance of change in the enterprise (organizationnal rigidities) 16.58% 3.27%
OBS5 Lack of skilled personnel 21.15% 3.38%
OBS6 Lack of information on technologies 17.03% 5.58%
OBS7 Lack of information on markets 16.19% 4.34%
OBS8 Legislation, regulations, norms, standards 14.21% 3.72%
OBS9 Lack of customer responsiveness to new products and processes 18.84% 9.19%
OBS10 Failure of cooperation 5.25% 6.66%

Met at least one obstacle 67:96% 40:78%


(1205) (723)
Source : SESSI (1997)

By taking account the size of the rm (table 2), the obstacles most met are appreciably the same
ones as previously. Being with delayed projects, the small rm primarily face with the obstacles of
innovation costs (OBS2), economic risk (OBS1) and lack of skilled personnel (OBS5) to a lesser
extent the lack of customer responsiveness (OBS9). The medium-size rms meet more frequently
obstacles relating to the lack of skilled personnel (OBS5), economic risk and customer responsiveness
(OBS1 and OBS9). However, the excessive costs of innovation as well as the lack of information
on markets (OBS2 and OBS7) remain prevalent obstacles. Lastly, for the large rm, economic risk
(OBS1), lack of information on technologies and customer responsiveness (OBS6 and OBS9) set up
the principal group of obstacles. Come to be added the lack of skilled personnel and the innovation
costs (OBS5 and OBS2).
The frequencies of answers to the obstacles according to the size of the rm are not signicantly
dierent with critical value at the 5% level12 . The size thus does not seem discriminating factor.
However, the small rm are more sensitive to the high costs of innovation (OBS2) and to the lack of
appropriate source of nance (OBS3). The failure of cooperation (OBS10) and lack of information
on technologies (OBS6) are brakes more frequently met by the rm of large size compared to small.
Obstacles with the projects given up for the small rm remain the same ones, but to a lesser
extent. The small rm most frequently react to the excessive costs of the innovation (OBS2) and
to the economic risk (OBS1). The lack of customer responsiveness (OBS9), but especially the
lack of appropriate source of nance (OBS3) and the failure of cooperation (OBS10) come to
supplement the preceding obstacles. The problems of nancing and co-operation seem signicant
here, which is not the case for the delayed projects. The rm of average size as for them face
the same concerns : economic risk (OBS1), excessive costs of innovation (OBS2), lack of customer
responsiveness (OBS9), lack of information on technologies (OBS6) and failure of cooperation
(OBS10). Lastly, the large rm meet the same obstacles as the preceding rm except for the
obstacle relating to nancing (OBS3).

12
The test carried out is a Student test of frequencies comparison to the threshold of 5%. The law of Student is
approximated by a Normal distribution.

7
Table 2 : Obstacles distribution by size (%)
Delayed Given up
Obstacles
project project
SF MF LF SF MF LF
OBS1 20.96 20.50 22.48 13.83 19.43 24.38
OBS2 22.42 17.65 18.86 14.41 13.37 16.38
OBS3 16.16 10.70 13.90 6.55 4.99 5.52
OBS4 14.99 16.58 18.67 3.06 2.67 4.19
OBS5 20.82 23.17 19.43 3.64 3.21 3.24
OBS6 15.14 16.58 20.00 4.37 7.31 5.33
OBS7 14.41 17.65 16.95 5.09 3.21 4.57
OBS8 15.57 14.08 12.57 4.22 3.92 2.86
OBS9 16.59 20.50 20.00 7.86 8.38 11.81
OBS10 3.78 5.35 7.05 6.40 6.60 7.05

Met at least
62.74% 70.05% 72.57% 34.64% 42.07% 47.43%
one obstacle (431) (393) (381) (238) (236) (249)

N 687 561 525 687 561 525


Source : SESSI (1997)

The distribution of obstacles according to the activities related to the innovation are in tables
3 and 4. Those are as follows : to carry out internal R&D, to acquire external R&D (including with
another rm of the group), to use training related directly to innovation and nally to have coop-
eration agreements with other rm or organizations for activities of innovation. The three-quarters
of the rm using these activities state to have had at least an obstacle delaying the project(s) of
innovation (table 3). The rm doing internal R&D (72%) are prone mainly to the economic risk
(OBS1 relating to 22% of the rm), with the lack of skilled workers (OBS5), with the high costs
of innovation (OBS2) and with the lack of customer responsiveness (OBS9). The rm not doing
internal R&D generally meet the following obstacles : lack of skilled workers (OBS5 for 19% of
them), high costs of innovation (OBS2), economic risk (OBS1) and lack of customer responsive-
ness (OBS9). Being the external R&D, the major obstacles are the lack of customer responsiveness
(OBS9), the lack of skilled workers (OBS5), information on the technologies (OBS6) and nally
economic variables (costs (OBS2) and risks (OBS1)). Training to innovation can provide a re-
sponse to the lack of skilled workers (OBS5), because 25% of rms engaged in this process state to
meet this obstacle. The lack of skilled workers (OBS5) constitutes the principal obstacle, with the
economic risk (OBS1), for the rm signatories of cooperation agreements (characterising 22% of
the sample).

8
Table 3 : Obstacles to innovation in delayed pro jects (%)
Internal No internal External No external
Obstacles
R&D R&D R&D R&D
OBS1 22.26 17.55 19.38 22.11
OBS2 20.33 18.09 19.57 19.98
OBS3 14.10 12.50 11.96 15.58
OBS4 17.68 12.50 17.21 16.30
OBS5 21.76 18.88 21.92 20.80
OBS6 18.04 13.30 21.92 14.82
OBS7 17.04 13.03 17.21 15.72
OBS8 15.18 10.64 15.40 13.68
OBS9 19.69 15.69 22.64 17.12
OBS10 5.23 5.32 6.16 4.83

Met at least
72.08% 52.66% 73.37% 65.52%
one obstacle

N 1397 376 552 1221

Cooperation No cooperation
Training No training
agreements agreements
OBS1 25.05 19.61 22.10 20.45
OBS2 21.71 19.04 21.75 17.99
OBS3 15.58 12.97 13.90 13.63
OBS4 18.00 15.96 17.31 15.87
OBS5 25.42 19.29 22.10 20.22
OBS6 18.74 16.29 20.05 14.08
OBS7 16.14 16.21 17.20 15.20
OBS8 13.91 14.34 14.24 14.19
OBS9 20.22 18.23 21.41 16.31
OBS10 4.64 5.51 6.95 3.58

Met at least
72.91% 65.80% 75.40% 60.67%
one obstacle

N 539 1234 878 895


Figures in bold are the case where frequency equality is rejected at 5%.
Source : SESSI (1997)

