Scholarly article on topic 'Determinants of reverse knowledge transfer for emerging market multinationals: the role of complexity, autonomy and embeddedness'

Determinants of reverse knowledge transfer for emerging market multinationals: the role of complexity, autonomy and embeddedness Academic research paper on "Economics and business"

Share paper
Academic journal
Revista de Administração
OECD Field of science

Academic research paper on topic "Determinants of reverse knowledge transfer for emerging market multinationals: the role of complexity, autonomy and embeddedness"

Available online at

Revista de Administrado raocsp

Management Journal

Revista de Administrado xxx (2016) xxx-xxx

Technology management

Determinants of reverse knowledge transfer for emerging market multinationals: the role of complexity, autonomy and embeddedness

Determinantes da transferencia reversa de conhecimento em multinacionais de mercados emergentes: o papel da complexidade, da autonomia e da integracao

Factores determinantes de la transferencia inversa de conocimiento en multinacionales de mercados emergentes: el papel de la complejidad, la autonomía y la integración

Franciane Freitas Silveira a*, Roberto Sbragiab, Henry Lopez-Vegac, Fredrik Telld

a Universidade Federal do ABC, Sao Bernardo do Campo, SP, Brazil b Universidade de Sao Paulo, Sao Paulo, SP, Brazil c Jonkoping University, Jonkoping, Sweden d Uppsala University, Uppsala, Sweden

Received 29 June 2016; accepted 19 August 2016 Scientific Editor: Maria Sylvia Macchione Saes


Subsidiaries conduct innovation activities in foreign markets either to capture valuable knowledge that is necessary to adapt their products to local markets or to create valuable knowledge for headquarters. For emerging market multinationals, most studies have overlooked the determinants of successful reverse knowledge transfer from subsidiaries located in emerging and developed markets. This paper analyzed the responses of a survey administered to 78 Brazilian multinationals that own subsidiaries in developed and emerging markets. We found that knowledge complexity developed at the subsidiary, its autonomy and embeddedness in the foreign market determine the successful reverse knowledge transfer to headquarters of emerging market multinationals. This paper contributes to previous studies of reverse knowledge transfer by underlying the main drivers for emerging market multinationals.

© 2016 Departamento de Administracao, Faculdade de Economia, Administracao e Contabilidade da Universidade de Sao Paulo - FEA/USP. Published by Elsevier Editora Ltda. This is an open access article under the CC BY license (

Keywords: Reverse knowledge transfer; Emerging multinationals; Brazilian multinationals


Subsidiarias realizam atividades de inovacao em mercados estrangeiros, quer para capturar o conhecimento valioso que e necessario para adaptar seus produtos aos mercados locais ou para criar conhecimento de alto valor para a sede. No contexto de multinacionais de mercados emergentes, a maioria dos estudos tem negligenciado os determinantes da transferencia de conhecimetno provenientes de subsidiarias (transferencia reversa). Foram analisadas as respostas de uma pesquisa realizada com 78 multinacionais brasileiras que possuem subsidiarias em mercados desenvolvidos

* Corresponding author at: Alameda da Universidade, s/n° - CEP 09606-045, Sao Bernardo do Campo, SP, Brazil. E-mail: (F.F. Silveira).

Peer Review under the responsibility of Departamento de Administracao, Faculdade de Economia, Administracao e Contabilidade da Universidade de Sao Paulo - FEA/USP.

0080-2107/© 2016 Departamento de Administracao, Faculdade de Economia, Administracao e Contabilidade da Universidade de Sao Paulo - FEA/USP. Published by Elsevier Editora Ltda. This is an open access article under the CC BY license (

2 F.F. Silveira et al. /Revista de Administrando xxx (2016) xxx—xxx

e emergentes. Verificou-se que a complexidade do conhecimento desenvolvido na subsidiária, bem como a sua autonomía e inserjao no mercado externo determinam o fluxo de transferencia reversa de conhecimento na empresa multinacional emergente. Este trabalho enriquece estudos anteriores sobre transferencia reversa de conhecimento destacando os principais drivers para as multinacionais dos mercados emergentes. © 2016 Departamento de Administracao, Faculdade de Economia, Administracao e Contabilidade da Universidade de Sao Paulo - FEA/USP. Publicado por Elsevier Editora Ltda. Este e um artigo Open Access sob uma licenca CC BY (

Palavras-chave: Trasnferencia reversa de conhecimento; Multinacionais emergentes; Multinacionais brasileiras


