An Evolutionary View of Innovation and Economic

Performance: A Firm Level Analysis

Haned, Naciba. European Conference on Innovation and Entrepreneurship: 273-XIV.
Reading: Academic Conferences International Limited. (Sep 2010)




Abstract. It is rather consensual that firm growth driven by financial long run profits seeking,
depending on market structures and how firms adapt to it thanks to their productive resources
(Penrose, 1959; Nelson and Winter, 1982; Teece, 2006). However, some differences emerge
when it comes to the origin of the productive resources that raise growth. Considering an
industrial market structure, evolutionary approaches describe innovations as a source in favour of
the emergence of "path dependencies" (because of lock-in effects and suboptimal behaviours on
the market and inside organisations) affecting innovativeness inside organisations. In fact, even
minor exogenous shocks may affect the occurrence of a particular path, changing the structure of
a given competitive model. This paper examines with two-stage least squares method economic
performance's impact on the persistent dynamics of innovation inside the French industry using
two waves of community surveys starting from 2002. It appears that the more firm growth raises
(at upper quantiles), the more innovativeness increases, and rises above the medium rate when
the productive opportunity of the firm is not limited within the studied period.
Keywords: innovation, capabilities, complementary assets, economic performance, two stage
least squares
(ProQuest: ... denotes formulae omitted.)
1. Introduction
There are no rules in terms of predicting if an innovation investment would get more benefit in a
firm. In fact, the central issue for empirical studies on the economic returns of innovation is to
identify and analyze the determinants of superior performance. The theoretical framework
supports this view, but empirical results on this dynamic are mixed since innovation is more
complex in actual facts.
Previous empirical findings shed light on first scientific or technological progress (empirically, it
can be considered regardless of the source of the invention), as the successful introduction on the
market of the new product, process or organizational innovation, that allows an enterprise to
ignore market and industry conditions to grow and gain market shares to the detriment of less
innovative enterprises. In fact, being given different levels of competitiveness of firms, building
a project of innovation without considering a loss of value once the product is put on the market
is not possible, being given the complexity of the innovation process. Supporting this
methodology, some empirical investigations controlling for other factors additionally to R&D
confirm that the R&D performance link is unclear and not straightforward (Schankerman, 1981;
Cuneo and Mairesse, 1984; 1985; 1988; Griliches and Mairesse, 1984; Hall and Mairesse, 1995).
Nevertheless, this conception of an innovation narrows the possibilities of accounting for
innovativeness and particularly its inputs.
The Schumpeterian vision of economics permeates the entire economics innovation framework,
in which innovation remains considered to be as a key success factor for economic performance
in the long term. In this framework, a successful innovation is seen as the positive outcome of a
dynamic process with many feedback circuits (Rosenberg, 1986). More precisely, technological
wherewithal improvements depend on a successful combination of firm specific capabilities
(Teece and al., 1997; Zollo et Winter, 2002) and industrial contexts (Metcalfe, 1995). The
baseline characteristics in this framework is opposed in many features to the analyses economic
agents' evolution with bounded rationality, as agents are not profit optimizers but profit seekers
(Simon, 1961).
Moreover, as R&D increases, companies do not always appropriate the results of R&D efforts in
order to increase corporate performance (Griliches, 1979) especially in highly competitive
markets. Most of the investigations measure only partially the effect of innovation on
productivity (Griliches, 1995), putting forward technological opportunities at the center of R&D
incentives (Dosi, 1988).
In this sense, the Penrosian and evolutionary theory of growth of the firm sheds light on the fact
that firms have the ability to allocate resources in productive resources, drawing this capacity
from its organization (Penrose,1959). As a consequence, intangibles are a very important feature
to develop such a capability. Hence, innovation inputs are the same that innovation outputs
meaning that new resources for the development of new process and product innovation earning
resources and all structures to access new knowledge and know-how.
The aim of this paper is to contribute in the analysis of the two-way relationship between
innovation and persistent corporate performance using a longitudinal data set which matches data
from innovation surveys and annual survey that provide data on companies' accounts from 2002
to 2006. The rest of the paper is organized as follows: the second section highlights a
background dressing the characteristics of the innovation process, and the dynamics of the
innovation output, and the second one gives a description of the data set. Finally the third and the
fourth section present the model and explore the results.
