Analysis of scientific research groups with greater productive applicability in Brazil: capacities and interactions with firms1

Thiago Caliari

Instituto Tecnológico de Aeronáutica, ITA, São José dos Campos, Brasil

[email protected]

Tulio Chiarini

Instituto Nacional de Tecnología, INT, Río de Janeiro, Brasil

Consiglio Nazionale delle Ricerche, CNR, Roma, Italia

[email protected]

Abstract. This paper analyzes the capacities of BR Survey Research Groups in generating science and technology, and how these groups may be associated with university-firm interactions. To this end, we perform a classification based on the diagram proposed by Stokes (2005), using the qualitative comparative analysis (QCA) statistical technique. As a primary result, we find that scientific and technological capacity on the part of the research groups is not a necessary condition for interactions with companies. Indeed, some research groups without high scientific and technological capacity make more interactions (in terms of the total amount) and yield similar values in the analysis (relative to the number of researchers) in comparison with the most qualified groups: those located in Pasteur’s quadrant.

Keywords: Academic-industrial collaboration; science and technology; scientific research; research workers; technological innovations; Brazil.

Acronyms

CAPES Council for the Improvement of Higher Level Personnel (Coordenação de Aperfeiçoamento de Pessoal de Nível de Superior)

CNPQ (National Council for Scientific and Technological Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico)

ISI Institute for Scientific Information

NIS National innovation system

QCA Qualitative comparative analysis

RGD Research group directory (diretório de grupos de pesquisa)

R+D Research and development

ST&I Science, technology, and innovation

SciELO Scientific Electronic Library Online

U-F university-firm

1.Introduction

Competition compels capitalist firms to grow. Two key aspects in achieving growth are the construction of aptitudes and the ability to learn. Accordingly, the Schumpeterian perspective stresses this construction as a way of creating competitive asymmetries based on product differentiation and the establishment of more efficient processes.

By broadening the availability of knowledge and capacities, firms increase their opportunities for innovation; this is possible, for example, through efforts to increase scientific and technological understanding. As Rosenberg (1990) notes, firms strive via their own resources to carry out basic as well as applied research activities; and, striving for growth, they form strategic alliances by expanding their learning capacity.

Universities are essential partners of firms because they are key institutions in the innovation system, given their creation and dissemination of new knowledge and inventions by way of basic and applied research, development, and engineering (Mowery & Sampat, 2005; Mazzoleni, 2005; Mazzoleni & Nelson, 2005). The generation and dissemination of science and technology has received special attention in the international and national literature, both in terms of creating new spin-off businesses and increasing the capabilities of already established firms.

In this regard, academic capabilities merit the attention of public policy makers, given that scientific and technological capacities tend to be important delimiters of efficiency in interactions with firms. An understanding of this allows articulations and alliances to be fostered between academia and firms, increasing capacities in a highly competitive environment.

Our objective here is to contribute to this area of investigation by analyzing universities (at the level of their research groups) and their science and technology output, as well as their interactions with firms, making comparisons between scientific and technical capabilities and the capacity for interaction with firms on the basis of the quadrants proposed by Stokes (1997/2005). To this end, we use data from the BR Survey, conducted in 2008 with the participation of research groups registered in the research group directory (diretório de grupos de pesquisa, DGP) of the National Council for Scientific and Technological Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico, CNPQ).

This study is divided into three sections in addition to this introduction and the final considerations. In section two, we describe the role of universities in the innovation system, as well as the specificities of university-firm (U-F) interaction in Brazil. In section three, we present the database and debate the methods of analysis. In section four, we discuss the results we obtained through our analysis.

2.Universities: scientific and technological capacities

Knowledge generation is a dynamic and interactive process. The main actors in this process are universities (Arocena & Sutz, 2001), which are fundamental to both the creation and dissemination of new knowledge and technologies by way of basic and applied research, development, and engineering. Therefore, universities serve as incubators of new ideas that can be transferred to new actors in the innovation system – by forming a network of interactions – and applied to production processes, often leading to innovations (Chiarini & Vieira, 2012a).

Thus, the recognition of the role of universities in creating new knowledge has stimulated the production sector’s interest in accessing them. The debate on U-F interaction is rather extensive in the international2 and national3 literature and, other variances notwithstanding, does not reduce the role of universities to that of merely supporting the technological development of certain sectors of economic activity, instead placing equal weight on their function as basic knowledge producers (Chiarini & Vieira, 2011).

U-F interaction – as well as research and development (R+D) collaboration, for instance – is justified as a means of reducing the uncertainties inherent to the innovation process (Tether, 2002), diluting risks related to innovative activities (Hagedoorn, Link, & Vonortas, 2000). The relationship can benefit companies by facilitating the development of capacities, learning, and acquisition of knowledge and technologies (Marques, Freitas, & Silva, 2007), thus contributing to the completion of industrial projects and/or facilitating the implementation of new projects.

Although U-F interactions have the potential to create and disseminate new knowledge and eliminate the risks associated with innovative activities, according to Arza (2010), they can also lead to the privatization of public research outcomes and color the research agenda (in the interests of private actors), distancing the process from the pursuit of more socially desirable knowledge. Thus, the production of new technical-scientific knowledge at universities casts aside the function of mere “enlightenment” without economic ends, to become more oriented towards creating economic value and expressing capitalist competition.

Although the role of universities in creating both basic (pure) and applied science is now recognized, for a long time the two approaches were regarded as conceptually distinct. The former has the objective of expanding comprehension of phenomena in a given scientific field; that is, it ought to be discharged without practical or profit-making ends, inspired by the pursuit of understanding alone. Known as natural philosophy, its results were centered on the discovery of metaphysical truths about the nature of the universe (Noble, 1979). On the other hand, practical men – “technologists” –- had little to no concern for abstract theories, their interest being centered more on utility and profits. Therefore, applied science was seen to be inspired by considerations of use; that is, oriented to a certain specific need to solve practical problems. According to Nelson (1959), applied research was unlikely to shift scientific paradigms, except by mere chance, and it was through basic research that additions were made to the store of knowledge.