Filiales realizan actividades de innovación en los mercados extranjeros, ya sea para capturar el conocimiento valioso que es necesario para adaptar sus productos a los mercados locales, o con el fin de crear conocimiento de alto valor para su sede. Respecto a las multinacionales de mercados emergentes, en la mayor parte de los estudios no se ha dado la debida atención a los factores determinantes de la transferencia de conocimiento a partir de filiales (transferencia inversa). En este estudio se analizan las respuestas de una encuesta realizada a 78 multinacionales brasileñas que poseen filiales en mercados desarrollados y emergentes. Los resultados indican que la complejidad del conocimiento desarrollado en la filial, así como su autonomía e inserción en el mercado externo determinan el flujo de transferencia inversa de conocimiento en la empresa multinacional emergente. Con este trabajo, se colabora al desarrollo de los estudios acerca de la transferencia inversa de conocimiento, con énfasis en los principales drivers para las multinacionales de mercados emergentes.

© 2016 Departamento de Administracao, Faculdade de Economia, Administracao e Contabilidade da Universidade de Sao Paulo - FEA/USP. Publicado por Elsevier Editora Ltda. Este es un artículo Open Access bajo la licencia CC BY (

Palabras clave: Transferencia inversa de conocimiento; Multinacionales emergentes; Multinacionales brasilenas

28 Introduction

29 The multinational enterprise (MNE) is a differentiated net-

30 work in which its controlled subsidiaries vary widely in terms

31 of duties and responsibilities (Nohria & Ghoshal, 1994). For

32 example, while some subsidiaries evolve through the headquar-

33 ters' mandates others focus on their own initiatives (Mudambi,

34 Piscitello, & Rabbiosi, 2014). Since the late 1990s, the recog-

35 nition that headquarters operate as knowledge receivers from

36 their internationally dispersed subsidiaries has gained signifi-

37 cance in international business research (Ambos, 2015). The

38 strategic importance of the MNE' subsidiaries has continued to

39 grow, in that it is an access pathway to knowledge and to the

40 technology situated at the subsidiaries' local markets (Borini,

41 Oliveira, Silveira, & Concer, 2012; Criscuolo & Narula, 2007;

42 Frost & Zhou, 2005), which can actively contribute to value

43 creation and subsequent gain of competitive advantage for the

44 entire MNE (Bartlett & Ghoshal, 1989; Cantwell & Mudambi,

45 2005; Yang, Mudambi, & Meyer, 2008).

46 An underlying idea is that MNE make use of knowledge

47 generated by foreign subsidiaries. From this perspective, sub-

48 sidiaries upgrade their competence enhancing role such as

49 market expansion, cost reduction and supplier adaptation and

50 begin to play a more active role through knowledge develop-

51 ment. For example, foreign subsidiaries might develop new

52 products, new technologies, create new practices, new skills

53 that will later shape their own competence creating pathways as

54 well as accumulate different degrees of technological capability

55 (Birkinshaw, 1997; Borini et al., 2012; Borini, Costa, Bezerra, &

56 Oliveira, 2014; Cantwell&Mudambi, 2005;Figueiredo &Brito,

57 2011; Frost, Birkinshaw, & Ensign, 2002; Ghoshal & Bartlett,

58 1988; Govindarajan & Trimble, 2012; Mudambi, Mudambi,

59 & Navarra, 2007; Nohria & Ghoshal, 1997). Moreover, com-

60 petence creating subsidiaries could enhance their innovation

outcomes which enables them to compete domestically and internationally (Bell & Pavitt, 1995; Cantwell & Mudambi, 2005; Figueiredo & Brito, 2011). From a subsidiary perspective, reverse knowledge transfer (RKT) gives visibility to subsidiaries that could leverage their strategic position in the multinational network (Borini et al., 2012; Holm & Pedersen, 2000).

These factors have highlighted that reverse knowledge transfer is a key variable in the study of cross-border knowledge flows in MNEs (Ambos, 2015). As a result, the knowledge transfer in the reverse direction, that is, fromsubsidiaries toMNEheadquar-ters, has emerged as a prominent theme in international business studies (Ambos, 2015; Ambos, Ambos, & Schlegelmilch, 2006; Criscuolo, 2005; Frost & Zou, 2005; Gupta & Govindarajan, 2000; Hakanson & Nobel, 2001; Rabiosi, 2008; Rabiosi & Santangelo, 2011; Rabiosi, 2011; Yang et al., 2008). While a number of articles explore the antecedents, success amount and success key-factors in different functional conFigurations at the multinational corporation (Ambos, 2015), additional research is needed (Michailova & Mustaffa, 2012). First, since the transfer of knowledge in MNEs has grown considerably in recent years, becoming therefore more prone to various definitions and measurements of the same constructs resulting in conclusions, often contradictory and ambiguous. Second, while recognizing the importance of investigating the relationship of the subsidiary with external companies located in the host countries, the literature often focuses only on the research of knowledge flows within the MNE. This narrow attention considers subsidiaries are primarily recipients of knowledge (Michailova & Mustaffa, 2012).