9
Table 4 : Obstacles to innovation in given up projects (%)
Internal No internal External No external
Obstacles
R&D R&D R&D R&D
OBS1 20.11 13.56 21.74 17.36
OBS2 15.75 10.64 15.58 14.25
OBS3 5.65 6.12 6.16 5.57
OBS4 3.22 3.46 3.26 3.28
OBS5 3.15 4.26 4.53 2.87
OBS6 5.80 4.79 6.88 5.00
OBS7 4.22 4.79 4.35 4.34
OBS8 3.44 4.79 3.44 3.85
OBS9 9.09 9.57 9.60 9.01
OBS10 7.37 3.99 6.88 6.55

Met at least
43.81% 29.52% 47.28% 37.84%
one obstacle

N 1397 376 552 1221

Cooperation No cooperation
Training No training
agreements agreements
OBS1 19.29 18.48 22.21 15.31
OBS2 16.33 13.94 14.92 14.41
OBS3 5.38 5.92 5.92 5.59
OBS4 2.78 3.48 3.53 3.02
OBS5 3.34 3.40 2.96 3.80
OBS6 5.38 5.67 6.72 4.47
OBS7 4.08 4.46 4.56 4.13
OBS8 3.71 3.73 4.33 3.13
OBS9 10.20 8.75 10.02 8.38
OBS10 7.42 6.32 8.88 4.47

Met at least
43.04% 39.79% 47.61% 34.08%
one obstacle

N 539 1234 878 895


Figures in bold are the case where frequency equality is rejected at 5%.
Source : SESSI (1997)

The withdrawal of innovation project for the rms engaged in internal R&D seems to be mainly
due to the economic risk (OBS1), the costs of innovation (OBS2) and to a lesser extent with the lack
of customer responsiveness (OBS9) (see table 4). Others obstacles being slightly met. These three
principals obstacles are the same ones for the rm doing external R&D and for the rm signatories
of cooperation agreements. The obstacles mainly met by the rm using the training to innovate

10
are the lack of skilled workers (OBS5), risk (OBS1), costs (OBS2) and customer responsiveness
(OBS9).

4. Complementarities between obstacles to innovation

The statistical analysis carried out previously revealed that obstacles to innovation seem to
be related to the rm size, the technological intensity and the activities in favour of innovation.
This analysis made it possible to highlight the obstacles most often met. In this section, we study
possible interrelationships and complementarities between the various obstacles. We seek to know
which obstacles go hand in hand, and which are the possible clustering. In order to answer this
question, we initially examine the binary correlations between the answers to obstacles. Then, we
carry out a Factor Analysis in order to determine clusters of obstacles to innovation.

4.1. Binary correlations analysis

Answers to obstacles are binary qualitative variables (1 if the rm considers to have met the
obstacle between 1994-1996, 0 if not), we carry out a Chi-2 test (2) of the obstacles independence.
The Chi-2 statistic do not provide a measurement of the intensity of the statistical link between the
obstacles, nor an indication on the direction even of the relation between two obstacles. For this
reason, we use the coecient of contingency or association C 13 . The correlations between obstacles
appear in table 5 and 6.
Concerning the delayed projects (table 5), the rm seem to answer positively and jointly the
obstacles related to economic risk, too high costs of innovation and lack of appropriate source of
nance (OBS1 with OBS3). The two obstacles related to the internal management of the rm,
organizational rigidities and lack of qualied workers (OBS4 and OBS5) seem to go hand-in-hand.
The brakes due to the lack of information on technologies, markets (OBS6 and OBS7) and to the
legislation, regulation, norms and standards (OBS8) are correlated. The external environment is
taken into account with the signicant relation between the lack of information on the markets and
the lack of customer responsiveness to new products and processes (OBS7 and OBS9). The others
obstacles comparatively slightly appear to be correlated, in particular, the excessive perceived
economic risk and the lack of skilled workers (OBS1 and OBS5). The obstacle relating to the
source of nance does not seem to be related to the lack of customer responsiveness (OBS3 and
OBS9). The four prevalent correlations invites us to perform the following gathering : the lack of
skilled workers and lack of information on technologies (OBS5 and OBS6) ; the lack of information
on markets and the lack of customer responsiveness to new products and processes (OBS7 and
OBS9) ; organizational rigidities and the lack of skilled personnel (OBS4 and OBS5) and nally
those related to costs of innovation and source of nance (OBS2 and OBS3).
It is not surprising to nd strong correlations between obstacles being followed in the order
of the questions in the survey (what corresponds to the diagonal of table 6). This can reect
an implicit will of classication of the obstacles that can introduce a bias into the answers. A
possible consequence of this is that in general the correlations decrease while moving away from
the diagonal. In spite of that, certain correlations seem signicant apart from the diagonal such as
13
p
The coecient of contingency C; is between 0 and 1; comes from 2. More precisely, C = 2 =2 + N, where
N is the number of observations (see Kendall and Stuart (1973) and Miravete and Pernias (2000)). Henceforth, we
will speak about correlation between obstacles.

11
excessive perceived economic risk and lack of customer responsiveness (OBS1 and OBS9), costs of
innovation and failure of cooperation (OBS2 and OBS10) and nally lack of appropriate source of
nance and failure of cooperation (OBS3 and OBS10).

Table 5 : Correlation between obstacles to innovation in delayed projects


OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9
OBS1 1
OBS2 0:21 1
OBS3 0:15 0:24 1
OBS4 0:06 0:04 0:18 1
(0:0157) (0:0888)
OBS5 0:06 0:08 0:17 0:26 1
(0:0094) (0:0010)
OBS6 0:0008 0:08 0:10 0:22 0:27 1
(0:9735) (0:0009)
OBS7 0:09 0:05 0:12 0:22 0:19 0:22 1
(0:0520)
OBS8 0:10 0:09 0:13 0:22 0:14 0:19 0:26 1
OBS9 0:13 0:08 0:08 0:15 0:11 0:16 0:27 0:19 1
(0:0006) (0:0008)
OBS10 0:09 0:12 0:21 0:16 0:13 0:15 0:16 0:18 0:16
Chi-2 statistic is 0.0001 for all pairs, except mention in brackets
Source : SESSI (1997)

For the given up projects, the correlations between obstacles are given in table 6. However,
the Chi-2 test is not valid for all considered pairs14 . A solution is to gather the modalities of
responses to obstacles in order to obtain a sucient number of theoretical observations. However,
this method does not have possible interpretation in the case of two dichotomic variables. In term
of correlation, the economic risk remains associated with the costs of innovation and the lack of
appropriate source of nance (OBS1 to OBS3). Apart from the diagonal, the correlated couples are
the following : excessive perceived economic risk and lack of customer responsiveness (OBS1 and
OBS9), the excessive costs of innovation and the failure of cooperation (OBS2 and OBS10) and
nally nancing and the failure of cooperation (OBS3 and OBS10). The signature of cooperation
agreements (OBS10) appears to be in direct relationship with the lack of information on technologies
(OBS6), which conrms the idea according to which an alliance seems favourable in the case of
insuciency or technological incapacity.
Caution is required in the interpretation of the binary correlations between obstacles, the cor-
relation between two obstacles does not inform us about the link of causality between them. A
third variable can strongly inuence the binary relation. In order to know the gathering between
obstacles, we carry out in the next section a Factor Analysis (FA).
14
Number of theoretical observations for one of the four possibles responses combinaisons ((0,0), (0,1), (1,0) and
(1,1)) is lower than 5.

