Some approaches that examine the role of natural resources in economic development: for example, the Dutch Disease, Resources curse, and Resource Blessing approach. On the one hand, there are two approaches with a negative view on natural resources. The Dutch Disease approach suggests that increasing profits in the primary sector induce a flow of resources for this sector and services, which negatively affects the development of the manufacturing sector, and exports can lead to increases in production costs (MATSUYAMA, 1992; GYLFASON, 2001; LARSEN, 2005, 2006). The Natural Resources curse approach refers to the findings of different empirical studies (e.g. MATSUYAMA, 1992; SACHS; WARNER, 1995, 2001) that sought to understand the phenomenon of the rate of economic growth in countries abundant in natural resources compared to countries without the same characteristic (Larsen 2006).

On the other hand, there is a group of researchers with a more positive view of the opportunities that natural resources can generate for development. Authors such as Maddison (1994) and Stijns (2005) argue that the conclusions of Sachs & Warner are distorted and that the proposition is not supported. Contrasting the idea of Natural Resources curse, Stevens (2003), Walker & Jourdan (2003) and Smith (2007) presented evidence that some countries have received a "blessing" and have been able to leverage their economic and technological progress through natural resources. However, for this development to take place, efforts of different natures are needed, for example: human capital, institutions, infrastructure, education, automation, learning and technological dominance (DE FERRANTI, PERRY, LEDERMAN; MALONEY, 2002; LEDERMAN; MALONEY, 2007; SMITH, 2007; TORRES-FUCHSLOCHER, 2010). In general, neither natural resource-based activities nor the related industries undertake intensive R & D efforts. Instead, they rely on knowledge flows from other capital goods and intermediary institutions and industries (SMITH, 2007). Perez (2010) argues that there are windows of opportunity for countries well provided with natural resources and experience. A combination of harnessing the benefits of hypersegmentation, application of IC technologies in the processing industries, enhancement of commodity prices, investment in technological upgrading and diversification of the export mix can improve the technological dominance of a country (or region) to prepare for the next technological revolution and for the creation of social welfare. However, none of these approaches is examined in depth. These studies carry out examinations at a macroeconomic level that do not correspond to the focus of analysis of this study.

In a more recent perspective (last 20 years), with a focus on the Latin American context, a group of economists associated with ECLAC (Economic Commission for Latin America and the Caribbean), concerned about the poor performance of the economy of Latin American countries after the reforms of the 1990s, takes a negative stance on the specialization in natural resources of these countries. This current of thought rescues Prebisch's arguments and postulates of the 1950s. Prebisch (1959) argues that natural resource-based activities face different kinds of supply and demand constraints, with trends of continuous decline in terms of trade. Hirschman's (1958) argument argues that the resource-based industry offers very limited connections to the development of activities in the economy as a whole.

Katz (2000) argues that after the institutional changes of the 1990s (market opening and abandonment of import substitution policy), Latin America has changed its pattern of productive specialization towards natural comparative advantages and capital- intensive firms (e.g. natural resource processing industries). The productive specialization of Latin America was consolidated in a productive pattern of maquila (in the case of Mexico and some Central American countries) and processing of natural resources (in the case of Brazil, Argentina, Chile and other countries in South America) (KATZ; STAMPO, 2001). It is advocated that these industries are specialized in standardized commodities and have low domestic value added (CASTALDI; CIMOLI; CORREA; DOSI, 2009). Some authors argue that these reforms have led countries like Argentina, Chile and Brazil to a "low-development trap" with negative consequences for technological development (KATZ, 2000; OCAMPO, 2001). This same thinking is replicated by ECLAC: "(...) reinforced the pattern of specialization in sector with static comparative advantages. The outcome has been production structure lock-in and technology lag "(ECLAC, 2012, p.44). Authors like Cavalieri, Torres & Hasenclever (2013, p.18) comment that the expressive growth of Brazilian specialization in industries related to natural resources can generate a "structural change limited to enclaves, with low aggregate demand growth and strong productivity increase in few sectors ". Ferraz, Souza & Kupfer (2010) emphasize that the performance of the natural resource intensive industry is a phenomenon of prices, subject to the ups and downs of commodity prices in the markets.