The categorical separation of basic and applied research led, erroneously, to the division of science and technology: science as the sole outcome of pure research, and technology as the result of applied research. This “divorce” led to a conception of basic research as the initial stage of the scientific development process, giving way to applied research and then to innovation. The process was known as a “linear model” (science-pushed), and has been widely criticized for its inherent lack of feedback; that is, applied research does not influence basic research, and neither marketing nor users influence basic or applied research (Kline & Rosenberg, 1986).

Despite such heavy criticism, the model served as the basis for countless public policies, and was even used as justification for public funding of scientific research and as an incentive for productive-sector investment in R+D. The premise was that the results of basic science would materialize once they were turned into technological innovations by way of technology transfer processes. From this perspective, science was regarded as the main source of technological innovation.

The distinction between science and technology led to the conclusion that science was only developed at universities and public research laboratories, while it was scientists working for private firms who were concerned with developing technology.(Nelson, 1982) However, challenging this conception, there are many examples throughout history of fundamental advances being made while working on practical or applied problems. Indeed, there is concrete evidence to show that both the pursuit of understanding and its application can influence research choices. The emblematic case is that of the French scientist Louis Pasteur, who:

[…] pursued a fundamental understanding of the processes of disease and other microbiological processes that he gradually discovered as he progressed through the successive studies of his notable career. But nor is there any doubt that he sought such understanding to achieve the applied objectives of preventing deterioration in the production of vinegar, beer, wine and milk, and to defeat flacherie in silkworm, anthrax in sheep and cattle, and rabies in animals and human beings. […] as Pasteur’s studies became progressively more fundamental, the problems chosen by him and the lines of research adopted were progressively more applied. […] Pasteur […] never carried out a study that was not applied, while he shaped a completely new branch of science (Stokes, 1997/2005, p. 31).4

There are other examples. In the field of economics, John Maynard Keynes had a strong desire to understand macroeconomic dynamics at a fundamental and basic level (to contribute to the advancement of economic theory), but he also sought to overcome the economic depression of 1921 by coming up with practical solutions to real problems (Stokes, 1997/2005).

These cases fit within what Stokes (1997/2005) calls Pasteur’s quadrant. Drawing on historical examples, Stokes conducts a bidimensional analysis of research objectives; that is, to expand knowledge for practical ends. Stokes plots two axes: a vertical one, representing the pursuit of pure understanding, and a horizontal one, taking into consideration how knowledge will be used. On this basis, he proposes four different quadrants: those of Pasteur, Bohr, Edson, and Ruetsap.

Pasteur’s quadrant refers to the scientific fields dedicated to solving specific practical problems and furthering understanding at the same time. As such, it concedes the possibility that significant advances in scientific knowledge can also have practical value. New knowledge can serve as an input for future research. Therefore, basic research can create substantial externalities for applied research.

This quadrant provides theoretical support for the idea that the progress of knowledge is founded on basic research, which in turn is inspired by its utilization By placing knowledge in “motion” and meeting social demands, this conceptualization can act as the basis for a new understanding between the scientific and political communities. A breakdown of the other three quadrants proposed by Stokes (1997/2005) is provided in Table 1.

Table 1
Quadrant model of scientific research

Is the research inspired by usage considerations?
		No	Yes
Is the research inspired by the pursuit of fundamental knowledge?	Yes	Pure basic research (Niels Bohr’s quadrant)	Basic research inspired by use (Louis Pasteur’s quadrant)
	No		Pure applied research (Thomas Edison’s Quadrant)

Source: Stokes (1997/2005, p. 118).

Edison’s quadrant considers applied research as that which aspires towards technological development without pursuing advances in understanding. Such research possesses some scientific importance, and science is used with strategic perspectives. And oft-cited example is the electric light system developed by Thomas Edison.

In turn, Bohr’s quadrant represents the utilization of basic research with no immediate application. There is no commitment to developing a specific product or process. The objective is to interpret natural phenomena. One such example is the research of Niels Bohr, whose contributions to quantum theory earned him the Nobel Prize in Physics in 1922.

Finally, Ruetsap’s quadrant is the anti-science quadrant (the blank quadrant shown in Figure 1); that is, it represents societal needs that are not covered by the other quadrants. It is used to classify social, scientific, and technological phenomena in areas outside academia, as well as research motivated by the curiosity of the researcher; Stokes (1997/2005) defines these as “particular” events, and cites birdwatching as an example.

Table 1 can be used to classify individual researchers and the research groups in which they are active, according to their scientific and technical capacities; that is, their capacity to further basic knowledge and their capacity to carry out applied research with the aim of achieving technological developments (Table 1). Thus, it is expected that the greater the scientific and technological capacity, the greater the interactions between basic and applied research; that is, between universities and the productive sector. In this way, according to Stokes’s (1997/2005) logic, if there is a high level of scientific capacity but a low level of technological capacity, few interactions between universities and firms will be expected. The same is expected to hold true if there is a high level of technological capacity and a low level of scientific capacity. Most interactions occur in Pasteur’s quadrant, where there is a high level of both scientific and technological capacity.

Table 2
Classification of scientific and technological competences in the quadrants proposed by Stokes

		Is the research inspired by usage considerations?
		No	Yes
Is the research inspired by the pursuit of fundamental knowledge?	Yes	Pure basic research Bohr’s quadrant High level of scientific capacity Low level of technological capacity	Basic research inspired by use (Pasteur’s quadrant) High level of scientific capacity High level of technological capacity
	No	Low level of scientific capacity Low level of technological capacity	Pure applied research (Edison’s Quadrant) Low level of scientific capacity High level of technological capacity

Source: Stokes (1997/2005, p. 118); compiled by authors.