12
Table 6 : Correlation between obstacles to innovation in given up projects
OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9
OBS1 1
OBS2 0:26 1
OBS3 0:13 0:23 1
OBS4 0:11 0:11 0:27 1
OBS5 0:06 0:16 0:14 0:21 1
OBS6 0:07 0:11 0:15 0:17 0:22 1
(0:0024)
OBS7 0:12 0:11 0:14 0:16 0:16 0:15 1
OBS8 0:13 0:16 0:14 0:19 0:14 0:17 0:17 1
OBS9 0:18 0:14 0:11 0:11 0:08 0:09 0:15 0:21 1
(0:0007)
OBS10 0:17 0:17 0:19 0:14 0:08 0:16 0:15 0:18 0:17
Chi-2 statistic is 0.0001 for all pairs, except mention in brackets

Chi-2 test non valid : number of theoretical observations for one of the four possible responses combinations
((0,0), (0,1), (1,0) et (1,1)) is lower than 5.
Source : SESSI (1997)

4.2. Factor Analysis

Following the paper of Mohnen and Rosa (2000), we seek to form groups (or clusters) of obstacles
to innovation. These groups are formed using the following criterion : their linear combination must
explain as much as possible the variance of the responses to the obstacles to innovation.
The variables resulting from the survey are of qualitative nature (especially binary qualitative
variables). Thus, we carry out a Factor Analysis (FA). The FA is a descriptive method, which makes
it possible to study the connections between qualitative variables. The algorithm of the FA was
applied to the complete disjunctive table resulting from the table of condensed coding. We gave
the statute of active element to the set of the ten obstacles. Since the active questions have both
methods, no obstacle has a priori more importance than another.
The eigenvalues resulting from the FA applied to the complete disjunctive data set are very
pessimistic measurements of extracted information15 . For this reason, we choose the square of the
eigenvalues that are the eigenvalues of the FA applied to the Burts table associated with our data.
The rates of inertia for the obstacles with the delayed project are the following :

15
Lebart et al. (1997).

13
Table 7 : Factor Analysis in obstacles to innovation in delayed projects
Eigenvalues Variance (%) Cumulative Variance (%)
0:0575 46:26 46:26
0:0158 12:71 58:97
0:0107 8:61 67:58
0:0079 6:36 73:93
0:0068 5:47 79:40
0:0064 5:15 84:55
0:0056 4:51 89:06
0:0049 3:94 93:00
0:0044 3:54 96:54
0:0043 3:46 100:00
Source : SESSI (1997)

In the framework of the FA, the quality of the adjustment provided by the rst factorial plan
is satisfactory (see graphic 7.1 in appendix) since this rst two dimensional factor space restores
approximately 59% of total inertia. The criterion of Catell16 (called sometimes algorithm of the
elbow) invites us to retain the rst three factorial axes.
The Factor Analysis, on obstacles in delayed projects, conrms the results obtained by the
binary correlations. In table 8, the rst factorial axis gathers the problems of rigidities as well
internal (organizational rigidities (OBS4) and lack of qualied personnel (OBS5)) that external
(legislation, regulation, standards, standards (OBS8)) and lack on information on technologies
(OBS6) and markets (OBS7). With regard to the second axis, the economic variables : economic
risk (OBS1), the costs of innovation (OBS2) and the lack of appropriate source of nance (OBS3)
which take part in the formation of this factor. For the third axis, we record the importance of
the lack of customer responsiveness (OBS9) and the contribution of the lack of skilled personnel
(OBS5). The lack of appropriate source of nance (OBS3) also intervenes in the determination of
this axis.

16
The procedure seeks to detect an elb ow on the histogram of the eigenvalues to facilitate the choice of the number
of factorial axes to use. In dierences second, the elbow i is : i = (i+1 i ) (i i1) with i the eigenvalue
of the axis i. In dierences third, the elbow is : i = ( i+2 i1) 3 (i+1 i ).

14
Table 8 : Contribution (%) to axis is variance - Delayed projects
Axis 1 Axis 2 Axis 3
Obstacles Rigidities and Risks, Costs and Customer
Information Source of Finance Responsiveness

OBS1 3.7 26.1 8.4


OBS2 4.6 33.7 1.1
OBS3 9.3 15.6 11.7
OBS4 13.2 4.3 5.5
OBS5 11.6 3.2 19.3
OBS6 11.6 8.8 4.8
OBS7 13.9 5.1 11.6
OBS8 12.5 1.2 6.2
OBS9 9.7 0.5 31.4
OBS10 9.9 1.5 0.1
Source : SESSI (1997)

In conclusion, we can release three groups of obstacles in delayed projects of innovation. A


rst which gathers rigidities and the lack of information, a second made up of purely said eco-
nomic variables (risk, costs and nance) and nally a third one dominated by the lack of customer
responsiveness.
We apply the same methodology as previously in order to establish groups of obstacles for the
given up projects of innovation. The eigenvalues obtained are as follows :

Table 9 : Factor Analysis in obstacles to innovation in given up projects


Eigenvalues Variance (%) Cumulative Variance (%)
0:0576 46:60 46:60
0:0128 10:37 56:96
0:0098 7:96 64:92
0:0081 6:58 71:50
0:0073 5:94 77:44
0:0070 5:67 83:11
0:0059 4:73 87:85
0:0057 4:65 92:49
0:0053 4:32 96:82
0:0039 3:18 100:00
Source : SESSI (1997)

Once again by helping us on the criterion of Catell, we retain three factorial axes (table 10
below and graph 7.2 in appendix).

15
Table 10 : Contribution (%) to axis is variance - Given up projects
Axis 1 Axis 2 Axis 3
Obstacles Costs, Source of Finance Risks and Source of Finance
and Rigidities Skilled Personnel and Customer

OBS1 8.5 26.1 3.1


OBS2 11.1 9.5 18.1
OBS3 12.0 0.5 22.8
OBS4 11.5 10.8 3.3
OBS5 8.4 20.8 0.6
OBS6 9.0 14.9 1.4
OBS7 9.0 1.4 14.2
OBS8 11.7 0.0 14.2
OBS9 8.3 11.8 21.4
OBS10 10.5 4.2 0.9
Source : SESSI (1997)

For the given up projects, the rst axis is characterized at the same time by obstacles related
to the source of nance of the innovation projects, problems of organizational rigidities and failures
of cooperation. More precisely, obstacles of nancial nature gather the too high costs of innovation
(OBS2) and the absence of suitable source of nancing (OBS3). Rigidities are composed of organi-
zational rigidities (OBS4), legislation, regulation, norms and standards (OBS8). Lastly, the failures
of cooperation (OBS10) also take part in the formation of this factor. We notice that the obstacles
of nancial nature (OBS 2 and OBS3) also intervene in the formation of the third axis, where their
contributions are higher. The second axis gathers the following obstacles : the excessive perceived
economic risk (OBS1), which most strongly contributes to its formation, the lack of skilled personnel
(OBS5) and lack of information on technologies (OBS6). The third axis concentrates the obstacles
of nancial nature (OBS2 and OBS3), the lack of customer responsiveness to new products and
processes (OBS9) and lack of information on markets (OBS7).
The implementation of a typology makes it possible to conrm these gathering (see Galia and
Legros (2002)). Moreover, our results are in line with those of Mohnen and Rosa (2000) obtained
on a sample of Canadian service rms. The authors retain three groups characterized respectively
by : high risks and costs ; risks and lack of nance ; and nally by risks and internal resistance.