Katz (2007) and Castaldi et al (2009) argue that Latin American companies have so far not shown significant interest in engaging in technology-generation efforts, R&D activities, interactions with universities, laboratories, research centers or knowledge-intensive companies. It is argued that the trajectory of specialization in resource and capital intensive activities is particularly disappointing, since such activities are typically characterized by low technological content and few learning opportunities:

“In terms of specialization patterns, following the trade reforms, many Latin American economies increased their share of production in (i) natural resources and natural resource processing industries (such as pulp and paper, iron and steel, vegetable oil, etc)... […] The last couple of decades have been disappointing. […] The end result is a widening dualism whereby an increasing share of the whole economy is composed of activities typically characterized by a low knowledge content and low opportunities for technological and organizational learning” (CASTALDI et al, 2009, pp. 64-65).

There is a widespread view that highly complex technology sectors are the "answer" to development problems in Latin America. Ocampo (2004, p.38) concludes: "(...) if the region is seeking to achieve the rapid rates of structural change (including penetration into dynamic technology- intensive sectors) that are essential to gradually bridge the gap separating it from the industrialized world". In the taxonomy of Ferraz, Kupfer and Iootty (2004) only the sector of “innovation carrier industry” is considered as real creator of technological development opportunities. Cimoli, Dosi and Stiglitz (2009, p.556) argue that "(...) manufacturing and other increasing return activities such as knowledge-intensive services are at the core of technological learning with reduced demand for skilled labor."

In short, the negative view on natural resources advocated by these approaches has limitations and some considerations can be raised: (i) the definition of what constitutes "Natural Resources" is not clearly defined. That is, this sector is generally treated without due attention to understand the heterogeneities among the different sub- sectors, and these generalizations can lead to the wrong conclusions; (ii) Studies tend to generalize sectors, countries and even Latin America as a whole, focusing on macro and/or sectorial analyzes, without entering the level of firms and their temporal evolution, neglecting activities and phenomena that escape this type of analysis methodology; (iii) This current of thinking ignores the technological advances necessary for natural resource- processing industries to be competitive in the international market, reducing their success factor only in price; (iv) The authors argue that the potential for technological advancement, industrial diversification, and economic growth is mainly (or only) achieved through the high-tech product-based firms; (v) the existence of a set of studies that consider natural resource-processing industries as having very limited opportunities to generate: (a) product and process innovations, (b) technological and organizational learning, (c) spillovers, and (d) industry diversification (GONZALEZ, 2016).

The contributions and insights of the literature dedicated to examining the importance of low- and medium-tech industries (LMTs) can serve as inspiration to understand the role of agroindustry in creating technological development opportunities, once that this type of firm and/or sector possesses some (if not most) of the characteristics of the said LMT's industries. These scholars criticize the negative generalizations about the innovative performance of LMT firms and industries and present evidence that there is considerable innovation capabilities in this particular type of organization and/or industry. That is, significant differences of intrasectoral heterogeneity are found in terms of R&D intensity. LMT's industries systematically use distributed organizational knowledge elements, where innovation is largely the result of processes of search, transformation and configuration of internal & external knowledge, components and technologies widely known and developed in others places (VON TUNZELMANN; ACHA, 2004; SMITH, 2004; HIRSCH-KREINSEN, JACOBSON; ROBERTSON, 2006; HIRSCH- KREINSEN, 2008; HIRSCH-KREINSEN, HAHN; JACOBSON, 2008; ROBERTSON; SMITH, 2008; ROBERTSON; VON TUNZELMANN, 2009).

In summary, the literature devoted to the examination of LMT's industries not debate directly with the school of thought of the economists who advocate the idea that specialization in natural resource processing industries limited opportunities for technological advancement, but other characteristics of this body of knowledge are important: (i) Emphasize the importance of sector-level and firm-level studies to understand how innovative activities are being carried out in environments where conventional indicators of innovation measurement - e.g. number of patents and R&D expenditures - are not the most indicated; (ii) They point out the technological opportunities of LMT's industries - where agroindustry can be represented - and their possible impacts on the economy. However, this literature still lacks studies in contexts other than advanced economies (GONZALEZ, 2016).

Thus, in order to provide an alternative interpretation and to add new evidence to the debate about the limited opportunities of technological and economic development in the agroindustry, an empirical examination is justified that seeks to increase the knowledge related to the flows of knowledge, technological mastery, innovative performance, organizational growth and internationalization.