Stokes’s (1997/2005) efforts to distinguish between basic and applied research notwithstanding, his main contribution lays in exposing the inadequacy of the linear model and showing that pure and applied science are practices with different characteristics. For Dasgupta & David,(1994), the distinction between basic and applied research on the basis of their respective objectives is irrelevant. For them, the difference resides in the social behavior of different groups of scientists in the research community. In sum, the social organization of science and technology are separate from each other.

Indeed, the main difference between these two types of research communities is not the research method, nor the nature of the knowledge obtained, nor even the source of financial resources that makes the research possible, but the behavioral norms of each community – especially when it comes to publicizing research results and the compensation system. Hence, Dasgupta & David’s (1994) contribution has been to separate “academic scientists” from “industrial scientists.” From their perspective, what distinguishes one group from the other is the structure of the socioeconomic rules in which the research is carried out, and what each group does with its research results. Therefore, neither the cognitive ability of each nor the research objective in question – whether to further knowledge or for practical ends – is important in this regard. This being the case, distinguishing between basic and applied research based on their objectives is inconsequential.

3.Database and methodology

3.1Data

The DGP-CNPQ5 began in 1992 and since then it has published a biannual census of installed research capacity in Brazil, measured by the groups that are active in each period. A research group is defined as a set of researchers, students, and technical support teams organized around the execution of research lines that follow a hierarchal law based on specialty and technical-scientific capacity. As such, the DGP-CNPQ gathers information from various institutions, such as federal, state, and private universities; research institutes; public technological institutes; public and private R+D laboratories; and non-governmental organizations permanently involved in scientific and technological research.

The information gathered, broken down over time by region, federal unit, and institution, includes: the human resources6 that comprise the groups, such as researchers, students, and technicians; research lines pursued by the groups; areas of knowledge; sectors of activity involved; scientific and technological output of the researchers and students in the groups; and patterns of interaction with the private sector.

In 2002, the CNPQ introduced specific questions about interactions between firms and other research institutions, establishing itself as a source of information about U-F relations in Brazil. However, it should be noted that the research group leaders underestimate the number of interactions (Rapini, 2007a, 2007b). That is, more interactions take place than those actually reported by the leaders, since, according to Rapini (2007b), there are inherent differences in the questionnaire itself and in the content of the optional answers, which can limit its completion.

Registration in the DGP-CNPQ is voluntary, but researchers are strongly motivated to participate, largely because an up-to-date curriculum vitae on research is a prerequisite for access to public financing and scientific and technological research. It should be noted that interaction with the productive sector is not a criteria used by development agencies for assessing researcher performance, which may also explain the apparent underestimation. Despite these limitations, the DGP-CNPQ universe has grown in recent years, and now covers a sizable proportion of Brazil’s scientific community (Carneiro & Lourenço, 2003).

After the 2004 census, participating entities were sorted into two groups: firms,7 and research groups affiliated with universities and research institutes. Based on this classification, a survey8 was prepared for each of the groups and was sent to them to fill out. Its objective was to determine the characteristics of U-F interactions in Brazil. This was known as the “BR Survey” and was conducted in 2008.

The BR Survey included questions about U-F interactions, covering aspects such as:

a)Forms of interaction;

b)Outcome of the interaction;

c)Benefits of the interaction;

d)Difficulties with the interaction;

e)Channels of information; and

f)Sources of financing.

Moreover, certain characteristics of the research groups, such as scientific production, technological production, number of interactions, time active, and main scientific area of operation, were included as part of the survey.

Both the firms and the research groups answered questions about interactions that took place in the three years immediately preceding the survey. The university questionnaire was sent to the heads of the 2,151 groups registered in the 2004 census; responses were received from 1,005 groups, equaling a response rate of 46.7%. Meanwhile, the firm questionnaire was sent to 1,622 firms, of which 319, or 18.9%, responded.

The BR Survey provides information relevant for analyzing the characteristics of research groups in Brazil based on their science and technology output, and their capacity for interaction with the productive sector. In the next section we present exercises based on this information, taking into account approximations based on the quadrants proposed by Stokes and observations of the groups’ behavior based on the major areas of scientific knowledge. First, we will introduce the methodology for the model.

3.2Methodology

The objective of this study is to understand the characteristics of research groups in terms of their scientific and technological capacity, as well as the way in which these capacities relate to the capacity for interaction with firms, taking into account the positioning of the groups in the quadrants proposed by Stokes.

Thus, we classify the research groups in order to capture their different capacities and sort them into the four quadrants, with the aim of exploring the relationship between these scientific and technological capacities and the capacity for interaction with the productive sector. In sum, we seek to verify whether, in the Brazilian case, the research groups located in Pasteur’s quadrant – with high scientific and technological capacities – are those with the greatest capacity for interaction with the other actors, as inferred by the work of Stokes.

For this exercise, we opt to use the statistical technique known as qualitative comparative analysis (QCA), commonly used in experiments in the social sciences. We selected this technique for its ability to find different patterns of behavior that lead to the same result: in this case, capacity for interaction (Rihoux & Ragin, 2009). It is therefore unnecessary to establish a single specific causal model that best fits the data, instead allowing for the identification of outliers and/or opposite characteristics that give rise to the same result, the so-called conjunctural causation (Ragin, 1987; Rihoux & Ragin, 2009).

Moreover, we take into account the ongoing discussion of causation in science, technology, and innovation (ST&I) regarding the following questions: Does science generate technology, and does this in turn generate interaction? Does interaction generate more technology? Does technology generate science? Thus, the QCA method is more appropriate than a conventional econometric model, since in this method it is not necessary to determine the causal direction, only to prove the relationship between the variables.

As to conjunctural causation, we consider the possibility that variables A, B, C, D, and E exist as possible variables that generate the same output, Y. Thus, the application of QCA can lead to the conclusion that the configuration of factors AB or ACD give rise to Y (AB or ACD → Y). Or, depending on the context, the lesser value may be important to attaining the same result: AB → Y, but also aC → Y, where [a] is the designation for values below factor [A].9

For the purposes of this study, we use the fuzzy QCA (fsQCA) method, which enables estimation of the membership score of a group within the interval between 0 and 1. The rule used to assess the relationship between the output and the related variables, in the case of the fuzzy method, is the inclusion ratio (Longest, Vaisey, & Fuzzy, 2008):

where:

X denotes the configuration of the predictor (i.e., AB)

Y denotes the output

defines each membership in the configuration X

denotes each membership in the configuration Y.