5. Complementarities analysis between obstacles to innovation : an econometric


approach

The empirical literature studying the complementarities between various strategies17 proposes
two types of approaches18 (see Athey and Stern (1998)). The rst studies the rms productivity
and consists in modelling (using several techniques) a measurement of rms productivity with a
17
i.e. if doing more of any subset of them increases the returns from doing more of any subset of the remaining
strategies, we thus have mutual reinforcement between these strategies. It provides a way to capture the intuitive
ideas of synergies and systems eects, i.e. that the whole is more than the sum of the parts.
18
A third one, using the theorical concept of lattices and supermodularity, introducing in economics by Milgrom and
Roberts (1990), see also Vives (1999) and Topkis (1998), and Mohnen and Rller (2000) for an empirical application.
Specially, a lattice is a partially ordered set where each pair of elements fx; yg has a least upp er bound, noted x ^ y;

16
set of regressors, including the interactions eects between several practices or strategies, inter-
preted like parameters of complementarities. The most convincing example of the application of
this methodology is the article of Ichniowski, Shaw and Prennushi (1997) (see also Becker and
Huselid (1998), Lhuillery (1999), Laursen and Foss (2000) and Leiponen (2000a at 2000b)).
The second approach tests the positive correlation between the various practices or strategies.
Arora and Gambardella (1990, 1996) show that certain strategies of external linkages, of large
chemical and pharmaceutical rm, are indeed complementary. The study focused on four types of
linkages : agreements with other rms, agreements of research with the universities, investment
in the stock of capital of new biotechnological rms (NBF) and acquisitions of NBF. By also
applying this methodology, Colombo and Mosconi (1995) highlights complementarities between
the application of new technological processes on the one hand and the organizational innovations
and managerial on the other hand. Laursen and Mahnke (2001) study, with the same method, the
complementarities between the human-resources practices in Denmark. Another study concerning
the exibility of the production and the innovation is Jung (2001).
The approach of Arora and Gambardella (1990) proves to be advantageous when the value of the
complementarities does not appear directly testable, because the intensity even of the strategies
remains not easily measurable. In this case, we are constrained to test a signicant number of
possible combinations between the various strategies. If, to take again the example of the authors,
an increase in the internal competence of the rm, due to an acquisition of NBF, increases the
marginal value of the links with the universities (equivalent to saying that they are complementary),
it seems intuitive to expect to observe the fact that the rm which makes more acquisitions also
tends to have a greater number of interactions with universities. In other words, if two strategies
are complementary, we must expect a positive correlation between them. Our preceding discussion
on the obstacles to innovation is in agreement with this.
However, we must also take into account the eects of the indirect reactions by the intermediary
of the other strategies. Let us suppose, for example, that x1 and x2 are complements, but that x3
is a substitute of x1 and a complement of x2 . It is then possible that x1 and x2 move in opposite
direction if both are more strongly connected to x3 than one the other. The relation between these
strategies and a set of rm-specic characteristics, the environment or the institutional framework,
thus do not have to be ignored. This implies that we have to account this kind of factors. This
methodology is developed in order to show that if the external strategies of links are complementary,
then covariance between any strategies, conditionally with a set of characteristics of the rm, proves
to be positive. This model primarily seems a formalized way to take into account the two problems
discussed above and provides us a set of conditions according to which the complementary strategies
will positively be correlated.
The application to our study of this procedure in two stages, leads us initially to regress the
ten obstacles with the innovation by a set of common exogenous variables (variables in table 11
below). We estimate a probit model explaining each obstacles (tables 11 and 13). In the second
time, we carry out a correlation analysis of the generalized residuals resulting from the rst stage
in order to reach some conclusions concerning whether or not our ten obstacles can be seen to be
complementary, i.e. depends one on the other or is reinforced mutually (tables 12 and 14).

and a greatest lower bound, noted x _ y: Two objects x and y are then called complementary if the corresponding
payo function f : Rn ! Rn is sup ermodular if for all x; y 2 Rn ; f (x) f(x ^ y) f (x _ y) f (y):

17
Table 11 : Probit model explaining obstacles to innovation in delayed projects
OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9 OBS10
Intercept ? 0. 7852 ? 0. 8197 ? 1. 0259 ? 1. 1784 ? 0. 9451 ? 1. 2260 ? 1. 1962 ? 1. 3446 ? 1. 0644 ? 1. 7419
0.0993 0.1000 0.1088 0.1103 0.1010 0.1106 0.1108 0.1181 0.1053 0.1543

[50;99] empl. ? 0. 2168 ? 0. 2830 ? 0. 2051 0. 0523 ? 0. 0413 0.0750 ? 0. 0498 ? 0.0527 ? 0. 0656 ? 0. 0113
0.1117 0.1104 0.1187 0.1187 0.1103 0.1184 0.1206 0.1191 0.1166 0.1811

[100;249] empl. ? 0. 1018 ? 0. 3430 ? 0. 3609 0. 0380 ? 0. 0166 ? 0.0892 0. 2405 ? 0.0709 0. 1318 0. 1240
0.1086 0.1112 0.1235 0.1182 0.1092 0.1213 0.1140 0.1195 0.1116 0.1739

[250;499] empl. ? 0. 2124 ? 0. 3639 ? 0. 3841 0. 0675 0. 0434 0.1095 ? 0. 1398 ? 0. 2658 ? 0. 0431 0. 1733
0.1192 0.1198 0.1340 0.1261 0.1164 0.1243 0.1317 0.1339 0.1232 0.1802

[500;999] empl. ? 0. 1567 ? 0. 3987 ? 0. 3245 0. 0886 ? 0. 1712 0.1418 0. 0155 ? 0. 2832 ? 0. 0531 0. 0820
0.1163 0.1183 0.1291 0.1241 0.1187 0.1219 0.1250 0.1302 0.1218 0.1840

[1000;1999] empl. ? 0. 0330 ? 0. 2979 ? 0. 2099 0. 1937 ? 0. 1883 0.1078 ? 0. 0517 ? 0. 4998 ? 0. 0632 0. 5945
0.1427 0.1457 0.1590 0.1507 0.1493 0.1501 0.1586 0.1788 0.1510 0.1917

