Public Research Organisations (PRO) – Industry interactions in developing countries: Insights...
-
Upload
fabricio-martins -
Category
Documents
-
view
637 -
download
0
description
Transcript of Public Research Organisations (PRO) – Industry interactions in developing countries: Insights...
Public Research Organisations (PRO) – Industry interactions in developing
countries:
Valeria Arza & Andrés LópezFundación CENIT
Insights from the literature and empirical evidence of Argentina
OPEN SESSION
PRO-I interactions in ArgentinaOrganisation
I. Historical roots of PRO-I linkages in the world
II. Conceptual framework to analyse PRO-I linkages in developing countries
III. Case study on an interaction in biotechnology
IV. Research design for the analysis of firms’ survey
V. Determinants of linking and linking actively
VI. Average effect of linking
VII. Average effect of linking actively
VIII. Conclusions
I. Historical developments of PRO-I interactions
Linear model of innovation (separation of S&T) became questioned Changes in the production of scientific
knowledge, especially in some fields of research Geopolitical, economic and ideological changes
about the role of the StateIn developing countries: liberal ideology
Market should guide any inquireBudget constraints to public researchPRO took the place of the state in local/regional development
II. Conceptual framework of PRO-I linkages
Mode 2, Triple Helix, Sabato’s triangle, NIS literature, etc. -> provided a general context to understand PRO-I linkages - no methodologies or tools to derive hypotheses
No framework on PRO-I in developing countries, which differ from developed countries in: policy strategies and needs, historical evolution of S&T institutions structure and behaviour of the private sector
II. PRO-I linkagesRisks & Benefits Benefits
Contribution to the stock of public K (science learn from applications)
Funding for public research Solution of concrete problems Increase the potential for frontier-moving innovations Spillovers – local development
Risks Adequacy of PRO’s research and timing Privatisation of research outputs Interference in lines of research Opportunity costs (of time forgone in interactions) Conflict of interests between primary functions of public research
(teaching and research) and secondary functions (interactions)
II. Motivations and modes of interactions
Motivations Universities
Intellectual imperatives Financial imperatives
Firms Long-term innovative strategies fed by external sources of
knowledge Short-term problem solving strategies
Modes of interaction Bi-directional: shared intellectual resources and
outputs (e.g. joint research) Unidirectional: unilateral provision of intellectual
resources in exchange for money (e.g. consultancy, training, testing, etc.)
II. Motivations & Modes =>Risks & Benefits Main claims
When PRO-I are motivated by intellectual imperatives and long-term strategies => benefits>risks Usually instrumented by bidirectional modes Sophisticate knowledge bases in both ends
When PRO-I are motivated by financial imperatives and short-term strategies => benefits<risks Usually instrumented by unidirectional modes Firms’ supplement their knowledge bases with K from PRO
In LA first types of linkages are scarce Production structure do not draw from fields of research of the
learn heavily from application Use of mature technology Less sophisticate demand from industry Uncertainty + inconsistent S&T policy => short-term demand from I
II. Risks & Benefits Main goals
Given the likelihood of risks in PRO-I in developing countries (LA), we must investigate further which behaviours, practices, and characteristics of PRO & I that prioritises benefits over risks
This projects aims at exploring benefits and risks and both ends by analysing Survey on firms Survey on research In-depth case studies
III Case study on PRO-I interaction in biotechnology Actors: Biosidus, leader firm in biotechnology, and Institute of
Antarctic Research (IIA) Goal: to discover native bacteria which have not been described
before => mission oriented research on the IIA traditional lines of research
Potential application: enzymes from bacteria adapted to extremely low temperatures may substitute enzymes that need high temperatures to be effective very much used in the food and textile industry => potential reduction of the use of energy. 15 million USD market only in Argentina => seeking radical innovations
Outcome: new bacteria was discovered, described and published. The bacterium’s genome was fully described for the first time in Argentina. Several further interactions followed-up with other public and private institutes.
III Case study on PRO-I interaction in biotechnology Modes of interaction: during the first four years => general
agreement; when concrete results were foreseen => specific agreements, stating rights and obligations Rights on patents and commercialisation for the firm, which
would pay royalties to the institute Further publications / cooperation need the consensus of both
parts Conclusion: Benefits were clear: the success of the research is explained by:
Curiosity-led / frontier moving motivations (long-term) Instrumented by bi-directional cooperation Leader actors in their fields / strong knowledge bases
Yet the case raises concerns about issues of privatisation of public knowledge and potential conflict of interests regarding publications => may they be overcome by S&T regulation?