Given the conditional probability, the closer the value of is to the unit, the greater the consistency of the data when it is affirmed that X is a subset that defines Y (X → Y). Several methods can be used to decide which configuration of predictors X is sufficient to explain Y; we base this study on Ragin (2006), who defines a benchmark of 0.700, which means that all configurations in which are sufficient for determining the membership between X and Y.

We implement an internal minimization procedure to identify the sets of conditions that encompass all possible results. If found, these sets define the conditions X, which define Y. Finally, the model allows the best configuration (the best fit) to be presented for each observation (in the specific case of this study, the research group), considering the variables in comparison with the result Y and the other groups analyzed.

For the presentation, we perform two analyses using the QCA method. The first, Model 1, seeks to associate the results of scientific and technological capacities that have total values with the research group’s total number of interactions; while the second, Model 2, makes the same association but with values relative to the number of researchers in the group as variables. The variables for each model are set out in tables 3 and 4.

Table 3
Variables investigated using QCA, Model 1

Model 1 Total values

Output variable

1.1 Number of interactions with the productive sector

Input variable

1.2 Publications in the Institute for Scientific Information (ISI) and the Scientific Electronic Library Online (SciELO) databases. 1.3 Applications for patents, licenses and software.(1) 1.4 Group’s time in operation (stated period for which the group in the database has been active, up to 2008).

Note: (1) We use the sum of the different technological outcomes given the distinctions in their uses across research groups, without making judgements about the relevance of each one. Of the 902 groups analyzed, only eight (i.e., 0.9%) possessed the three types of outcomes, while just 29 (3.2%) held patents and licenses (outcomes normally regarded as highly correlated). Moreover, this form of measurement, which considers technology apart from expectations, does not create problems for the analysis, as we will show in the following section.

Table 4.
Variables investigated using QCA, Model 2

Model 2 Relative values

Output variable

2.1 Interactions with the productive sector per operator: total number of interactions divided by the number of registered researchers(1) in the research group

Input variables

2.2 Publications per researcher: total number of publications recorded in the ISI and the SciELO divided by the number of registered researchers(1) in the research group. 2.3 Applications for patents, licenses, and software per researcher: total number of patent, license, and software applications divided by the number of registered researchers(1) in the research group. 2.4 Group’s time in operation (stated period for which the group in the database has been active, up to 2008).

Note: (*) Only researchers with doctoral and postdoctoral qualifications are taken into account.

In the two models proposed, we consider the group’s operating time as a way of controlling for longevity and path dependence in technological and scientific capacities. Of the 1,005 research groups in the BR Survey, we took into account 902 (89.8%): those working in the following major areas of scientific knowledge: earth and biological sciences, engineering, and exact and Earth sciences.10

4.Discussion and results

Our presentation of the results takes into account two aspects: the minimum configurations found by the minimization process, and the best configurations (best fits) for each research group analyzed. As to the minimization process, the application of QCA did not find possibilities of minimum results for any of the models, so it was not possible to present any results.11

The non-existence of results through the minimization process gives rise to an important consideration for U-F interactions in Brazil: statistically, there are no results that define capacity for interaction. Any combination of values concerning the scientific and technological capacity of the research groups can result in high levels of interaction with firms. In sum, this means that, for example, a group with relatively low scientific capacity (indicator [s] in the model) and relatively low technological capacity (indicator [t] in the model) – located in Ruetsap’s quadrant, the anti-science quadrant in Table 1 – can engage in a large number of interactions, in the same way as a combination group [ST], that is, one with high scientific and technological capacity, located in Pasteur’s quadrant in Table 1. Moreover, consideration of the research group’s time in operation likewise did not allow a minimum configuration, indicating the limited importance of questions related to the accumulation of scientific and technological knowledge as well.

The impossibility of analysis, given the different configurations, leads to high levels of interaction and opens up a second analytical possibility: the observation of the configurations that arise in each research group. In this analysis, we consider the methodology’s approximations to the theoretical positioning of the quadrants. Given the definitions of technological and scientific capacities set out in Table 2, the connections between the minimum configurations of science ([S] for high and [s] for low) and technology ([T] for high and [t] for low) with the quadrants can be understood as being the following:

Configuration St → Bohr’s quadrant

Configuration ST → Pasteur’s quadrant

Configuration sT → Edison’s quadrant

Configuration st → Ruetsap’s quadrant

Thus, we distinguished the groups according to these minimum configurations, as set out in tables 5 and 6.12

Table 5
QCA Models

Variables	Model 1. Total values				Model 2. Relative values
	Bohr’s Quadrant	Pasteur’s Quadrant	Edison’s Quadrant	Ruetsap’s Quadrant	Bohr’s Quadrant	Pasteur’s Quadrant	Edison’s Quadrant	Ruetsap’s Quadrant
	St	ST	sT	st	St	ST	sT	st
Research group	192	140	124	446	189	145	120	448
Publications	10,319	12,444	156	501	10,336	12,431	169	484
Technology	0	688	374	0	0	686	374	0
Interactions with the production sector	877	935	384	2,267	858	946	378	2.281
Group’s average time in operation	12.54	14.19	12.91	11.66	12.41	13.97	13.05	11.74
Publications per researcher(1)	11.52	16.39	0.40	0.26	11.96	16.01	0.20	0.14
Patents, licences and software per researcher(1)	0	1.06	0.79	0	0	1.10	0.74	0
Interactions with the production sector per researcher (1)	1.05	1.53	0.86	1.60	1.19	1.58	0.77	1.54

Note: (1) Average values.
Source: BR Survey; prepared by authors.