2. The persistence of innovative performance: A conceptual framework
2.1 The technological capacity of firms
The seminal work of Schumpeter emphasizes vital issues in the innovation process as its costly,
risky and uncertain nature, and the importance of appropriability of the economic benefits of
innovation in a large sense. In this framework the successful innovation is described as the
introduction of technological advance at the origin of profit generation rather than qualifying
innovative activities in the large sense as a performance (Schumpeter, 1934). However, there are
no conclusive analyses on the determinants of innovation for firms. Also, even in perfect
competitive conditions, firms are disadvantaged because of the constraint the nature of modern
industries creates. Thus, firms have to be endowed with increasingly sophisticated specific and
technological competences ones in order to meet the nature of modern industries (Schumpeter,
1950).
2.1.1 The uncertainty of innovation resources
"Innovative" companies that have engaged and R&D effort, or that allocate resources for
innovation support a much higher risk than "imitators", because their projects are not always
successful. In fact, it is difficult for an investor to select "good" R&D projects because R&D
investments lead to long term profitability, because R&D investments are risky and their results
is uncertain (Stiglitz and Weiss, 1981).
However, there are factors having an impact on the probability of success of a given innovation
project. We come back hence to the first Schumpeter hypothesis (1942) who stated that in a
world of increasing competition, innovation activities increase more than proportionally with
firm size. Therefore, we can identify the set of factors that are likely to influence the decision to
engage in innovation activities and those that impinge on the probability of success of an
innovation.
2.1.2 Internal resources for innovation
The existence of firm specific resources (Penrose, 1959; Nelson and Winter, 1982) and the use of
external resources (Pisano, 1990) have an impact on the technological capacity of the firm,
meaning the propensity to innovate. Firm specific assets (as R&D or business intelligence for
modest firms) are at the core of the development of such capacity (Klein, Crawford et Murphy,
1978; Williamson and Riordan, 1999). Actually, innovation follows the strengths of a firm;
meaning using its internal capabilities to survive "internal process of development leading to
cumulative movements" (Penrose, 1959, p.1). The capabilities are presumed to be key resources
personnel that act as a team outside of which effectiveness of decisions is reduced. These
practices rely on tacit forms of knowledge that can be easily transmitted because of their
experiential nature. The latter criterion allows the mobilization of fundamental competences for
the renewal of the base of knowledge notably the process of technological paradigms (Dosi,
1982). As each company uses knowledge as a free and non exclusive good, trajectories and also
conflicting aims in the paradigm itself, that creates a need for codification, in turn generating
new forms of processes, because the need of transfer is created by its own tacit form (Polanyi,
1962; Schutz, 1962). Consequently, changes in the knowledge and know-how bases imply
significant discontinuities in the ways that knowledge is generated and economically exploited.
In such circumstances, as the path of performances broadens, the heterogeneity of firm regarding
the diversity of innovation performances is no longer a temporary situation in the tradition of
Knight (1921). To narrow the analysis towards the frame of evolutionary dynamics, we consider
that innovations are no longer linear but sequential and developed through evolutionary learning
complex processes with many feedback circuits. In fact, specific decisions emerge from
routinized search methods -as rules of thumb pouce- are described in the framework of the
Darwinian metaphor (Metcalfe, 2008). These routines also known as "the best it knows and can
do", include not only operating rules as regard with production and supply factors for production
but also strategic and investment rules aiming at decision-making. They are considered as a
repository of knowledge and skills with an ability to replicate, even when no effort is provided
by the firm (Nelson and Winter, 1982; Nelson, 1995). Within this framework, innovations are
described as not linear but sequential and developed through evolutionary learning complex
processes with many feedback circuits: "Search and selection are simultaneous, interacting
aspects of the evolutionary process: the same prices that provide selection feedback also
influence the directions of search. Through the joint action of search and selection, the firms
evolve over time, with the condition of the industry in each period bearing the seeds of its
conditions in the following period" (Nelson and Winter, 1982).
2.1.3 External resources for innovation
In addition to their eventual R&D activities (or more modestly business intelligence), companies
use external knowledge by either purchasing licenses or attracting relevant skills for business
services as collaboration activities altogether dropping transaction costs.
These complementary assets are also important even for leaders (or first movers) since they
involve the use of external knowledge sources (Pisano, 1990). That is the case of licensing or he
decentralization of R&D (Cassiman and Veugelers, 2006; Arora et Gambardella, 1994); or
collaborating with other firms (D'Aspremont et Jacquemin, 1988) or research institutes (Cohen,
Nelson and Walsh; 2002).