IV Research design on firms’ surveysGoals of research
1. To compare the characteristics, behaviour and performance of firms that connect and firms that do not connect to PRO.
2. To compare the characteristics, behaviour and performance of firms that connect to PRO with bi-directional linkages and those that connect with unidirectional ones.
IV Research design on firms’ surveysData
Especial section added to the National Innovation Survey (fieldwork 2007) Firms that connected to PRO in 2005: 590 answered by
354 Firms that did not connect to PRO in 2005 (sub-sample
that resembles sector & size of linked group): 384 answered by 238
Data refers to 2005 Broad definition of interactions with PRO, including
active cooperation and information exchange
IV Research design on firms’ surveysMethodology
Propensity score matching techniques Construct valid counterfactual groups against which to
compare outcomes of the treated (e.g linked) group It finds a twin for every linked firm in the unlinked group.
Three steps Probit to calculate the propensity score (i.e. the probability
of linking to PRO) Matching methods on the propensity scores (i.e. matching
linked with unlinked firms with similar probabilities of linking) Four methods used: Nearest neighbour, Caliper, Kernel, Radius
Calculation of the average treatment effect (i.e. difference in an outcome variable -e.g. innovative expenditures- between linked firms and their twins) for different matching methods
V Linking actively We identify four modes of interactions that involve
active participation of firms (bi-directional linkages)
Average importance 0-1 scale
% of linked firms with importance
>0.5
Informal Exchange 0.58 51%
Publications 0.56 47%
Conferences 0.54 46%
Hiring graduates 0.44 27%
Consultancies 0.44 26%
Research contracts 0.42 26%
Joint R&D 0.42 25%
Licences 0.38 16%
Networks 0.37 15%
Patentes 0.37 15%
Scientific parks 0.35 12%
Internships 0.34 10%
Incubators 0.30 5%
University owned firms 0.27 3%
Spin off 0.27 2%
Any active mode 0.36 37%
37% of linked firms considered at least one of these modes as moderately or highly important
V Determinants of linking actively
marginal
effects 0.355*
Skills Professionals over total employment [0.192]
0.022** Size
Deciles based on employment for the full sample [0.010]
0.819 IA_sector
Innovative activities over sales for the sector full sample [2.323]
0.006** Linked_act_sector
Quantity of actively linked firms in the sector [0.002]
0.085 Link_vert Vertical linkage
[0.060] 0.018
Link_int Linkage within the firm network [0.058]
Observations 354 Pseudo R-squared 0.04 Wald 18.51***
Robust standard errors in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Source: The Survey 2005 & The Survey 2006
Firms that link actively have more sophisticate knowledge bases
VI Average effect of linkingOutcome variables
Variable group Variable Name Type of dataIA_sales Ratioimaq_sales Ratioinhouse_sales Ratioinn_prod Dummyinn_proc Dummy
c) Appropriability behaviour
patent1-4 scale normalised to 0-1
fin_int Percentage
fin_pro Percentage
import_pro_pub import_pro_protimport_pro_techimport_pro_lab
1-4 scale normalised to 0-1
role_pro_edurole_pro_resrole_pro_socrole_pro_entr
1-4 scale normalised to 0-1
a) Innovative behaviour
b) Innovative outcome
d) Sources of financing innovation
e) Perception of PRO (outcomes and roles)
VII Average effect of linking activelyOutcome variables
The same outcome variables plus:
Variable group Variable Name Type of data
f) Goals of the interaction
goal_absgoal_contrgoal_supplgoal_hrgoal_capgoal_k
1-4 scale normalised to 0-1
g) Firms' payment for linking
pay_abspay_contrpay_supplpay_hrpay_cappay_k
Dummies
VII Average effect of linking activelya) Innovative behaviour
Actively linked firms (ALF) invest 1.4% of sales in in-house innovative activities, which is significantly more than Non-actively linked firms (NALF)
They invest as much in in-house as in machinery
ATTNearest Neighbour 0.0258 0.0211 0.0316 0.0105
Kernel (normal) 0.0258 0.0237 0.0316 0.0079Radius 0.0258 0.0223 0.0321 0.0098Caliper 0.0258 0.0212 0.0321 0.0109
Nearest Neighbour 0.0161 0.0129 0.0133 0.0004Kernel (normal) 0.0161 0.0149 0.0133 -0.0016
Radius 0.0161 0.0136 0.0136 0.00005Caliper 0.0161 0.0133 0.0136 0.0004
Nearest Neighbour 0.0060 0.0064 0.0139 0.0075*Kernel (normal) 0.0060 0.0056 0.0139 0.0083**
Radius 0.0060 0.0057 0.0143 0.0086**Caliper 0.0060 0.0062 0.0143 0.0081*
AI_vtas
inhouse_vtas
imaq_vtas
a) Innovative behaviour
Variables Weight methods
All non-active firms before
matching (NALF)
NALF after matching (control
group)
Actively linked firms (ALF)
Difference of means of ALF - NALF firms
Means
VII Average effect of linking activelyb) Innovative outcome
No difference was found.