Table 6
Number of research groups in each quadrant, per major scientific area

Scientific areas	Model 1 Total values				Model 2 Relative values
	Bohr’s Quadrant	Pasteur’s Quadrant	Edison’s Quadrant	Ruetsap’s Quadrant	Bohr’s Quadrant	Pasteur’s Quadrant	Edison’s Quadrant	Ruetsap’s Quadrant
	St	ST	sT	st	St	ST	sT	st
Agricultural sciences	40	22	28	110	39	24	26	111
Percentage	20.0	11.0	14.0	53.5	19.5	12.0	13.0	55.5
Biological and health sciences	53	32	29	107	62	31	31	102
Percentage	24.0	14.5	13.1	48.4	27.4	13.7	13.7	45.1
Engineering	64	55	47	157	57	55	46	160
Percentage	19.8	17.0	14.6	48.6	17.9	17.3	14.5	50.3
Exact and Earth sciences	35	31	20	72	31	35	17	75
Percentage	22.2	19.6	12.7	45.6	19.6	22.2	10.8	47.5

Source: BR (2006); compiled by authors.

For the analysis presented in tables 5 and 6, we consider the models that measure capacity by total quantity (model 1), and capacity relativized by human capital, measured by the total number of doctoral and postdoctoral researchers (model 2). Finally, we present the number of research groups per major scientific area in each quadrant. As can be noted, the results of the two models are close in terms of the capacity for interaction of research groups in Brazil, and the related variables. Taking this into consideration, for the most part our analyses will take both results into account.

In sum, the quadrants clearly correspond to the theoretical proposition. For instance, the research groups with high levels of scientific capacity and no technological capacity are located in Bohr’s quadrant. In Edison’s quadrant, the reverse is true: it hosts those with high levels of technological capacity and low scientific capacity. Pasteur’s quadrant includes research groups with high scientific and technological capacities and, as such, can be expected to contain high capacity for interactions with the productive sector, while the reverse is the case in Ruetsap’s quadrant.

Another important observation is the divergence between the scientific and the technological results. Even those groups in Pasteur’s quadrant, in comparison to those with a higher technological profile, have low technological capacity in relation to their scientific capacity, which coincides with the findings of Ribeiro, Albuquerque, Franco, & Moura (2009) and of Suzigan & Albuquerque (2009).

Observing the distribution of the research groups in the quadrants gives an idea of the non-observance of a minimum configuration in the explanation of interaction with firms. The research groups are mainly located in Ruetsap’s quadrant (49.7%), and have low values for scientific and technological capacities but, against expectations, exhibit a large number of interactions with firms. In relative terms, capacity for interaction is similar to that of the research groups situated in Pasteur’s quadrant, even though these score well for publications and technology (total and relative); indeed, they are the groups with the highest values for these variables. Moreover, the research groups in Edison’s and Bohr’s quadrants also possess high relative values for interaction, despite their differences in terms of science and technology (S&T).

This result is unprecedented among studies of Brazilian academia since it shows that research groups without scientific or technological capacity have engaged in more interactions in the total quantum and in the nearby values of the analysis relative to the number of researchers, in comparison to the groups with greater capacity: those in Pasteur’s quadrant. This observation illustrates the need to study the quality of U-F interactions, since the expectation is that these research groups will be less likely to achieve significant developments in the field of ST&I.

Establishing the group’s time in operation does not alter the observations. Although the average operating time of the research groups in Ruetsap’s quadrant is lower than the others, this does not amount to a significant difference in academic performance; this is because the difference in comparison with the research groups in Pasteur’s quadrant is 2.38 years, based on the average of the two models. It is not reasonable to consider that this period can make a difference to S&T capacity for groups with an average of 11 to 14 years in operation.13

Finally, the observation of differences in the number of groups while taking into consideration the major scientific areas gives an indication of the minimal differentiation between these areas. It can be established that: the biological and health sciences are present in Bohr’s quadrant in relatively greater proportion than the groups from the remaining areas; the research groups representing the exact and Earth sciences are proportionately greater in Pasteur’s quadrant; engineering is most represented in Edison’s quadrant; while agricultural sciences are concentrated in Ruetsap’s quadrant. These proportions, however, vary little, and are not any stronger than the observation that all major scientific areas are represented to the greatest extent by groups in Ruetsap’s quadrant.

Having observed the characteristics of the research groups in the classification using the quadrants proposed by Stokes, our next step is to establish whether there are significant differences between these groups in terms of the characteristics of U-F interactions, especially as regards the following:

a)importance of the types of interactions established;

b)importance of the different results;

c)importance of the benefits.

The leaders of the research groups who responded were given the option of classifying the levels of importance of their group on a scale of one to four, where:

1 = Unimportant

2 = Somewhat important

3 = Moderately important

4 = Very important

Based on this classification, we calculated the average for each cohort of research groups in the quadrants. The specific question we seek to answer is: Might the similarities in the capacity for interaction across research groups in the four quadrants – especially where the extremes, those of Pasteur and Ruetsap, are concerned – be a product of different characteristics in the types of interaction engaged in (in terms of forms of interaction, outcomes achieved, and benefits generated)? We regard this as an important question since it could lead to the identification of distinctions in the types of interactions most linked to different capacities, which would involve the understanding of the same “quantity” but different “quality” across the four quadrants. In figures 1 to 3, we proceed with the comparative analysis of these characteristics.

In Figure 1, we find that there are very close average values for the relationship types, with greater significance in the case of: R+D with immediate outcomes, consulting, and training, whereby the latter two forms of interaction are those that coincide with the results found by Rapini (2007a). As to the average importance of the outcomes, Figure 2 clearly illustrates the trend of higher importance attributed to those values closest to scientific outcomes such as training of human resources/students; theses and dissertations; and new research projects. Figure 2 also shows that the questions concerning technology-related outcomes, such as patents, software, design and spin-off, present the lowest average values. As regards the benefits, Figure 3 points to a lower discrepancy in the average comparisons of the values; but even then, we find somewhat higher average values for new research projects, ideas for new research projects; and exchange of knowledge, again highlighting the scientific profile to which the research groups commit in their interaction with the firms.