From emerging multinationals enterprises (EMNEs)'s viewpoint, the ability to transfer knowledge in reverse direction seems to be even more crucial. For example, authors say that the EMNEs strategic models are guided by the pursuit of foreign capabilities, such as technological knowledge, which can

80 8i 82

F.F. Silveira et al. /Revista de Administrafao xxx (2016) xxx-xxx 3

95 be combined with the existing resources (Bartlett & Ghoshal, on organizational performance and innovation capacity (Lyles 150

96 2000). That is so because, instead of internationalizing to utilize &Salk, 1996; Powell, Koput, & Smith-Doerr, 1996; Tsai, 2001; 151

97 existing advantages, emerging market multinationals will inter- Van Wijk et al., 2008). The underlying idea is that the transferred 152

98 nationalize aiming at acquiring new advantages and capabilities knowledge contributes to the development of organizational 153

99 (Guillen & Garcia-Canal, 2009; Mathews, 2006; Ramamurti & capabilities that are difficult to imitate and can later lead to 154

100 Singh, 2009) and should to do it quicker than traditional multi- better performance (Szulanski, 1996). Knowledge transfer sti- 155

101 nationals did in their expansion paths (Mathews, 2006). mulates the combination of the existing knowledge with the 156

102 In the context of Brazilian multinationals, recent studies have newly acquired one and increases the capability of a unit for 157

103 sought to understand the primary factors that influence the RKT. carrying out new combinations (Jansen, Van Den Bisch, & 158

104 The study of Borini et al. (2012) argues that the reverse knowl- Volberda, 2005). 159

105 edge transfer is a function of the strategic guidance of the: However, transferring knowledge between units of a same 160

106 (1) subsidiaries' R&D laboratories, (2) integration (communi- organization is not easier than conducting external knowledge 161

107 cation) between headquarters and subsidiaries, (3) subsidiary transfers (Kogut & Zander, 1992). This is particularly the case 162

108 entrepreneurial orientation, (4) subsidiary lifetime and (5) entry when it comes to RKT. This process can be even more challeng- 163

109 via greenfield investments. Moreover, the study of Bezerra and ing, since while "[...] the conventional transfer is a process of 164

110 Borini (2015) tests the impact that a nation development exerts teaching, the reverse transfer is a process of persuading (Yang 165

111 on the reverse innovation transfer in products and processes. In et al., 2008)". In this case, the effort is much higher because 166

112 this study, they sought to understand which determinants of RKT its effectiveness depends on convincing headquarters. There- 167

113 are present in Brazilian multinationals. In ourstudy, we showthat fore, the transfer depends on headquarter's assessment that the 168

114 RKT is related to the degree of: (1) knowledge complexity that is features and relevance of the subsidiary's knowledge is crucial 169

115 being transferred, (2) subsidiary autonomy and (3) external so that the reverse transfer does occur. The RKT is defined as 170

116 embeddednes. As highlighted by numerous authors (Mimbaeva "an intra-organizational exchange of information, technology 171 11Q3 et al., 2007; Van Wijk, Jansen, & Lyles, 2008), such aspects are or know-how from international subsidiaries (located in host 172

118 identified as key ones for understanding the RKT phenomenon. countries) to corporate headquarters (home countries). The term 173

119 Although there are many kinds of knowledge to be transferred 'reverse' is used to distinguish these transfers from the more 174

120 through conventional and/or reverse direction, this study focuses conventional form of 'forward' transfers - from headquarters 175

121 specifically on the technological type of knowledge (of product to subsidiaries -, and 'lateral' transfers between subsidiaries" Q6 176

122 and process). Our findings are based in an analysis of the survey (Ambos, 2015). 177

123 responses administered to78 Brazilian multinationals that own Some studies have highlighted that subsidiaries create com- 178

124 subsidiaries in developed and emerging markets. petitive advantages for MNEs when valuable knowledge is 179

125 As a contribution, it is expected that our study adds knowl- transferred to the headquarters (e.g. Gupta & Govindarajan, 180

126 edge to the international businesses theory, since the knowledge 2000; Hakanson & Nobel, 2001; Rabbiosi, 2011; Yang et al., 181

127 transfer has been treated as a key factor of competitive advantage 2008). For MNEs, some of the determinants of RKT include 182

128 of MNEs (Borini etal., 2012,2014; Govindarajan & Ramamurti, the: (1) knowledge features being transferred (Mimbaeva et al., 183

129 2011) and, specifically, of the emerging multinationals compa- 2007), (2) organizational characteristics (size, age, autonomy) 184