Data about linkages and innovation refer both to 2005. Time lags
ATTNearest Neighbour 0.5000 0.5615 0.6077 0.0462
Kernel (normal) 0.5000 0.5367 0.6077 0.0710Radius 0.5000 0.5320 0.6032 0.0712Caliper 0.5000 0.5635 0.6032 0.0397
Nearest Neighbour 0.5357 0.5308 0.5923 0.0615Kernel (normal) 0.5357 0.5642 0.5923 0.0281
Radius 0.5357 0.5549 0.5794 0.0245Caliper 0.5357 0.5317 0.5794 0.0476
inn_prod
inn_proc
Actively linked firms (ALF)
Difference of means of ALF - NALF firms
Means
Variables Weight methods
All non-active firms before
matching (NALF)
NALF after matching (control
group)
b) Innovative outcome
VII Average effect of linking activelyc) Appropriability behaviour
ALF are more prone to patenting. 8% of ALF obtained at least one patent, while only around 1.5%-3.5% of twins obtain patents.
ALF are in a privileged position to access public K. Concerns about privatisation of K (Nelson 2004)
ATTNearest Neighbour 0.0357 0.0154 0.0769 0.0615**
Kernel (normal) 0.0357 0.0355 0.0769 0.0414Radius 0.0357 0.0295 0.0794 0.0498*Caliper 0.0357 0.0159 0.0794 0.0635*
c) Appropriability behaviour
Variable Weight methods
All non-active firms before
matching (NALF)
NALF after matching (control
group)
Actively linked firms (ALF)
Difference of means of ALF - NALF firms
Means
patent
VII Average effect of linking activelye) Goals of the interaction
ALF consider all goals more important than NALF
The difference is the most important for goals_k and goals_contr
The difference is the least important for goals_cap and goals_hr
This confirm the active nature of ALF: they value K of PRO and they link to contribute rather than to supplement
ATTNearest Neighbour 0.3549 0.3585 0.4942 0.1357***
Kernel (normal) 0.3549 0.3577 0.4942 0.1365***Radius 0.3549 0.3567 0.5000 0.1433***Caliper 0.3549 0.3620 0.5000 0.138***
Nearest Neighbour 0.3359 0.3547 0.5543 0.1996***Kernel (normal) 0.3359 0.3448 0.5543 0.2095***
Radius 0.3359 0.3628 0.5500 0.1872***Caliper 0.3359 0.3540 0.5500 0.196***
Nearest Neighbour 0.3158 0.3256 0.4787 0.1531***Kernel (normal) 0.3158 0.3200 0.4787 0.1586***
Radius 0.3158 0.3244 0.4760 0.1516***Caliper 0.3158 0.3280 0.4760 0.148***
Nearest Neighbour 0.4012 0.4554 0.5669 0.1114***Kernel (normal) 0.4012 0.4147 0.5669 0.1522***
Radius 0.4012 0.4200 0.5660 0.1460***Caliper 0.4012 0.4550 0.5660 0.111***
Nearest Neighbour 0.4397 0.4638 0.5963 0.1324***Kernel (normal) 0.4397 0.4449 0.5963 0.1514***
Radius 0.4397 0.4580 0.5907 0.1327***Caliper 0.4397 0.4673 0.5907 0.1233***
Nearest Neighbour 0.3304 0.3430 0.5349 0.1919***Kernel (normal) 0.3304 0.3345 0.5349 0.2004***
Radius 0.3304 0.3361 0.5370 0.2009***Caliper 0.3304 0.3450 0.5370 0.192***
goals_cap
goals_k
goals_abs
goals_contr
goals_suppl
goals_hr
e) Goals of interaction
Variables Weight methods
All non-active firms before
matching (NALF)
NALF after matching (control
group)
Actively linked firms (ALF)
Difference of means of ALF - NALF firms
Means
VII Average effect of linking activelyf) Payment (only important goals)
ALF more often pay for they goals they seek than NALF
This may imply that ALF collaborate more formally with NALF