2000 empl.
0. 0225 ? 0. 3515 ? 0. 0389 0. 1743 ? 0. 00947 ? 0.1815 0. 1939 ? 0.1335 0. 2131 0. 0512
0.1641 0.1703 0.1760 0.1764 0.1675 0.1856 0.1736 0.1809 0.1672 0.2587
and more

Medium Low
? 0. 1821 0. 0187 ? 0. 0239 ? 0. 0487 0. 0962 ? 0.00355 0. 0365 0. 1738 ? 0. 00933 ? 0. 0970
0.0872 0.0900 0.0989 0.0931 0.0878 0.0940 0.0956 0.1050 0.0905 0.1352
Tech

Medium High
? 0. 0964 0. 1168 0. 1162 0. 0685 0. 0738 0.0794 0. 2266 0. 4700 0. 1005 0. 0749
0.0953 0.0979 0.1066 0.1004 0.0967 0.1015 0.1023 0.1102 0.0980 0.1417
Tech

High Tech ? 0. 1855 0. 0326 0. 0417 ? 0. 3091 ? 0. 2401 ? 0.0781 ? 0. 0563 0. 5838 ? 0. 3950 0. 0785
0.1330 0.1353 0.1472 0.1505 0.1414 0.1418 0.1470 0.1424 0.1506 0.1888

Internal R&D 0. 2043 0. 0949 0. 1204 0. 1995 0. 0646 0.0904 0. 1207 0. 2040 0. 0725 ? 0. 1545
0.0921 0.0921 0.1011 0.0990 0.0907 0.0981 0.0986 0.1041 0.0945 0.1341

External R&D ? 0. 1722 ? 0. 0641 ? 0. 1578 ? 0. 0370 ? 0. 0129 0. 1870 ? 0. 00420 0.0787 0. 1263 0.00899
0.0806 0.0811 0.0906 0.0838 0.0794 0.0816 0.0842 0.0876 0.0803 0.1163

Training 0. 1837 0. 1153 0. 1400 0. 0412 0. 2160 0.0195 ? 0. 0526 ? 0.0635 0. 0132 ? 0. 1631
0.0752 0.0770 0.0841 0.0799 0.0751 0.0797 0.0816 0.0855 0.0782 0.1175

Coop. Agreements 0. 0720 0. 1930 0. 0528 0. 0103 0. 0697 0.1525 0. 0532 ? 0.0389 0. 1488 0. 2847
0.0747 0.0759 0.0833 0.0790 0.0746 0.0788 0.0796 0.0832 0.0770 0.1136

Likelihood 26.10 24. 76 23.44 15. 83 21.50 28. 59 23. 47 44.87 30. 36 26.96
Ratio

%correct 56.5 57. 7 58.8 56. 3 56.5 58. 6 57. 8 62.0 58. 8 64.6
predictions
N 1773 1773 1773 1773 1773 1773 1773 1773 1773 1773

Between brackets are standard deviation. In bold, signicant value at the 10% level.
Source : SESSI (1997)

18
Table 12 : Correlation among generalized residuals - Obstacles to innovation in delayed projects
OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9
OBS1 1
OBS2 0:21 1
OBS3 0:14 0:24 1
OBS4 0:06 0:04 0:19 1
(0:0174) (0:0683)
OBS5 0:06 0:07 0:17 0:27 1
(0:0174) (0:0018)
OBS6 0:003 0:08 0:11 0:23 0:29 1
(0:8859) (0:0006)
OBS7 0:09 0:05 0:12 0:23 0:19 0:23 1
(0:0002) (0:0523)
OBS8 0:10 0:09 0:13 0:23 0:15 0:20 0:27 1
(0:0002)
OBS9 0:13 0:08 0:08 0:15 0:11 0:16 0:28 0:20 1
(0:0011) (0:0007)
OBS10 0:10 0:13 0:22 0:17 0:14 0:15 0:17 0:20 0:16
Chi-2 statistic is 0.0001 for all pairs, except mention in brackets.
Source : SESSI (1997)

Table 11 shows the results of the ten probit models explaining obstacles to innovation in delayed
projects. The principal results are as follows : compared to the size class of reference (20 to 49
employees), the other classes of size inuence signicantly and negatively the rst three obstacles
(excessive perceived economic risk, innovation costs too high and lack of appropriate source of
nance) and to a lesser extent variable OBS8 (legislation, regulation, norms and standards). The
small rm thus meet in a more intense way the three rsts obstacles. The rm with 100 to 249
employees have a probability more raised to meet a lack of information on markets (OBS7). The
large rm (1000 to 1999 employees) encounter signicantly and positively the failure of cooperation
(OBS10). The rm engaged in an activity of internal R&D are more sensitive to obstacles related to
the economic risk (OBS1), organizational rigidities (OBS4) and nally to the legislation, regulation,
norms and standards (OBS8). The external R&D makes it possible signicantly to reduce the
economic risk related to the innovation (OBS1) and the lack of appropriate source of nance
(OBS3). However, the externalisation of research increases the lack of information on technologies
(OBS8).
From these estimates, we calculated the correlations between the generalized residuals of each
equation. In table 12, all the paire-wise complementarities are signicant (with an exception).
Most marked are located on the diagonal. In particular, the complementarities arise between the
obstacles related to the excessive perceived economic risk (OBS1) and the innovation costs too high
(OBS2) ; organizational rigidities (OBS4) and lack in skilled personnel (OBS5) as well as the lack of
information on technologies (OBS6). Let us note also the strong complementarity between the lack
of information on the markets (OBS7) and what we can describe as administrative or institutional
obstacles (OBS8). Concerning the element out of the diagonal, the lack of appropriate source of
nance (OBS3) are complementary with the failure of cooperation (OBS10).

19
Table 13 : Probit model explaining obstacles to innovation in given up projects

OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9 OBS10
Intercept ? 1. 1793 ? 1. 1769 ? 1. 4466 ? 1. 7550 ? 1. 6943 ? 1. 7218 ? 1. 6732 ? 1. 7586 ? 1. 3987 ? 1. 8676
0.1099 0.1128 0.1356 0.1674 0.1618 0.1528 0.1526 0.1602 0.1265 0.1554

[50;99] empl. 0.1188 0.0142 ? 0. 0538 ? 0.00068 0. 1263 0. 1467 0. 1807 0. 0994 0.1716 0. 1198
0.1208 0.1204 0.1514 0.1932 0.1815 0.1706 0.1614 0.1737 0.1413 0.1531

[100;249] empl. 0.0760 ? 0.1334 ? 0. 0478 0.0334 0. 0314 0. 1960 ? 0. 2251 0. 00875 0.0138 ? 0. 00754
0.1210 0.1236 0.1507 0.1924 0.1882 0.1663 0.1903 0.1816 0.1486 0.1566