130 nies (Cuervo-Cazurra, 2012; Immelt, Govindarajan, & Trimble, (Frost et al., 2002; Gupta & Govindarajan, 2000), (3) role of 185

131 2009; Ramamurti, 2008). Since most research that explains this organizational mechanisms (Hakanson & Nobel, 2001; Rab- 186

132 phenomenon is based on MNEs with subsidiaries and headquar- biosi, 2011), (4) the subsidiaries' roles (Ambos et al., 2006; 187

133 ters in developed countries, less affected by institutional distance Rabbiosi, 2011; Yang et al., 2008), (5) the host country eco- 188 13Q4 (Rabbiosi, 2011; Yang et al., 2008), one can not assume that the nomic development (Cantwell & Mudambi, 2005; Frost et al., 189

135 factors that influence RKT from for MNEs are the same as those 2002; Gupta & Govindarajan, 2000), (6) the absorptive capacity 190

136 for EMNEs (Borini et al., 2016). A practitioner contribution of (Ambos et al., 2006), (7) the knowledge relevance (Yang et al., 191

137 this study seeks to inform EMNE managers about the strategic 2008), (8) the internal embeddedness (subsidiary/headquarters) 192

138 drivers of RKT. and (9) the external embeddedness (subsidiary/partners) 193 13Q5 This paper is structured as follows: the next section, "Con- (Figueiredo, 2011; Meyer, Mudambi, & Narula, 2011). 194

140 ceptual framework" section presents the proposed determinants Following, it is explained how knowledge characteristics, 195

141 of reverse technology transfer. "Methodology" section outlines such as complexity, autonomy, and external embeddedness influ- 196

142 our research strategy and field procedures. "Findings" sec- ence RKT from subsidiaries to headquearters of EMNEs. 197

143 tion presents our results and discusses the implications of our

144 findings for firms in emerging markets. Finally, "Conclusion" Subsidiary's autonomy 198

145 section presents our main conclusions, some limitations of the

146 study, and avenues for further research. Subsidiary's autonomy could be defined as the extent to 199

which a subsidiary is allowed to make decisions on its key 200

147 Conceptual framework strategic issues (Mudambi & Navarra, 2004), without a head- 201

quarters direct intervention (Roth & Morrison, 1992). A higher 202

148 The literature argues that knowledge transfer, whether aris- level of autonomy is often related to knowledge creation and 203

149 ing from internal or external sources, has an important impact development at the MNE (Ghoshal & Nohria, 1989; Gupta 204

4 F.F. Silveira et al. /Revista de Administrafao xxx (2016) xxx—xxx

205 & Govindarajan, 1991; Nohria & Ghoshal, 1994), since inde- 1990). The greater the number of techniques, organizational 259

206 pendent subsidiaries, (1) have strategic mandates (Birkinshaw, routines, people and resources involved that are connected to a 260

207 Hood, & Jonsson, 1998), (2) make quick decisions (Cantwell particular knowledge, the more complex it becomes. These con- 261

208 & Piscitello, 1999), (3) recognize and take advantage of local ditions moderate the information amount that must be processed 262

209 opportunities (Frost et al., 2002), (4) develop new knowledge as for the understanding of components involved (Simonin, 1999). 263

210 oflocal knowledge bases (Andersson,Forsgren, & Holm, 2002), Thus, management scholars tend to agree on the idea that 264

211 (5) generate intrinsic motivation on individuals (Mudambi et al., complexity hinders knowledge transfer since it decreases the 265

212 2007), (6) have initiative and willingness to share the knowledge receiver's ability to identify, understand and integrate the knowl- 266

213 acquired (Gupta & Govindarajan, 2000; Tsai, 2002). On the con- edge to be acquired (Simonin, 1999). Yet, opposite results have 267

214 trary, a low level of autonomy, may limit the subsidiary freedom, been found in the literature (Mimbaeva et al., 2007). Since, 268

215 hindering its knowledge creation and development capability complex knowledge is the most valuable to the company's com- 269

216 (Ghoshal & Bartlett, 1988). Foss and Pedersen (2002) also petitiveness. Studies have shown, for example, that global teams 270

217 explain that high levels of subsidiary autonomy-associated with are able to share complex knowledge through rules and codes 271

218 loss of control - could be overcomed by the increase in knowl- common to the exchanging area (Reddy, 2008). Q7 272

219 edge exchange amongst subsidiaries. While opposite results In the EMNE perspective, additional efforts to share this kind 273

220 have also been reported (Frost etal., 2002; Gammelgaard, Holm, of knowledge can be advantageous since, as its imitation and 274

221 & Pedersen, 2004), most researches have suggested mainly a substitution is hampered, it may be useful to the building of 275