Studies about technological mastery and technological knowledge flows originate in the 1960s. In the late 1960s, Latin America showed considerable growth in its industrial productivity. However, after almost three decades of a policy of industrialization by import substitution in the region, the perspective of Dependency Theory argued that this productivity growth had not developed industrial innovation in the same way (BELL, 2006). The combination of the continuous import of foreign technology and the perception that Latin American firms and industries failed to internalize the innovation process was described as the process of self-perpetuation of technological dependence. However, the Katz Program challenged this perspective end-of-history (BELL, 2006)

Subsequently, studies about technological mastery and technological knowledge flows were influenced by two phenomena: in Asia, by the emergence of the "Asian tigers"; and in Latin America, by replacing the policy of industrialization by imports due to the liberalization of the economy. From the 1990s and 2000s, there was a profusion of studies that sought to focus on the organizational and managerial dimensions of technological mastery and how this mastery was built from knowledge flows. These studies were influenced by the contributions of the literature on strategic management, innovation and competitiveness in firms from advanced economies (GONZALEZ, 2016) (e.g. BELL et. al., 1982; HOBDAY, 1995; KIM, 1997a, 1997b, 1998; ARIFFIN, 2000; DUTRENIT, 2000; LEE; LIM, 2001; LEE, LIM; SONG; 2005; MARCELLE, 2005; LIU, QIAN; CHEN, 2006; DANTAS, 2006; TSEKOURAS, 2006; FAN; 2006; YORUK, 2009).

From 2010´s, studies about knowledge flow and technological domain advanced the understanding on the dynamics of technological accumulation patterns and the form that firms used the knowledge flows for this development (CUSMANO; MORRISON; RABELLOTTI, 2010; GUO; GUO, 2011; WHANG; HOBDAY, 2011; GUENNIF; RAMANI, 2012; XIAO, TYLECOTE; LIU, 2013; CHONG, HWANG; SONG, 2014; HANSEN; OCKWELL, 2014). However, literature is limited in understanding how firms explore new directions of innovation (technological trajectory) (BELL, 2010), especially in industries related to natural resources (MARIN; STUBRIN; VAN ZWANENBERG, 2014).

In addition, there have been significant advances in the understanding of the organization of the technological domain. The literature shows that the process of innovation in firms has been substantially disintegrated - or decomposed as advocated by Schmitz and Stramback (2009). The domain of technological capabilities is a process in which the organizations do not act in isolation and are interdependent of other institutions (GONZALEZ, 2016). Several studies demonstrate the importance of local suppliers and the diffusion of knowledge flows with companies for the innovation process (ARIFFIN, 2000; TSEKOURAS, 2006). Some studies have sought to understand how technological mastery could be distributed in knowledge networks with suppliers, universities and research institutes (DANTAS, 2006; GIULIANI; ARZA, 2009; ZENG, CHIE; TAM, 2010; DANTAS; BELL, 2011; YORUK, 2009; URZUA, 2011; BESSANT, ALEXANDER, TSEKOURAS, RUSH; LAMMING 2012; GUO; CHEN, 2013; CHOUNG, HWANG; SONG, 2014; ANDERSEN, 2015).

Finally, Mazzoleni and Nelson (2007) argue that technological development in agriculture, for example, “could not simply copy technologies and practice in countries at frontier, but needed to develop technologies suited to their own conditions.” (p. 1516). In other words, as this area of knowledge is subject to specific conditions (e.g. soil, climate, water, diseases, pests, etc.), it is necessary (if not mandatory) for countries and/or industries to create their own technological trajectory; while the technological development in the manufacturing industry can be adapted with modest modifications and with not so expressive costs (MAZZOLENI; NELSON, 2007).

In summary, the relevant literature on these themes concluded that: (i) there was a significant advance in the understanding of how traditional (manufacturing) industries in emerging economies (especially in the Asian context) carried out the process of technological domination in technological trajectories already mapped and traveled by world leaders (KIM, 1997a, 1997b; GUENNIF; RAMANI, 2012; XIAO; TYLECOTE; LIU, 2013); (ii) However, to what extent is it possible to apply this approach in resource-intensive industries, since the edaphoclimatic conditions of this particular type of industry preclude a mere technological replication? (BELL, 2010), especially in natural resource- related industries (MARIN; STUBRIN; VAN ZWANENBERG, 2014; GONZALEZ, 2016);(ii) Much of the literature about technological mastery examines individual firms and does not investigate the role of related organizations (universities, research institutes, suppliers, competitors, etc.); (iv) The understanding of the relative importance of knowledge flows in terms of which partners (universities, research institutes, suppliers, competing companies, customers, etc.) were accessed for technological mastery building (GONZALEZ, 2016), and; (v) there is little evidence about how the relative importance of different types of collaboration and how different types of partners change over time while the company builds technological dominance (GONZALEZ, 2016).