Exception: using PRO’s infrastructure (service provision), around 60% of both ALF and NALF pay (no sig. diff)
ATTNearest Neighbour 0.1186 0.0986 0.2113 0,1127*
Kernel (normal) 0.1186 0.1044 0.2113 0,1069*Radius 0.1186 0.0790 0.1935 0,1145*Caliper 0.1186 0.0806 0.1935 0.1129
Nearest Neighbour 0.1111 0.1084 0.4096 0,3012***Kernel (normal) 0.1111 0.1196 0.4096 0,2901***
Radius 0.1111 0.1451 0.3836 0,2385***Caliper 0.1111 0.0959 0.3836 0,2877***
Nearest Neighbour 0.1628 0.1364 0.3485 0,2121**Kernel (normal) 0.1628 0.1771 0.3485 0,1714**
Radius 0.1628 0.1926 0.3214 0.1289Caliper 0.1628 0.1607 0.3214 0.1607
Nearest Neighbour 0.2566 0.2427 0.4563 0,2136***Kernel (normal) 0.2566 0.2593 0.4563 0,1970***
Radius 0.2566 0.2506 0.4444 0,1938**Caliper 0.2566 0.2424 0.4444 0,2020***
Nearest Neighbour 0.5664 0.5364 0.6364 0.1Kernel (normal) 0.5664 0.5750 0.6364 0.0614
Radius 0.5664 0.5736 0.6214 0.0478Caliper 0.5664 0.5243 0.6214 0.0971
Nearest Neighbour 0.1250 0.1915 0.3723 0,1809*Kernel (normal) 0.1250 0.1292 0.3723 0,2431***
Radius 0.1250 0.1068 0.3614 0,2546***Caliper 0.1250 0.1084 0.3614 0,2530***
pay_cap
pay_k
pay_abs
pay_contr
pay_suppl
pay_hr
f) Firm's payment for linking
Variables Weight methods
All non-active firms before
matching (NALF)
NALF after matching (control
group)
Actively linked firms (ALF)
Difference of means of ALF - NALF firms
Means
VIII) Conclusions2nd goal: to compare the characteristics of bi-directional linkages (actively linked firms - ALF) and unidirectional linkages (NALF)
i. ALF higher intensity of investments in in-house innovative activities than NALF. ALF invest more or less the same proportion in incorporated and in-house technologies.
ii. ALF are particularly prone to protecting their intellectual property. Issues of concern about privatisation of public knowledge
iii. ALF value more largely all goals of cooperation. iv. Low incidence of formal payment for all interactions. ALF are
more likely to pay when interacting with PRO: more formal interactions? Especially, for HR at PRO.
v. There are better opportunities for PRO when interacting with ALF than with NALF because more knowledge value is involved
ALF invest more heavily in in-house innovative activities ALF value particularly knowledge resources exclusively available
in PRO rather than other inputs that could be found in other service providers.
ALF interact with PRO to contribute rather than to supplement innovative activities
VIII) ConclusionsMain findings for policy making Bi-directional interactions are more likely to optimise
risks and benefits of PRO-I linkages There is more knowledge value involved in interactions
with ALF than with NALF There is more learning potential for PRO. They invest their
time more productively: they could reach better research outputs and find sources of inspiration for new lines of research
Incentives should be created to promote bi-directional linkages while restricting unidirectional ones.
Linked firms, and ALF in particular, seem to be more prone to using IPR
Patenting by PRO is cumbersome Issues of concern about the privatisation on publicly
created knowledge. S&T policy must avoid tragedy of the scientific commons (Nelson 2004)