222 positive relationship between knowledge decentralization and strategic capabilities (Nair, Demirbag, & Mellahi, 2015) due 276

223 transfer (Cantwell & Mudambi, 2005; Foss & Pedersen, 2002; to the prevailing need to use the available foreign resources 277

224 Van Wijk et al., 2008). (Mathews, 2006). A study conducted in Indian multination- 278

225 Recently, Rabiosi (2008) argued that RKT is coupled with als, for example, found that RKT happens regardless of the 279

226 subsidiary autonomy, i.e. mechanisms of personal communi- knowledge complexity. In Brazilian MNEs, it is suspected that 280

227 cation between subsidiary and headquarters. Yet, regarding only less complex knowledge from subsidiaries is transferred in 281

228 subsidiaries of EMNEs, it is argued that, due to their recent reverse direction, considering that the foreign subsidiaries role 282

229 progress in the international market and, therefore, due to their is determined by the Brazilian headquarters (Galina & Moura, 283

230 early age, they are strongly dependent on headquarters' deci- 2013) which still holds greater centralization in the decisions 284

231 sion making power (Dunning, 1993). This might not be different and innovations. Accordingly it was formulated the following 285

232 in Brazilian MNEs, that tend to be more centralizing, limiting hypothesis: 286

233 therefore their subsidiaries' knowledge creation possibilities

234 (Chu & Wood, 2008). This is an unfavorable situation for the H2. The lower the complexity of the subsidiary's R&D knowl- 287

235 development of existing and new knowledge at the headquarter. edge, the greater the degree of reverse knowledge transfer. 288

236 However, in the same way as traditional MNEs, the international-

237 ization process of EMNEs requires the capability to acquire and

238 develop knowledge (Mathews, 2006). Hence, subsidiaries play

239 a central role in the pursuit of new knowledge (Borini & Fleury, Local embeddedness 289

240 2011). Different authors state that EMNEs survival depends

241 even more heavily on resources that have been developed abroad Embeddednes is related to the notion that MNE's competitive 290

242 when compared to the multinationals from developed countries performance can be facilitated through the social relationships 291

243 (Guillen & Garcia-Canal, 2009; Mathews, 2006). Therefore, this they create with several business players such as customers, 292

244 study advocates that subsidiary autonomy is critical for RKT in universities and local research institutions (Grabher, 1993; 293

245 EMNEs, which allows us to hypothesize that: Granovetter, 1985; Uzzi, 1996). More specifically, embed- 294

dedness refers to the mutual adaptation of activities between 295

246 m. The greater the subsidiary autonomy, the greater the two companies as much as a common understanding of the 296

247 reverse knowledge transfer. collective targets and appropriate ways to work in a social 297

system (Tsai & Ghoshal, 1998). Therefore, it is considered 298

248 Knowledge complexity as a strategic resource for MNEs. It provides easy access to 299

the resources and capabilities that are outside the company 300

249 The increasing specialization and sophistication in R&D (Andersson et al., 2002; Uzzi & Gillespie, 2002) that are able 301

250 requires companies to integrate distinct knowledge areas to to generate a large knowledge transfer among the partners 302

251 develop new products. As a result knwoledge turns to be highly (Figueiredo, 2011; Uzzi & Gillespie, 2002). 303

252 complex and difficult to conduct intra-knowledge transfers. A The degree of embeddedness by foreign subsidiaries, mea- 304

253 paradox emerges: the greater the number of functional areas sured by the proximity to local partners, reflects subsidiary's 305

254 and scientific disciplines necessary to develop new products, ability to absorb knowledge from its local network, which some- 306

255 the more complex it is to transfer the knowledge (Ciabuschi & times might result in new knowledge creation (Anderson et al., Q8 307

256 Martin, 2012). Knowledge complexity is associated to the ampli- 2002). This scenario tends to directly foster subsidiary's innova- 308

257 tude which is the extent of specialization fields (Grant, 1996) and tive capacity, i.e. improvement of existing products and services 309

258 the ambiguity of the referenced knowledge (Reed & DeFillippi, or new product, service, technology development (Andersson,

F.F. Silveira et al. /Revista de Administrafao xxx (2016) xxx—xxx 5

310 Bjorkman,&Forsgren,2005;Cantwell&Mudambi,2005;Frost population, 39 Brazilian multinationals participated in the sur- 363

311 et al., 2002; Hakanson & Nobel, 2001; Yamin & Otto, 2004). vey (61.9%), with 78 responses, corresponding to 32.5% of 364

312 Indirectly, local subsidiary embeddedness can foster knowledge all subsidiaries. This means that in some cases responses were 365