That is the case of licensing or he decentralization of R&D (Cassiman and Veugelers, 2006;
Arora et Gambardella, 1994); or collaborating with other firms (D'Aspremont et Jacquemin,
1988) or research institutes (Cohen, Nelson and Walsh; 2002).
In this context, organizations with high innovation capabilities seize more important
technological opportunities. These same companies easily increase their innovation capacity
because it is incremental (Clark and Griliches, 1984). And large firms have a higher propensity
to innovate in opposition to small ones. In this sense, empirical studies show the importance of
market power and firm size for innovation (Cohen and Levin, 1989), but their results shed light
on conflicting and overall undermined evidences (Scherer, 1992; Cohen, 1995). In fact, these
studies are constrained by the use of cross-sectional data and consequently use rather input
oriented indicators of innovation as R&D, which account for the early stages of the innovation
profits that lead to rather uncertain flows of results.
The literature on the determinants of innovations also considers the dichotomy created by
industrial dynamics, as technology push and demand-pull. Technology pushed innovation
activities are impulse by the race to innovate based on the environment selection criteria
(Scherer, 1982; Coombs and al., 1987). Whereas demand pulled ones acknowledges are
stimulated by other factors on the demand side increasing the incentives of innovation
production and similarly R&D expenditures (Schmookler, 1962, 1966).
2.2 Innovation rewards
Incentives for innovation have traditionally been linked to competition, especially in the
Schumpeterian tradition. Firms compete in the market for products, in part, by becoming more
effective in their production processes (i.e., by innovating) and thereby gain market power. The
appropriability of new knowledge meets difficulties.
2.2.1 The economic profitability of innovations
Some empirical studies linking innovation to corporate performance concentrate either the
persistence of innovation "success breeds success" (Mansfield, 1968; Peters, 2005; Duguet et
Manjon, 2002 ; Raymond et al., 2006); or emphasise the importance R&D for productivity
(Crpon et al., 1998), or financial results (Roberts, 2001).
The other set of studies explore the allocation of resources (Mairesse and Griliches, 1983;
Griliches, 1986), entry and exit (Geroski, 1995, Caves 1998). Whereas others empirical studies
analyze the underlying factors of productivity growth - as technology (Mairesse and Sassenou,
1981; Hall and Mairesse, 1995) or human capital (Crpon, Duguet, Mairesse, 1998).
According to this argument, it is the final stage of innovation that must be explored, especially
when it allows at the same time to measure the impact with the flows of economic performance
that are generated by a successful innovation.
For example, results from innovation French survey data find a positive link between innovation
output and R&D spending, whereas firm size has no impact (Crpon and al., 1998). Also, a
number of empirical studies acknowledged for the impact other firm specific indicators as
financial capability (essentially cash flow) which results shed light on a rather weak (Mairesse
and al., 1999) or no effect at all (Heshmati and Lf, 2006) of financial performance on the
propensity to innovate.
Since the product is successfully brought onto the market, and imitators begin to enter the
market, innovators are no leaders (Teece, 1987). The economic value of assets depends on the
crucial issue of "appropriability" and opportunities on the market.
Because intellectual property deposits raise the economic value of assets and similarly
innovation rents. Then, many empirical studies, in particular the "Yale survey" of Levin et al.
(1987) and, more recently, Cohen and Al (2000) suggest that the first mean of protecting
innovations is not patenting but rather other ways such as the secrecy or first mover advantage
according as the industrial sector.
2.2.2 Satisficing and the generation of knowledge
The coordination of the actors and factors on the market is not always optimal for innovation
outputs (Arrow, 1962), more especially as the output of innovation is uncertain.
In fact, as the coordination of actors and factors on the market is not always optimal (Arrow,
1962), resulting in uncertain flows of outcomes especially being given the uncertainty of
innovation output whatever the amount of the initial investment is. Since novelty is not
predictable (cannot affect a probability to each eventuality), it is not possible to insure its outputs
(no risk mutualisation); which is possible for risky investments through insurance premiums
(Knight, 1921). In this sense the Schumpeterian vision puts forward a dynamic rather than static
efficiency, where the reward from a successful innovation a superior value for a given
expenditure, and confers to the firm market power and growth, at the expense of non innovating
firms.
Contemporary theorising abandons the idea of the Marshalian quasi-rent in favour of competitive
strength and viability of a company that depends on "core competences" and its relations with its
customers and suppliers. These assets are represented as "a set of technological skills
differentiated, complementary assets and routines that are the basis of competitive capacity of an
undertaking in a particular activity" (Dosi, Teece et Winter, 1990).