Figure 1
Average level of importance of forms of interaction by quadrant models

Source: BR (2006); compiled by authors.

Figure 2
Average level of importance of outcomes by quadrant models

Source: BR (2006); compiled by authors.

Figure 3
Average level of importance of benefits by quadrant models

Source: BR (2006); compiled by authors.

Beyond these observations, what needs to be highlighted in the results is that the similar capacities for interaction across research groups is mimicked by the characteristics of these interactions; indeed, we found no significant differences between the research groups from different quadrants in terms of the types, results, and benefits of interactions. Despite the different science- and technology-related skills identified in the analysis, the research groups classified do not differ in their responses to important questions about the characteristics of their interactions with firms.

One interesting result is that while it is possible to identify a cohort of groups with greater capacity in technological terms (those in Pasteur’s and Edison’s quadrants), their responses in relation to the characteristics of the interaction are no different from those of the groups with low scientific capacity (Bohr) or low scientific and technological capacity (Ruetsap’s quadrant).

5.Final considerations

The scientist detached from mundane concerns has gradually given way to the reality of the scientific community, of intellectual workers organized at universities and firms, forming part of so-called big science. Science ceased to be seen as a process governed by mere laws of creativity – as an autonomous and independent entity of society – and came to be regarded as a product of society. Thus, advances in scientific research increasingly came to be oriented towards social and economic objectives.

Along the way, the generation of new knowledge at universities – the main locus of production of new scientific knowledge – has once again become the focus of interest of firms and the state alike. Universities thus cease to be, in the words of Mowery & Sampat (2005), “ivory towers” oriented to the pursuit of knowledge per se, and are seen as strategic assets. Thus, the main contribution of this study is to advance research in the area of the economics of science and technology, aiding understanding of the dynamics of output in this area in the case of Brazilian academia.

Our concern therefore lies in the capacities of the research groups that participated in the BR Survey to generate science and technology, and in how these capacities might relate to interactions with firms. To this end, we utilized a classification based on the quadrants proposed by Stokes (1997/2005) through the QCA statistical technique, which enables verification of conjunctural causation and makes it possible to find different patterns of behavior that lead to the same results.

By using this technique, we were able to prove that there is no well-defined pattern of scientific and technological capacity among the research groups that would lead to high levels of interaction with firms, both in terms of the total quantum of interactions and of capacity relative to the number of researchers. This non-existence of a minimum pattern of necessary conditions is an initial result, allowing us to conclude that the scientific and technological capacity of the research groups is not a necessary condition for their inclusion in U-F interactions.

The verification through QCA of better science and technology configurations for research groups serves to emphatically corroborate this result. We found, first, that the largest number of interactions occurred in Ruetsap’s quadrant; and, second, that capacity relative to the number of researchers is similar between groups in Ruetsap’s and Pasteur’s quadrants, the two extremes of scientific and technological capability.

According to this result, unprecedented among studies of Brazilian academia, research groups with low scientific or technological capacity have engaged in more interactions in the total quantum, with close values in the analysis relative to the number of researchers, as compared to the groups with greater capacity: those in Pasteur’s quadrant. This observation illustrates the need to study the quality of U-F interactions, since according to the theory we would expect the research groups in this quadrant to have less interactions with firms.**

Having tested for the possibility of distinctions in the types, outcomes, and benefits of interaction across the research groups in the different quadrants, the findings corroborate the similarity between profiles, with little to no difference in the average pattern of responses. Finally, we find insufficient difference between the major scientific areas to mark them apart.

This information attests to the already noted immaturity of Brazil’s national innovation system, as expressed by Albuquerque, and the low trade-off between science and technology in this system, as argued by Ribiero et al. (2009), Suzigan & Albuquerque (2009), and Fernandes et al. (2010). One the one hand, there is a reasonable number of research groups that develop a good deal of science but little technology, whether in Bohr’s quadrant or in Pasteur’s. On the other hand, there is an even greater number of groups that develop neither science nor technology but do undertake U-F interactions to a greater extent than the aforementioned groups, with a relative percentage similar to that of the higher-capacity research groups.

It should be noted here that the results are a product of important but now dated research, and that the findings ought to be corroborated with more recent information, which will also serve to provide robust support for the use of these results in public policies aimed at promoting science and technology in Brazilian academia. However, when we used Stokes’s proposal to find differences in capacity between research groups in Brazil, a notable gap was found between capacity and the number of interactions engaged in. Caliari et al. (2016) note that the greater the number of interactions with firms, the greater the capacity of the university or research institution to generate technology. We would expect this result to hold for the base of research groups observed here, with the addition of an improvement in science, thus establishing the causation necessary to promote better capacities in the Brazilian NIS.

References

Albuquerque, E. M. (1998). Produção científica e sistema nacional de inovação. Ensaios FEE, 19(1), 156-180.

Albuquerque, E. M. (1999). National Systems of Innovation and Non-OCED Countries: Notes About a Rudimentary and Tentative “Typology.” Brazilian Journal of Political Economy, 19 (4-76), 35-52.

Albuquerque, E. M., Silva, L. A., Rapini, M. S., & Souza, S. G. A. (2005). Interactions Between Firms and Universities in an Immature System of Innovations: A Survey of Industrial R&D-Performers Firms in Minas Gerais. Texto para discusión N° 280. Belo Horizonte: UFMG/Cedeplar.

Arocena, R., & Sutz, J. (2001). La universidad latinoamericana del futuro: tendencias, escenarios, alternativas. Mexico: Udual.

Arza, V. (2010). Channels, Benefits and Risks of Public-Private Interactions for Knowledge Transfer: A Conceptual Framework Inspired by Latin America. Science and Public Policy, 37(7), 473-484.