313 transfer to other MNE's units (Powell et al., 1996; Yamin & Otto, received from more than one subsidiary per headquarter. 366

314 2004), constructing, in turn, the subsidiary's power relationships In the attempt to identify possible shortcomings or misunder- 367

315 within the MNE (Andersson, Forsgren, & Holm, 2007). standings in the survey, a pre-test was conducted together with 368

316 Higher levels of subsidiary embeddedness are related to specialists from academia and industry (Cooper & Schindler, 369

317 an understanding of the context in which the local knowl- 2003) which help to generate new insights and adjustments in 370

318 edge resides. Frequently, subsidiaries interact with its closest the questionnaire. Following, the electronic survey was sent to 371

319 network of local companies and institutions in order to learn participants, with a follow-up phone-call to clarify any questions 372

320 about customers and technologies and, therefore 'capture' the from respondents. The total period of data collection was five 373

321 local knowledge (Figueiredo, 2011). Subsequently, it must use months, from October 2013 up to February 2014. Responses 374

322 the connectivity already established within the MNE network were collected from R&D offices and the respondents ranged 375

323 for transferring the knowledge in reverse direction (Meyer from subsidiary director, international business and R&D direc- 376

324 et al., 2011; Najafi-Tavani, Giroud, & Andersson, 2013). tor, and engineering managers. 377

325 Regarding MNEs of emerging markets, Child and Rodriguez

326 (2005), Mathews (2006) and Luo and Tung (2007) emphasize MeasUres ™

327 the importance of relationships and knowledge opportunities

328 available at subsidiaries hosting markets. For example, provid- Dependent variable 379

329 ing easy access to technol°gies found in devel°ped markets The dependent variable (reverse knowledge transfer - RKT) I

330 (Figueiredo, 2005). Ramamurti and Singh (2009, pp. 126-127) represents, over the last three years, the rate of RKT of tech- 381

331 show that EMNEs can pursue several different strategies, such nology and market knowledge that the subsidiary transferred 382

332 as "low-cost partners", "global consolidators" and "global first back to the headquarters. In order to detail the types of tech- 383

333 movers".Based on these arguments,the following hypothesis is nological content, it was applied the Iammarino, Padilla-Perez, 384

334 suggested: and Von Tunzelmann (2008) scale, which was validated previ- 385

335 H3. The greater the embeddedness of a foreign subsidiary, the ously by other authors (LaH, 1992; BeU & Pavitt, 1995; Ariffin 386

336 higher the reverse knowledge transfer and Figueiredo (2003)), who rank the technological knowledge Q9 387

transfer in terms of product and process. On a five-point scale 388

(ranging from 1 "not at all" to 5 "to a very great extent"). For 389

337 Methodology ensuring the robustness results, it was also inserted a dummy 390

variable which allowed the respondent to indicate the cases in 391

338 Satnple and data cMectim which the subsidiary had never done or had done the reverse 392

transfer of a specific product or process knowledge (0 or 1). 393

339 The sample of this study consists of Brazilian MNEs with

340 manufacturing, sales or R&D subsidiaries abroad. We expected

341 that subsidiaries with more strategic activities would have more Independent variables 394

342 opportunities of transferring knowledge to the headquarters in The knowledge complexity construct measures the num- 395

343 reverse direction. The data was collected using an eletronic sur- ber of interdependent technologies, routines, individuals, and 396

344 vey with Brazilian MNEs subsidiaries established abroad (see resources linked to a particular knowledge or asset (Simonin, 397

345 Appendix 1). Due to the non-existence of an official number of l999);Moreover,the complexity construct was measured using 398

346 Brazilian multinationals owning subsidiaries with either man- a six-item Likert scale based on responses (l-strongly dis- 399

347 ufacturing or R&D centers installed abroad. The first step was agree; 5 - strongly agree) (adapted from Simonin,2004; Zander 400

348 to identify Brazilian multinationals presenting these character- & Kogut,1995).The subsidia7 autonomy measure indicates the 401

349 istics from secondary data sources, such as GINEBRA Project extent to which a subsidiaryis allowed to make decisions about 402

350 (Management System for the Internationalization of Brazilian lts key strategic issues (Rabbiosi, 2011). The measure of sub- 403

351 Enterprises) that resulted in the publication 'Business Man- sidiary autonomy was based on a scale originally developed by -

352 agement for the Internationalization of Brazilian Companies' Ghoshal and Nohria (1989) and later used by Birkinshaw et al. 405

353 (coordinatedbyFleury, 2010), an annual survey oftheFundacao (1998) and Rabbiosi (2011).A five-itemLikert scale assessed it. 406

354 Dom Cabral (Dom Cabral Foundation), Valor Economico (Eco- The subsidiary embeddedness indicates the collaboration degree 407

355 nomic Value), and SOBEET (Brazilian Society of Transnational with the local networks.In partfculrc,this study focuses on the 408

356 Corporations) surveys as well as data from the Brazilian Multi- subsidiary embeddedness with local customers and suppliers. 409

357 nationals Observatory (Center of Brazilian Multinationals) of This construct was developed based on Andersson et al.(2002, 410

358 the ESPM (School of Higher Education in Advertising and Mar- 2005). A five-item Likert scale assessed it. 411

359 keting).