As individuals (An individual can be an organization, a legal entity or even an institution and
makes research according to its own features (Penrose, 1959; Hodgson, 1993) are "profit
seekers" instead of "profit optimisers", the first stage of this mechanism is characterised by the
introduction of a successful innovation on the market, in turn giving birth to new markets and/ or
new companies. At this stage innovators are leaders that benefit from monopolistic rents,
yielding supernormal profits. That how creative destruction proceeds, and may yield to a
transitory but not necessarily persistent profits as imitators enter the market.
This measure problem is acknowledged in empirical studies that explore the impact of a large
range of firm-specific and industry specific factors on innovation output.
An innovation is not necessarily introduced on the market even if R&D is made, if it is the case,
the impact on economic performance is weak, but the impact on corporate performance is high,
by the introduction of new knowledge.
3. Model and empirical specifications
Before describing the econometric model, we examine the data set and the principal indicators
explored in the empirical analysis. The aim of this section is to emphasize the proxies
constructed on the basis of not only CIS data but also other data basis provided by INSEE (The
French public Institute of Statistics), and to describe the methodological problems we had to
cope with.
3.1 Data set and variables
Our empirical study is based on a longitudinal data set obtained by merging two waves of
Community Innovation Surveys (hereafter CIS) : CIS4 (2002-2004) and CIS2006 (2004-2006).
The specific architecture of innovation surveys requires some clarifications, as for it contains
different types of variables on innovative and non-innovative industrial and service firms.
These surveys gather the same information on a three time span period, and are conducted at
regular intervals: CIS4 (2002-2004) was conducted in 2005 and CIS2006 (2004-2006) in 2007.
The first part of surveys provides general information on companies, as their main activity, size
and turnover; if the company is part of a group or not, and also the localisation of its most
significant market.
Then, the second parts of surveys are related to the introduction of product and process
innovations and R&D expenditures (Internal and external sources, equipments and materials).
The first period (2002-2004) is covered by CIS4, which results are decomposed in basic
information (as activity, turnover, employees, geographical situation, etc) and various aspects of
innovation activity (products and process innovation, R&D expenditures, intramural and
extramural R&D, knowledge acquisition, factors hampering innovation, etc).
Most of the data is qualitative, which means that to obtain results on innovation performance
over time, we have to match CIS data with other data, by using accounting data.
In this sense, to obtain financial information on our sample data, we have to use annual
enterprise surveys (2002-2006) conducted by Sessi, the Ministry of Agriculture (for IAA) and
INSEE in industry for gathering information regarding the main firm economic indicators
(staffing, value added, investment, profitability, and so on).
The resulting sample from merging all the data sets result in 2354 companies in the
manufacturing tool industry, expressed by the NACE two digit level in the European
classification of activities (Table 1), with 20 or more employees.
We can consider that there are two main segments: the first one includes low-tech markets with,
say a weak percentage of innovation product put on the market: textile (2,9 %,), wood paper and
printing (3,7 %), plastic and rubber (2,8 %), glass and ceramic (2,8 %), Metals (4,4 %), furniture
and recycling (1,8 %). Whereas the second one includes high-tech markets as chemicals ();
electrical and electronic engineering (8,2 %) and vehicles (5,9 %).
3.2 The econometric analysis
The main constraint of our empirical work arises from innovation the complexity of the process
arising from an idea and evolves to output. We acknowledge for this complexity using not a
unique relation between the innovation effort and economic performance. In order to fulfil this
goal we introduce several factors that interact to identify the origin of innovations in a first step,
and then we capture the interrelationship between various indicators.
3.2.1 The dependent variable: the efficiency of innovation
We model the efficiency of innovation to assess the feedback loop between innovations and
economic performance, this model explains firm's sales growth between 2004 and 2006 with
several independent variables proper to institutional and organisational features companies have
adjust to. We will use a multivariate regression with one time lag starting from 2004 up to 2006
(see figure 1).
Our model aims at testing the hypotheses of the two stage mechanism previously described: the
first one aims at determining the propensity for a firm to innovate for our sample (n=2354).
... (1)
Where Y^sub i^ is the estimated performance measure in our study (growth of sales in the
turnover compared to industry between 2004 and 2006).
X^sub i^ is the percentage of innovative products in the sales turnover in logarithms for i=1,..., n.