In this research, learning is understood along the lines of Bell et al. (1982) and Bell (1984), who point out that learning consists of conscious, intentional, costly, non-automatic, active and deliberate processes in which skills and technical knowledge are acquired by individuals and by the organization. Malerba (1992) argues that learning is cumulative and increases the stock of knowledge or technological capabilities of the company. Technological learning can happen internally, by creating knowledge by the company itself, or externally, by searching for sources located outside the organization, which can be in and/or outside of its market and country. This external relationship (technological knowledge flows) can happen with several actors such as Universities, Research Institutes, Suppliers, Competitors, Users, Partner Companies and etc. In this way, the importance of integrating external and internal learning is imperative (Kim 1997a). Bell & Pavitt (1993, p. 163) comment that: "Technological accumulation (or technological learning) refers to any process by which the resources for generating and managing technical change are increased or strengthened" - therefore, if an organization aims to deepen its technological domain rapidly and overcome technological discontinuities effectively, it is necessary that learning efforts (or knowledge flows) are carried out in an intense way.

The technological knowledge flows will be examined in this research by the different ways of acquiring and assimilating knowledge. Bell (1984) developed a typology of learning activities divided into: Learning-by-doing, operating, changing, searching, hiring, training and system performance feedback. This typology makes an important distinction between active and passive modes of learning. However, the literature of the subject advances in the proposition of other mechanisms of learning (For a review of the main learning mechanisms, consult Queiroz (2006). Malerba (1992) and other authors argue about five other types of activity: Learning by searching, using, interacting, from inter-industry spillovers e from advances in science and technology. However, other authors examine other learning mechanisms, for example: (i) linkages (LUNDVALL, 1988); (ii) from users (VON HIPPEL, 1988); (iii by competitors (WHIPP, ROSENFELD; PETTIGREW, 1989). (iv) R&D (COHEN; LEVINTHAL, 1989); (v) joint ventures & strategic alliances (DODGSON, 1993); (vi) before doing (PISANO, 1996); (vii) sharing (NELSON; WINTER, 1982; MARCELLE, 2004); (viii) field experimentation (MARCELLE, 2004) e; (ix) large-scale project management (MARCELLE, 2004). These internal and external knowledge flows are useful for understanding the possible sources that organizations, especially those within emerging market contexts, can use to carry out innovative activities.

Nevertheless, the literature on technological knowledge flows in emerging economies is not enough to elucidate some questions about innovation in Brazilian agroindustry. In this way, the literature on learning organizations, organizational learning (OL) and strategic management presents a series of merits that can help innovation scholars of firms in emerging economies understand how knowledge flows occur in this particular type of organization. This literature becomes useful when organizations of emerging economies already master innovative technologies and activities of high complexity. The focus of this literature is to understand how organizations exploit, increase and renew their innovative technological capabilities and advance the international technological frontier.

Finally, this work starts from the premise that the technological knowledge flows play the role of a variable with greater proximity and with a greater degree of influence in the technological mastery (LALL, 1992; BELL; PAVITT, 1993; BESSANT; ALEXANDER; TSEKOURAS; RUSH; LAMMING 2012). Malerba (1992) comments that, since there are many mechanisms (flows) of knowledge, different types of learning will affect the stock of knowledge (and therefore the technological mastery) of firms differently. Therefore, Table 1 presents an operational framework to investigate technological knowledge flows.

Table 1
Types of technological knowledge flows.

Based on Bell et al (1982); Malerba (1992); Kim (1997a, 1998); Guo and Guo (2011); Gonzalez (2016)

Table 2
Tipology of technological domain for Brazilian agroindustry.

Based on Bell et al (1982 e 1995), Lall (1992), OECD (1992), Arnold & Thuriaux (1997), Bell (2006), Gonzalez (2016)

Table 2 (cont.)
Tipology of technological domain for Brazilian agroindustry.

Based on Bell et al (1982 e 1995), Lall (1992), OECD (1992), Arnold & Thuriaux (1997), Bell (2006), Gonzalez (2016)