360 In this secondary sources, 63 multinational companies were Controls variables 412

361 listed, being possible to identify 240 subsidiaries with for- The MNE literature suggests several factors that might be 413

362 eign manufacturing operations and/or R&D centers. Of this correlated to RKT. In particular, it is expected that subsidiaries 414

F.F. Silveira et al. /Revista de Administraçao xxx (2016) xxx—xxx

415 located in developed countries and more ancient subsidiaries are

416 more likely to transfer reverse knowledge.

417 Subsidiary location. The host country has been related to fac-

418 tors that impact the subsidiary development and positioning

419 (Birkinshaw & Hood, 1998; Gupta & Govindarajan, 2000; 42Q10 Mudambi & Cantwell, 2005; Rabiosi, 2011) as well as the nature

421 of RKT (Yang et al., 2008). Particularly, this happens because the

422 subsidiary's capabilities and skills could reflect the country tech-

423 nological and institutional forces, such as legal and institutional

424 factors (for example patent protection and industrial incentives)

425 that ensure the proliferation of innovation. The assumption,

426 therefore, is that companies in emerging markets get involved in

427 less innovation than companies in developed markets, due to the

428 lack of high technology in emerging markets (Vernon-Wortzel &

429 Wortzel, 1998). Thus, the higher the economic development of

430 the subsidiary's host country, the greater the benefits earned by

431 the headquarters arising from the transferred knowledge (Frost

432 et al., 2002).

433 For emerged market MNEs, subsidiaries located in devel-

434 oped or high-income countries can impact the rate and speed

435 of RKT, since the resources available in these markets can help

436 increase the headquarters breadth and novelty (Mathews, 2006). 43Q11 On the contrary, Aulakh (2007) and Cuervo-Cazurra and Genc

438 (2008) argue that emerging market MNEs have the same knowl-

439 edge resources than those operating in developed countries. In 44Q12 the specific case of Brazilian companies, Bezerra et al. (2015)

441 have concluded that MNEs' Brazilian subsidiaries located in

442 developed countries transfer more knowledge in reverse direc-

443 tion than subsidiaries located in emerging countries. In order

444 to capture the subsidiary location effects on the levels of RKT,

445 the dummy variable low-income countries (0) and high-income

446 countries (1) were added to the model.

447 Subsidiary's age. More ancient subsidiaries could have some

448 advantages over newer ones due to (1) the increased informa-

449 tion and resources, (2) the higher development of R&D skills,

450 (3) acquired experience and expertise, and (4) increased learning

451 curve effects. Therefore they might be less dependent on knowl-

452 edge from headquarters (Foss & Pedersen, 2002; Yamin and

453 Otto, 2004). Previous studies show both positive and negative

454 effects of organizations's age regarding the learning and innova-

455 tion outcomes (S0rensen & Stuart, 2000). While positive effects

456 are justified by the knowledge increase, accumulated experi-

457 ence and possession of stronger relationships with suppliers,

458 and customers that enable the innovation process improvement

459 (Cohen & Levinthal, 1990). Negative effects are associated with

460 upgrade difficulties of more mature companies with external

461 technological advances, at the risk of becoming inert and limited

462 for learning and adapting to new circumstances. In other words,

463 there is a loss of innovative capacity (S0rensen & Stuart, 2000;