Britto, G., Santos, U. P., Kruss, G., & Albuquerque, E. M. (2015). Redes globais de inovação e interações universidade-empresa: uma análise exploratória de dados. Revista Brasileira de Inovação, 14(1), 163-191.

Bush, V. (1945). Science: The Endless Frontier. Washington: United States Government Printing Office.

Caliari, T., & Rapini, M. S. (2017). Diferenciais da distância geográfica na interação universidade-empresa no Brasil: um foco sobre as características dos agentes e das interações. Nova Economia. Forthcoming.

Caliari, T., Santos, U. P., & Mendes, P. S. (2016). Geração de tecnologia em universidades/institutos de pesquisa e importância da interação com empresas: constatações através da base de dados dos grupos de pesquisa do CNPQ. Análise Econômica, 34, 285-312.

Carneiro, S. J., & Lourenço, R. (2003). Pós-graduação e pesquisa na universidade. En E. B. Viotti, & M. Macedo (Orgs.). Indicadores de Ciência, Tecnologia e Inovação no Brasil (pp. 169-227). Campinas: Editora da Unicamp.

Chiarini, T., Oliveira, V. P., & Silva Neto, F. C. C. (2013). Spatial Distribution of Scientific Activities: An Exploratory Analysis of Brazil, 2000-10. Science and Public Policy, 41, 625-640.

Chiarini, T., Rapini, M. S., & Vieira, K. P. (2014). Produção de novos conhecimentos nas universidades federais e as políticas públicas brasileiras recentes de CT&I. Revista Economia & Tecnologia, 10, 71-98.

Chiarini, T., & Vieira, K. P. (2011). Alinhamento das atividades de pesquisa científica e tecnológica realizadas pelas IES federais de Minas Gerais e as diretrizes da política industrial, tecnológica e de comércio exterior? Pitce. Revista Brasileira de Inovação, 10, 301-342.

Chiarini, T., Vieira, K. P. (2012a). As universidades federais mineiras estão se tornando mais desiguais? Análise da produção de pesquisa científica e conhecimento (2000-2008). Revista Educação e Pesquisa, 38(4), 897-918.

Chiarini, T., & Vieira, K. P. (2012b). Universidades como produtoras de conhecimento para o desenvolvimento econômico: sistema superior de ensino e as políticas de CT&I. Revista Brasileira de Economia, 66, 117-132.

Cruz, C. H. B. (2004). A universidade, a empresa e a pesquisa. En Seminário Brasil em Desenvolvimento da Universidade Federal do Rio de Janeiro, UFRJ, Río de Janeiro. Retrieved from http://www.ifi.unicamp.br/~brito/artigos/univ-empr-pesq-rev102003b.pdf

Cruz, C. H. D. B. (2014). “Sência: a fronteira sem fim”, uma apresentação. Revista Brasileira de Inovação, 13(2), 241-247.

Dagnino, R. (2003). A relação universidade-empresa no Brasil e o “argumento da hélice tripla.” Revista Brasileira de Inovação, 2(2), 267-307.

Dasgupta, P., & David, P. A. (1994). Toward a New Economics of Science. Research Policy, 23, 487-521.

Dutrenit, G., & Arza, V. (2010). Channels and Benefits of Interactions Between Public Research Organisations and Industry: Comparing Four Latin American Countries. Science and Public Policy, 37(7), 541-553.

Esteves, L. A., & Meirelles. J. G. P. (2009). Innovation and Cooperation Between Firms and Universities: Evidence from Brazil. In Third Conference on Micro Evidence on Innovation and Development, Meide. Río de Janeiro, Brasil.

Fernandes, A. C, Campello de Souza, B., Stamford da Silva, A., Suzigan, W., & Albuquerque, E. (2010). Academy-Industry Links in Brazil: Evidence About Channels and Benefits for Firms and Researchers. Science and Public Policy, 37(7), 485-918.

Freeman, C., & Soete, L. (1974/2008). A economia da inovação industrial. Campinas: Editora da UNICAMP.

Friedman, J., & Silberman, J. (2003). University Technology Transfer: Do Incentives, Management, and Location Matter? The Journal of Technology Transfer, 28(1), 17-30.

Hagedoorn, J., Link, A. N., & Vonortas, N. S. (2000). Research Partnerships. Research Policy, (29), 567-586.

Harmon, B., Ardishvili, J., Cardozo, R., Elder, T., Leuthold, J., Parshall, J., Raghian, M., & Smith, D. (1997). Mapping the University Technology Transfer Process. Journal of Business Venturing, 12(6), 423-434.

Kline, S. J., & Rosenberg, N. (1986). An Overview of Innovation. In R. E. Landau, & N. Rosenberg (Orgs.). The Positive Sum Strategy: Harnessing Technology for Economic Growth (pp. 275-306). Washington: National Academy Press.

Longest, K. C., Vaisey. S., & Fuzzy: (2008). A Program for Performing Qualitative Comparative Analyses (QCA) in Stata. The Stata Journal, 8(1), 79-104.

Markmana, G. D., Gianiodisa, P. T., Phanb, P. H., & Balkinc, D. B. (2005). Innovation Speed: Transferring University Technology to Market. Research Policy, 34(7), 1058-1075.

Marques, R. A., Freitas, I. M. B., & Silva, E. M. P. (2007). Colaboração com universidade e as atividades para inovação de empresas brasileiras. Engevista, 9(2), 112-128.

Marx, K. (1890/1985). O capital: crítica da economia política. Livro 1, O processo de produção do capital (10th ed.). São Paulo: Difel Difusão Editorial.

Mazzoleni, R. (2005). Historical Patterns in the Coevolution of Higher Education. Public Research and National Industrial Capabilities. Viena: Unido.

Mazzoleni, R., & Nelson, R. (2005). The Roles of Research at Universities and Public Labs in Economic Catch Up. Pisa: Laboratory of Economics and Management Sant’Anna School of Advanced Studies.