It expresses the technological capability of the firm. If the percentage of sales in the turnover of
innovative products is positive for a given firm, i, it means that it has a high technological
capability.
This technological capability can be enhanced by several factors that we explained in the
analytical framework; including innovation inputs as size (SIZE). We also include internal
capabilities for innovations as training (TRAIN); internal and external R&D (RDINT, RDEXT),
and also take into account cooperation (COOP).
The second stage aims at the intensity of innovation outputs depending on firm resources, and
industry related features. Finally we measure to what extent our dynamic loop works in actual
facts.
These links materialize into a simultaneous system of equations, to connect innovation output to
innovation input, and in the same time, accounts for the selectivity biases (Johansson and Lf,
2008, 2009).
The method of instrumental variables operates into two steps. First, the innovation variable is
substituted X is substituted because it is endogenous. We expect to get little correlated results by
choosing instrumental variables for X estimation.
As X is correlated with the error term in equation 1, consequently, the second part of the model
is formulated as an instrumental variables equation, partly uncorrelated to ui in order to estimate
innovation output and the model comes as follows:
... (2)
Xi is estimated by the following explanatory variables:
Perf^sub i^ is past performance measured as sales growth rate compared between the years 2002
and 2004. Most of empirical studies have found a positive correlation between financial
capability (generally expressed by cash flow, see Mueller, 1967) and a strong R&D intensity
(Cohen, 1995).
Size in 2004 because the incentive to innovate should increase when the company grows
according to the Schumpeterian view of innovation dynamics.
Financial resources for innovation in 2004: internal R&D expenditures (RDINX) and external
ones (RDEX) even if most of firms finance innovation by internal funds because of hazard
situations (Stiglitz and Weiss, 1981).
We also control for other variables related to knowledge and scientific resources used between
2002 and 2004 because there is a lag period for which investments should be effective.
Therefore, we include firm specific resources that is to say internal (product developed by the
enterprise: PDINT), and external knowledge (PDEXT).
3.2.2 Estimating economic performance with two stages least squares
The third stage of our model is the estimation of innovation output in 2004 (equation 1) using
instruments related to innovation inputs, and then we include it in equation (2); we obtain
equation (3):
... (3)
As for equation 3; it is estimated by the innovation output ... and by a set of exogenous variables
(uncorrelated with the error term) explaining growth of turnover compared to industrial activity:
Firm size in 2006 is measured by the total number of employees in logarithms (SIZE06);
Two sets of dummy variables indicating the technological or innovative capabilities (Dosi, 1988)
used between 2004 and 2006: internal ones (Train0406 which stands for training; RDINT0406
which stands for internal R&D); and external ones (RDext0406 which stands for external
knowledge appropriated by the firm, and coop0406 which stands for cooperation);
Several dummy variables for industry (DUMIND), and innovation protections.
In order to measure the effect the protection of innovation rents, we use several measures,
because lagged effects of intellectual property rights assets are captured effects while taking into
account patent, trademarks, and designs deposits for two periods (2002-2006); but secrecy is also
important: in the first estimation ("model 1" in table 2), we include only two measure globally
including patent, designs, trademarks, and copyright deposits (Leg0206). But the lack of data
induced a binary proceeding for these variables whilst in the literature they are usually estimated
with count data models. Indeed, there is a biases in data because there is a strong value
innovations products and much of low value innovations because few activities are really
protected from the imitation and much would remain unprotected from appropriation
mechanisms (Duguet and Lelarge 2004).This must be taken into account in analyses of the
influence of patents on downstream changes in economic performance parameters (Griliches
1979; Jaffe 1986); consequently we also include secrecy, complexity of design and lead-time
(Strat0206). We also account separately for these results by introducing a variable for each type
of protection (pat stands for patents, desg stands for designs, tdm stands for designs, cp for
copyrights, sct for secrecy, tech for lead-time); we call it "model 2" (see table 2).
4. Results and discussion
4.1 First results
Our intuition about the positive relationship between economic performance and technological
capability over time is confirmed. We obtain approximately the same results for model 1 and
model 2, but surprisingly, intellectual property rights do not have the expected effects. In fact,
only patents are significant in the model two, with a positive coefficient of 0.53; while the legal
and strategic protections of innovations- considered as a whole- are not significant. Moreover,
corporate performance strongly varies according to industrial sectors, with strongest rates are
recorded for some high-tech industries.