[250;499] empl. 0. 3902 ? 0.0322 ? 0. 4008 ? 0.2149 ? 0. 0490 0. 3665 ? 0. 0369 ? 0. 0305 0. 2465 ? 0. 1115
0.1223 0.1294 0.1882 0.2308 0.2125 0.1682 0.1886 0.1956 0.1492 0.1691

[500;999] empl. 0. 3002 ? 0.0664 ? 0. 1249 0.1800 ? 0. 1478 0. 0806 ? 0. 1316 ? 0. 2037 0. 3265 ? 0. 1225
0.1227 0.1285 0.1646 0.1940 0.2214 0.1819 0.1941 0.2060 0.1459 0.1662

[1000;1999] empl. 0. 4208 0.0581 ? 0. 2088 0.1500 ? 0. 0462 ? 0. 3294 0. 1325 ? 0. 2929 0.2318 ? 0. 0814
0.1471 0.1557 0.2134 0.2402 0.2625 0.2727 0.2168 0.2756 0.1827 0.2011

2000 empl.
0. 5143 0.2398 ? 0. 1167 ? 0.0138 0. 4403 0. 4808 0. 2602 ? 0. 1167 0. 4986 ? 0. 0292
and more 0.1671 0.1736 0.2385 0.2989 0.2524 0.2215 0.2403 0.2778 0.1961 0.2214

Medium Low
? 0. 1937 ? 0. 1674 ? 0. 1456 ? 0. 2704 ? 0. 0822 ? 0. 2381 ? 0. 0236 ? 0. 0375 ? 0.0501 ? 0. 1853
0.0914 0.0969 0.1240 0.1555 0.1452 0.1261 0.1354 0.1493 0.1087 0.1278
Tech

Medium High
? 0.0937 0.0345 ? 0. 0844 ? 0.0426 ? 0. 1771 ? 0. 1989 ? 0. 0981 ? 0. 00935 ? 0.0794 0. 1395
0.0986 0.1028 0.1361 0.1598 0.1677 0.1373 0.1530 0.1647 0.1200 0.1286
Tech

High Tech ? 0.1047 ? 0.1158 ? 0. 0481 0.0594 ? 0. 1571 ? 0. 2414 ? 0. 0854 0. 4155 ? 0.1132 ? 0. 00190
0.1336 0.1450 0.1862 0.2083 0.2339 0.1918 0.2092 0.1903 0.1635 0.1780

Internal R&D 0. 1649 0. 2535 0. 0117 ? 0.0188 ? 0. 1346 0. 0496 ? 0. 0365 ? 0. 1629 ? 0.1146 0. 2799
0.0974 0.1033 0.1272 0.1545 0.1477 0.1347 0.1379 0.1420 0.1105 0.1387

External R&D 0.0237 0.0149 0. 0832 ? 0.0158 0. 3147 0. 1335 0.000385 ? 0. 0476 ? 0.0256 ? 0. 1230
0.0809 0.0865 0.1135 0.1387 0.1343 0.1129 0.1262 0.1346 0.0988 0.1092

Training ? 0.0826 0.0570 ? 0. 0440 ? 0.1244 ? 0. 00416 ? 0. 0864 ? 0. 0240 0. 0344 0.0555 0. 0294
0.0788 0.0824 0.1113 0.1368 0.1336 0.1127 0.1222 0.1282 0.0939 0.1038

Coop. Agreements 0. 1322 ? 0.0504 0. 0679 0.0692 ? 0. 1786 0. 1534 0. 0565 0. 1984 0.0438 0. 3728
0.0775 0.0820 0.1074 0.1302 0.1326 0.1102 0.1185 0.1246 0.0934 0.1049

Likelihood 44. 82 19.28 7.57 9. 21 14.25 23. 85 8. 92 14. 19 13. 44 29.51


Ratio

%correct 60. 9 56.7 56. 4 58. 8 61.6 62. 6 57. 4 61. 3 57. 4 63.5
predictions

N 1773 1773 1773 1773 1773 1773 1773 1773 1773 1773

Between brackets are standard deviation. In bold, signicant value at the 10% level.
Source : SESSI (1997)

20
Table 14 : Correlation among generalized residuals - Obstacles to innovation in given up pro jects
OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9
OBS1 1
OBS2 0:27 1
OBS3 0:14 0:24 1
OBS4 0:11 0:11 0:28 1
OBS5 0:06 0:16 0:14 0:21 1
(0:0094)
OBS6 0:06 0:11 0:15 0:18 0:22 1
(0:0080)
OBS7 0:12 0:11 0:14 0:16 0:15 0:15 1
OBS8 0:14 0:16 0:16 0:22 0:14 0:17 0:20 1
OBS9 0:19 0:14 0:11 0:11 0:08 0:09 0:15 0:21 1
(0:0006) (0:0001)
OBS10 0:17 0:17 0:20 0:14 0:08 0:16 0:15 0:18 0:18
(0:0008)
Chi-2 statistic is 0.0001 for all pairs, except mention in brackets.
Source : SESSI (1997)

For the obstacles to innovation in given up projects, we apply the same methodology in two
stages. Table 13 shows the results of the estimation of each obstacle by a probit model. The
economic risk (OBS1) is inuenced to a signicant degree and increases with the size of the rm.
The large rm of more than 2000 employees face a lack of skilled personnel (OBS5), with a lack
of information on technologies (OBS6) and with a lack of customer responsiveness (OBS9). The
economic risk (OBS1), the innovation costs too high (OBS2) and the failure of cooperation (OBS10)
depend positively on the implication of the rm in a phase of internal R&D. The activity of external
R&D explains in a positive way the lack of skilled personnel (OBS5).
The complementarities between the obstacles which have occurred at the time of the given
up projects appear in table 14. The diagonal reports, again, the paire-wise complementarities i.e.
the highest coecients of correlation between the generalized residuals. Let us highlight, the com-
plementarity between the obstacles related to the economic risk (OBS1) and the excessive costs
(OBS2). Moreover, this latter (OBS2) is complementary with the lack of appropriate source of -
nance (OBS3). Organizational rigidities (OBS4) and the administrative obstacles (OBS8) go hand
in hand. Lastly, we nd complementarity between the lack of appropriate source of nance (OBS3)
and the failure of cooperation (OBS10).
These results tend to show that the paire-wise obstacles complementarities belong to the same
problem thus forming a system with whole share. Thus those plead in favour of the system of
policies of innovation. Thus, the political solutions suggested and implemented in order to ght the
obstacles to innovation should not consider the latter separately but overall, as an integral part of
a unit whose various components are in a reciprocal relation of dependence.