Mello, J. M. C., Maculan, A. M., & Renault, T. (2009). Brazilian Universities and their Contribution to Innovation and Development. In B. Goransson, & C. Brundenius (Orgs.). Developing Universities: The Changing Role of Academic Institutions in Development (pp. 1-25). London: Routledge e IDRC.

Mowery. D., & Sampat. B. (2005). Universities in National Innovation Systems. In J. Fagerberg, D. Mowery, & Nelson. R. The Oxford Handbook of Innovation. Oxford University Press.

Nelson, R. (1959). The Simple Economics of Basic Scientific Research. Journal of Political Economy, 67, 297-306.

Nelson, R. R. (1982). The Role of Knowledge in R&D Efficiency. The Quarterly Journal of Economics, 97(3), 453-470.

Noble, D. F. (1979). America by Design: Science, Technology, and the Rise of Corporate Capitalism. New York: Oxford University Press.

Pavitt, K. (1991). What Makes Basic Research Economically Useful? Research Policy, 20, 109-119.

Póvoa, L. C., & Rapini, M. S. (December 2009). Technology Transfer from Universities and Public Research Institutes to Firms in Brazil: What is Transferred and How the Transfer is Made. In Encontro Nacional de Economia, 37. Anpec, Foz do Iguaçu, Brasil.

Ragin, C. C. (1987). The Comparative Method. Moving Beyond Qualitative and Quantitative Strategies. Berkeley, Los Angeles and London: University of California Press.

Ragin, C. C. (2006). Set Relations in Social Research: Evaluating the Consistency and Coverage. Political Analysis, (14), 291-310.

Rapini, M. S. (2007a). Interação universidade-empresa no Brasil: evidências do diretório dos grupos de pesquisa do CNPQ. Estudos Econômicos, 37, 211-233.

Rapini, M. S., Suzigan, W., Fernandes, A. C., Domingues, E., Carvalho, S. S. M., & Chaves, C. V. (2007b). O diretório dos grupos de pesquisa do CNPQ e a interação universidade-empresa no Brasil: uma proposta metodológica de investigação. Revista de Economia Contemporânea, 11(1), 99-117.

Rapini, M. S., Suzigan, W., Fernandes, A. C., Domingues, E., Carvalho, S. S. M., & Chaves, C. V. (2009). A contribuição das universidades e institutos de pesquisa para o sistema de inovação brasileira. In Encontro Nacional de Economia, 37. ANPEC, Foz do Iguaçu, Brasil.

Rapini, M. S., Chiarini, T., & Bittencourt, P. F. (2015). University-Firm Interactions in Brazil: Beyond Human Resources and Training Missions. Industry & Higher Education, 29, 111-127.

Rapini, M. S., Oliveira, V. P., & Caliari, T. (2016). Como a interação universidade-empresa é remunerada no Brasil: evidências dos grupos de pesquisa do CNPQ. Revista Brasileira de Inovação, 15(2), 219-246.

Rapini, M. S., Oliveira, V. P., & Silva Neto, F. C. C. (2014) A natureza do financiamento influencia na interação universidade-empresa no Brasil? Revista Brasileira de Inovação, 13(1), 77-107.

Rapini, M. S., & Righi, H. M. (2006). O diretório dos grupos de pesquisa do CNPQ e a interação universidade-empresa no Brasil em 2004. Revista Brasileira de Inovação, 5(1), 131-156.

Renault, T., Mello, J. M. C., & Carvalho, R. (2008). Universities, Organizational Structure of the Research Activity and the Spin-Off Formation: Lessons from the Brazilian Case. In XXV IASP World Conference. ANPEC, Johannesburgo, Sudáfrica.

Ribeiro, L. C., Albuquerque, E., Franco, L. M., & Moura, I. A. (2009). The Scientific and Technological Trajectories of Four Latin American Countries: Mexico, Costa Rica, Argentina and Brazil [draft]. Belo Horizonte: Cedeplar and UFMG.

Rihoux, B., & Ragin, C. C. (2009). Configurational Comparative Methods: Qualitative Comparative Analysis (QCA) and Related Techniques. Applied Social Research Method Series. California: Sage Publications.

Rosenberg, N. (1974). Karl Marx on the Economic Role of Science. Journal of Political Economy, 82(4), 713-728.

Rosenberg, N. (1990). Why do Firms do Basic (With their Own Money)? Research Policy, 19(2), 165-174.

Schaeffer, P. R., Ruffoni, J., & Puffal, D. (2015). Razões, benefícios e dificuldades da interação universidade-empresa. Revista Brasileira de Inovação, 14(1), 105-134.

Schumpeter, J. A. (1911/1985). A teoria do desenvolvimento econômico. Uma investigação sobre lucros, capital, crédito, juro e o ciclo econômico. São Paulo: Nova Cultural.

Schumpeter, J. A. (1928). The Instability of Capitalism. The Economic Journal, 38(151), 361-383.

Schumpeter, J. A. (1942/2008). Capitalism, Socialism and Democracy. New York: Harper Perennial Modern Thought.

Shane, S. (2002). Selling University Technology: Patterns from MIT. Management Science, 48(1), 122-137.

Stokes, D. E. (1997/2005). O quadrante de Pasteur. A ciência básica e a inovação tecnológica. Campinas: UNICAMP.

Suzigan, W., & Albuquerque, E. M. (August 2009). The Underestimated Role of Universities for Development: Notes on Historical Roots of Brazilian System of Innovation. In World Economic History Congress, 15. International Economic History Association. Utrecht, Netherlands.

Tether, B. (2002). Who Co-operates for Innovation, and Why: An Empirical Analysis. Research Policy, (31), 947-967.

Tobin, J. (1958). Estimation of Relationships for Limited Dependent Variables. Econometrica, 26(1), 24-36.

Wright, M., Birley, S., & Mosey, S. (2004). Entrepreneurship and University Technology Transfer. The Journal of Technology Transfer, 29(3-4), 235-246.

---