In accordance with the Schumpeterian point of view of innovation, size has a positive impact on
economic performance via R&D intensity in our case. Innovation output and internal resources
for innovation are always significant with higher rates in the case of internal R&D and training.
That supposes correlations between internal factors enhancing innovation effort and its output. In
fact, we can observe in this sense that the estimated output of innovation is significant (-0,92) but
has that the coefficient is less important that the one for internal R&D resources (-0,82). The
capacities of the firm to adapt internal and external technologies play an important part in the
incentives to implement innovative projects. As for external resources do not have any effect on
economic performance, and the same results are observable for active cooperation. Nevertheless,
innovation protections have a positive effect on the growth rate of the turnover, which is an
interesting result, since patents encapsulate the information on the innovation process. In fact,
having a patent can be interpreted in two complementary ways: first, it means that the firm has
succeeded in the process of introducing a product on the market. But it also can be interpreted as
the innovation output.
4.2 Discussion
We had to cope with some methodological difficulties. In particular, a lot of firms prefer secret,
early movers' advantages or other means identified by the Yale survey for protecting the rents
from innovation. These firms may be as effective as others. We cope with this difficulty by an
appropriate choice of the sectors forming the sample on the one hand, and the use of firm
answers from CIS4 and CIS5 on the other. Secondly, there is a likely loop, a circular relationship
("a cumulative causation"), between fir complementary assets and firm growth. In fact the
reverse causation from economic growth to investments in IP assets is equally relevant: only the
firms with a high growth have a threshold level of profitability for funding an IP department, a
persistent flow of investment in R&D, and so on. It will be important to keep in mind that when
we will run estimations. One mean for bypassing this difficulty might be to put temporal lags
between the variables and estimate quantiles in order to capture the correct effects we want. The
last limitations would be the fact we could not get long times series. We will carry out only cross
section estimations. For this reason we have to define the specifications of our main variables
carefully in order to isolate the effects due to sectoral environments on firm growth (for instance
running regressions on the differences with the sectoral mean). It will be possible to control for
these issues by using a random effects model with a seemingly unrelated regression to capture
the effects of time lags.
It seems that we have to take into account the dichotomy between product and process
innovations to limit the selection biases that can be involved in the two steps process of
innovation output on the performance of R&D in the same tradition of Mansfield work (Mairesse
and Mohnen, 2004), particularly by reassessing the importance of R&D for innovation statistics.
References
References
Arora, A. and A. Gambardella (1994), "Evaluating technological information and utilizing it,"
Journal of Economic Behavior and Organization, 24, pp. 91-114.
Cohen WM, Klepper S. (1996) "Firm size and the nature of innovation within industries: the case
of product and process R&D", Review of economics and statistics, Vol 78, No. 2, pp 232-43.
Cohen, W.M., Nelson, R.R., Walsh, J.P. (2000) "Protecting Their Intellectual Assets:
Appropriability Conditions and Why U.S. Manufacturing Firms Patent (or Not)", NBER
Working Paper No. 7552.
Crpon B., Duguet, E., Mairesse J. (1998) "Research, innovation and productivity: an
econometric analysis at the firm level", Economics of Innovation and New Technology, Vol 7,
No. 2, pp.115-158.
Cassiman, B. and R. Veugelers (2006), "In Search of Complementarity in Innovation Strategy:
Internal R&D, Cooperation in R&D and External Technology Acquisition," Management
Science, 52, 1, pp. 68-82.
Metcalfe, J.S (1995) "Technology Systems and Technology Policy in an Evolutionary
Framework," Cambridge Journal of Economics, Vol. 19, No. 1, pp. 25-46.
Nelson, R.R, Winter, S.G. (1982) An Evolutionary Theory of Economic Change, Cambrige,
Harvard University Press.
Penrose, ED. (1959) The Theory of the Growth of the Firm, Oxford University Press: New York.
Pisano G., "The R&D Boundaries of the Firm: An Empirical Analysis", Administrative Science
Quarterly, n 35, 1990, p. 153-176.
Teece D, Pisano G,Shuen A. (1997), "Dynamic capabilities and strategic management", Strategic
management journal, Vol 18, No. 7.
Teece, David, J. (2006) "Reflections on profiting from innovation", Research Policy, Vol. 35, pp
1131-1146.
Zollo, M. and S. G. Winter (2002) "Deliberate learning and the evolution of dynamic
capabilities", Organization Science, Vol. 13, No. 3, pp.339-353.
AuthorAffiliation
Naciba Haned