21
6. Conclusion

Over the period 1994-1996, the French manufacturing rms gave up their projects of innovation
primarily for reasons related to the excessive perceived economic risk, the innovation costs too
high and the lack of appropriate source of nance. The delays in projects of innovation are being
explained by the same obstacles to which is added the lack of skilled personnel. Whatever the size
of the rm, the perception of the obstacles to innovation remains identical. The smaller the more
sensitive it is to the costs of the innovation and lack of appropriate sources of nance.
The analysis of the binary correlations highlights interrelationships between obstacles to innova-
tion. Indeed, the excessive perceived economic risk, too high innovation costs and lack of appropriate
source of nance are correlated. In the same way, organizational rigidities and the lack of skilled
personnel go hand in hand. Lastly, the legislation, regulations, norms, standards and the lack of
information (on the markets and on technologies) show signicant coecients of association.
The Factor Analysis conrms the existence of groups forming obstacles to innovation. For
the delayed projects, the groups of obstacles are identied without diculties. The rst group
of obstacles represents the lack of information and rigidities (internal and external). The second
consists of economic variables (excessive perceived economic risk, innovation costs too high and lack
of appropriate source of nance). The last one is dominated by the lack of customer responsiveness
to the new products and processes. For the given up projects, the rst group is characterized at
the same time by obstacles related to the projects nancing, problems of rigidities and failures
of cooperation. The second group includes the following obstacles : the economic risk, the lack of
skilled personnel and lack of information on technology. Finally, obstacles of nancial nature, lack
of customer responsiveness to new products and processes and lack of information on the markets
form the last group.
Our results show the existence of complementarities between obstacles to innovation. Moreover,
the excessive perceived economic risk and too high innovation costs ; organizational rigidities and
lack of skilled personnel as well as the lack of information on technologies ; lack of information
on markets and institutional obstacles ; the lack of appropriate source of nance and the failure
of cooperation are complementary. Hence, policies of technological innovation demand a systemic
approach capable of capturing all the important interactive factors that put up obstacles to inno-
vation. Taking into account the obstacles and the application of the solutions must be made in a
total way by holding account of the established complementarities.

22
7. Annexes

7.1. Descriptive statistics


Firms with
Between 1994 and 1996,
innovation pro jects

Firms with innovation projects still


95:94%
under development or marketing (1701)
Firms with innovation projects
26:28%
which were failures (466)

Internal R&D 78:79%


(1397)
External R&D
31:13%
(include a rm of the group) (552)
Training for innovation 30:40%
(539)
Cooperations agreements 49:52%
(878)

20-49 employees 22:28%


(395)
50-99 employees 16:47%
(292)
100-249 employees 17:43%
(309)
250-499 employees 14:21%
(252)
500-999 employees 16:13%
(286)
1000-1999 employees 7:95%
(141)
2000 employees and more 5:53%
(98)

Small rms (SM)


38:75%
20-99 employees (687)
Medium Firms (MF)
31:64%
100-499 employees (561)
Large Firms (FL)
29:61%
500 employees and more (525)

High Technology (HT) 9:48%


(168)
Middle High Technology (MH) 26:11%
(463)
Middle Low Technology (ML) 38:24%
(678)
Low Technology (LT) 26:17%
(464)

Number of rms 1773


Source : SESSI (1997)

23
7.2. First two dimensional factor space for delayed and given up projects to innovation

Fig. 7.1

24
Fig. 7.2

25
Rfrences

[1] AGHION, P. and J. TIROLE, (1995), The Management of Innovation, Quartely Journal of
Economics, Vol. CIX, No.4, November, pp. 1185-1209.
[2] ARORA, A. and A. GAMBARDELLA, (1990), Complementarity and External Linkages :
the Strategies of the Large Firms in Biotechnology, The Journal of Industrial Economics,
Vol. 38, pp. 361-379, June.
[3] ARORA, A., (1996), Testing for Complementarities in Reduced-form Regressions : A Note,
Economics Letters, 50, pp. 51-55.
[4] ATHEY, S. and S. STERN, (1998), An Empirical Framework for Testing Theories About
Complementarity in Organizational Design, NBER Working Paper No. W6600, June.
[5] BALDWIN, J. and J. JOHNSON, (1995), Dveloppement du Capital humain et Innovation :
la Formation dans les Petites et Moyennes Entreprises., Document de Recherche de Statistique
Canada, No. 74, March.
[6] BALDWIN, J. and M. Da PONT, (1996), Linnovation dans les Entreprises de Fabrication
Canadiennes, Document de Recherche de Statistique Canada, No. 88-513-XPB.
[7] BALDWIN, J. and V. PETERS, (2001), La Formation comme Stratgie en Matire de
Ressources Humaines : La Raction aux Pnuries de Personnel et au Changement Tech-
nologique, Document de Recherche de Statistique Canada, No. 154, April.
[8] BARLET, C., E. DUGUET, D. ENCAOUA and J. PRADEL, (1998), The Commercial Suc-
cess of Innovations : An Econometric Analysis at the Firm Level in French Manufacturing,
Annales dEconomie et de Statistiques, Spcial Issues 49/50, January/June, pp. 457-478.
[9] BECKER, B. and M. HUSELID, (1998), Human Resources Strategies, Complementarities,
and Firm Performance, mimeo, Jully.
[10] BELIN, J., (2001), Recherche-Dveloppement et Facteurs Financiers : Une Etude sur Donnes
dEntreprises Franaises, mimeo, Sminaire ERMES, January.
[11] CAROLI, E., N. GREENAN and D. GUELLEC, (2001), Organizational Change and Skill
Accumulation, Industrial and Corporate Change, Vol. 10, No. 2, pp. 481-506, June.
[12] CGP, (1999), Recherche et Innovation : la France dans la Comptition Mondiale, Rapport du
groupe de travail prsid par Bernard Majoie, Commissariat Gnral du Plan, La Documen-
tation Franaise, November.
[13] COHEN, W., (1995), Empirical Studies of Innovative Activities, dans Handbook of the Eco-
nomics of Innovation and Technological Change, Edited by STONEMAN, P., Blackwell Pub-
lishers Ltd., Oxford.
[14] COHEN, W. and R. LEVIN, (1989), Empirical Studies of Innovation and Market Structure,
dans Handbook of Industrial Organisation, Vol. II , pp. 1060-1107, Edit par SCHMALANSEE,
R. and R. WILLIG, North Holland, Elsevier Science Publishers BV, Amsterdam.
[15] COLOMBO, M. and R. MOSCONI, (1990), Complementarity and Cumulative Learning Ef-
fects in the Early Diusion of Multiples Technologies, The Journal of Industrial Economics,
Vol. 43, pp. 13-49, March.
[16] CRAMPES, C. and D. ENCAOUA, (2001), Microconomie de lInnovation, Cahiers de la
MSE, Srie Verte, No. 2001.67.
[17] CREPON, B., E. DUGUET et J. MAIRESSE, (2000), Mesurer le Rendement de lInnovation,
Economie et Statistique, No. 334, 2000-4, pp. 65-78.

26
[18] DUGUET, E., (2001), LEet des Aides Publiques sur le Financement Priv de la Recherche,
lInnovation et la Diusion des Connaissances : un Examen Microconomtrique Partir des
Mthodes dAppariement Slectif, Sminaire Fourgeaud, June.
[19] ELIASSON, G., (1990), The Firm as a Competent Team, Journal of Economic Behavior
and Organization, Vol. 13, No. 3, pp. 275-298, June.
[20] ELIASSON, G., (2000), The Role of Knowledge in Economic Growth, Report Department
of Industrial Economics and Management, Royal Institute of Technology, Stockholm.
[21] ENCAOUA, D., B. HALL, F. LAISNEY and J. MAIRESSE, (2000), The Economics and
Econometrics of Innovation, Kluwer Academic Publishers, Boston.
[22] FRANCOIS, J.-P., (1998), Les Comptences pour Innover dans lIndustrie, Chires Cls
Rfrence, SESSI.
[23] FRANCOIS, J.-P. and F. FAVRE, (1998a), LInnovation Technologique dans lIndustrie,
Chires Cls Rfrence, SESSI.
[24] FORAY, D. and J. MAIRESSE, (1999), Innovations et Performances, Approches Interdisci-
plinaires, Paris, Edition de lEHESS.
[25] GAILLARD, J.-M., (1997), Marketing et Gestion de la Recherche et Dveloppement, Econom-
ica, Paris.
[26] GALIA, F., (2001), Gestion des Equipes de R&D dans les Firmes Multinationales, dans
BALLOT G., G. BRESSON, F. FAKHFAKH and A. HAMMOUDI, (2001), Investissement en
Capital Humain et R&D, Formes de lInnovation et Exportations, Commissariat Gnral du
Plan.
[27] GALIA, F. and D. LEGROS, (2002), Les Obstacles lInnovation : Etude Empirique sur
Donnes Franaises, Document de Travail ERMES, No. 02-01, Janvier.
[28] GOURIEROUX, C., (1989), Economtrie des Variables Qualitatives, Economica, coll. ENSAE,
Paris.
[29] GASSMANN, O. and M. VON ZEDTWITZ, (1998), Organisation of Industrial R&D on a
Global Scape, R&D Management,Vol.28, No.3, pp. 147-161.
[30] GOURIEROUX, C., A. MONFORT, E. RENAULT and A. TROGNON, (1987), Generalised
Residuals, Journal of Econometrics, Vol. 14, No. 1/2, pp. 5-33.
[31] GREENAN, N., (1996), Innovation Technologique, Changements Organisationnels et Evolu-
tion des Comptences, Economie et Statistique, No. 298, pp. 15-33.
[32] GREENAN, N. and J. MAIRESSE, (1999), Organizational Change in French Manufacturing :
What Do We Learn From Firm Representatives and From Their Employees ?, NBER Working
Paper, No. W7285, August.
[33] GUILLAUME, H., (1998), Rapport de Mission sur la Technologie et lInnovation, March, 234
p., (www.nances.gouv.fr/innovation/guillaume).
[34] ICHNIOWSKI, C., K. SHAW et G. PRENNUSHI, (1997), The Eects of Human Resource
Management Practices on Productivity : A Study of Steel Finishing Lines, American Eco-
nomic Review, June, 87(3), pp. 291-313.
[35] JUNG, A., (2001), Are Product Innovation and Flexible Technology Complements ?, WZB
Discussions Papers, FS IV 01-07, Berlin.
[36] KENDALL, M. and A. STUART, (1973), The Advanced Theory of Statistics, Vol. 1, 2 et 3,
3rd Edition, Grin London Ed., Londres.

27
[37] KLEINKNECHT, A. and P. MOHNEN, (2001), Innovation and Firm Performance : Econo-
metric Explorations of Survey Data, Palgrave Publishers Ltd, November, 288 p.
[38] LAURSEN, K. and N. FOSS, (Forthcoming, 2002), New HRM Practices, Complementarities,
and the Impact on Innovation Performance, Cambridge Journal of Economics.
[39] LAURSEN, K. and V. MAHNKE, (2001), Knowledge Strategies, Firm Types, and Comple-
mentarity in Human-Resource Practices, Journal of Management and Governance, Vol. 5,
No.1, pp. 1-27.
[40] LEBART, L., A. MORINEAU et M. PIRON, (1997), Statistique Exploratoire Multidimension-
nelle, 2nd Edition, Dunod, Paris.
[41] LEIPONEN, A., (2000a), Collaboration, Innovation, and Firm Performance - Increasing Re-
turns from Knowledge Complementarities ?, Paper prepared for the Conference Supermodu-
larity and Innovation, UQAM, Montral, June.
[42] LEIPONEN, A., (2000b), Competencies, Innovation and Protability of Firms, Economics
of Innovation and New Technology, No. 9, pp. 1-24.
[43] LHOMME, Y., (2001), Le Financement de lInnovation Technologique dans lIndustrie, Le
4 Pages, SESSI, No.156, November.
[44] LHUILLERY, S., (2000), Impact and Complementarities of the Organizational Practices of
Innovation : an Empirical Investigation, Sminaire les Enjeux Economiques de lInnovation
- CNRS, Paris, November.
[45] LHUILLERY, S., (2001), Managing Surveys on Technological Knowledge : French Experi-
ence in the Nineties, Paper for the Conference on Knowledge Management in the Innovation
Process, Statistics canada, Ottawa, February 1-2.
[46] MILGROM, P. and J. ROBERTS, (1990), The Economics of Modern Manufacturing : Tech-
nology, Strategy, and Organization, American Economic Review, June 1990, 80(3), pp. 511-
528.
[47] MIRAVETE, E. and J. PERNIAS, (2000), Innovation Complementarity and Scale of Pro-
duction, mimeo, University of Pennsylvania, Last revised version, November.
[48] MOHNEN, P. and J. ROSA, (2000), Les Obstacles lInnovation dans les Industries de
Services au Canada, Cahiers du CIRANO, No. 2000s-14, April.
[49] MOHNEN, P. and L.-H. RLLER, (2001), Complementarities in Innovation Policy, Cahiers
du CIRANO, No. 2001s-28, April.
[50] MONJON, S. and P. WAELBROECK, (2000), The Eects of Internal and External Informa-
tion Sources on the Innovation Process : an Econometric Analysis at the Firm Level, Cahiers
de la MSE, Srie Verte, No. 2000.29.
[51] OCDE, (1994), La Mesure des Activites Scientiques et Techniques, Manuel de Frascati, Paris.
[52] TARONDEAU, J.-C., (1994), Recherche et Dveloppement, Vuibert, Paris.
[53] TIROLE, J., (1988), The Theory of Industrial Organization, MIT Press, Cambridge, Mas-
sachusetts.
[54] TOPKIS, D., (1998), Supermodularity and Complementarity, Princeton University Press.
[55] VINCK, D., (1991), Gestion de la Recherche : Nouveaux Problmes, Nouveaux Outils, sous la
coordination de VINCK D., Ed. De Boeck Universit, Bruxelles.
[56] VIVES, X., (1999), Oligopoly Theory, Cambridge University Press.

28

S-ar putea să vă placă și