464 Tushman & Anderson, 1986).

465 In this regard, in the Brazilian multinationals context, Bezerra

466 et al. (2015) found that the younger a subsidiary, the greater

467 its extent of RKT. Thus, despite the inconclusive findings of

468 subsidiaries' age, it is expected that older subsidiaries are more

469 likely to develop and transfer back knowledge to headquarters

than recently established subsidiaries. Particularly, due to the 470

period of existence of Brazilian subsidiaries is much lower when 471

compared to emerged market subsidiaries. In order to capture the 472

subsidiary age effects on the levels of RKT, the dummy variable 473

young (0) and old subsidiary (1) were added to the model. The 474

details of each variable, including indicators and authors, used 475

as background is presented in Appendix 1. 476

Data analysis 477

A descriptive analysis was carried out to identify the frequen- 478

cies of respondents' answers for all constructs comprised in the 479

survey. The Partial Least Square - Structural Equation Model- 480

ing (PLS-SEM) was used to assess the determinants' influence 481

of RKT (Hair, Hult, Ringle, & Sarstedt, 2014). The structural 482

model was estimated on SmartPLS 3.0 (Ringleet al., 2014) using Q13483

the 'path' weighting scheme. The decision to use this method 484

took into account a number of criteria, including (1) the fact 485

that the indicators do not have a normal distribution, which is 486

one of the assumptions for the use of the maximum likelihood 487

method (ML); (2) the use of interval scales (Joreskog & Wold, 488

1982); (3) its ability to deal with more complex models as com- 489

pared to LISREL (Henseler, Ringle, & Sinkovics, 2009); and 490

(4) the small sample size. 491

Since the PLS algorithm formulation (Hui & Wold, 1982; 492

Lohmoller, 1989) is recognized that it is biased and is only 493

"consistent at large", which means that the bias decreases as 494

the number of indicators by latent variable is increased. This 495

issue occurs because the relationships amongst latent variables 496

(correlations and path coefficients) are estimated as from the 497

factorial scores, which are obtained as a sum or a weighted aver- 498

age of their indicators, including the measurement errors. This 499

fact is treated as correlation attenuation in the methodological 500

references related to psychometrics, for example (Nunnally & 501

Bernstein, 1994, p. 212). However, despite this bias, Hair et al. 502

(2014, p. 79) mention some simulations where it is identified 503

that the bias is small for practical purposes. For four and eight 504

indicators by latent variable, Chin and Newsted (1999, p. 333) 505

found abias equal to 0.05. To minimize this bias (attenuation) the 506

latent variables were measured with five to six indicators each, 507

reaching reliability values (composite reliability and Cronbach's 508

alpha) higher than 0.8 (Table 2). 509

Additionally, to assess this bias size, the disattenuated cor- 510

relations were calculated (or "correction" for attenuation as 511

explained by Nunnally and Bernstein (1994, p. 241) of the 512

dependent variable (RKT) with the other independent variables 513

(Table 1). 514

It is observed that the highest bias was equal to 0.053. As this 515

is a small bias for practical purposes and is in the conservative 516

direction (underestimating the population parameter), the results 517

were considered adequate for purposes of results interpretation 518

from the point of view of statistical significance and practical 519

importance. 520

Another way to check the sample size adequacy is through 521

analyzing the statistical power sensitivity, performed with 522

G*Power 3 software (Faul, Erdfelder, Buchner, & Lang, 2009). 523

For a sample of 78 respondents, with a significance level of 5% 524

F.F. Silveira et al. /Revista de Administraçâo xxx (2016) xxx—xxx

Table 1

Disattenuated correlations of the dependent variable RKT.


1. Age

2. Country

3. Complexity

4. Autonomy

5. Embeddedness

6. RKT

Correlation with RKT Composite reliability

Disattenuated correlation with RKT (using CR)

-0.150 1

0.350 1

0.270 0.890 0.307

0.300 0.880 0.343

0.410 0.900 0.463

1000 0.870


and statistical power of 0.80 (Cohen, 1998), the test 'sensitivity analysis' found that the model is able to detect an effect size of 0.1574, which is considered a medium effect (Cohen, 1998). Using the population effect formula f2 = R2/(1 — R2) (Cohen, 1998), it was concluded that the research will detect a minimum R2 of 0.1507.

latent variables were smaller than the square root of the average variance extracted of their latent variables (Fornell & Larcker, 1981). Thus, it can be said that the model presented convergent, discriminant and reliability validity. The means, standard deviations, reliability estimates and factor correlations are reported in Table 2.

531 Findings

Assessment of the structural model

532 The respondents included a large variety of Brazilian multi-

533 nationals ranging from natural resources (12%), consumer

534 goods (21%), basic inputs (32%), manufacturing (19%), system

535 assembly (10%) and raw materials for construction (6%). The

536 responding subsidiaries locations were: Latin America (42%),

537 North America (24%), Asia (14%), Europe (14%) and Africa

538 (5%). At the country level, the largest number of subsidiaries

539 are in the U.S. (15%), Argentina (15%), Colombia (10%) and

540 Mexico (9%). Moreover, China (6%) already appears as an

541 important destination for Brazilian subsidiaries. As to the size

542 and number of employees at the subsidiary, 56% of responding

543 subsidiaries are in the range 100-1000 employees, followed by

544 14% of subsidiaries employing more than 1000 workers. This

545 descriptive statistics shows that a relative percentage of sub-

546 sidiaries consist of consolidated companies abroad. As regard

547 to the subsidiaries' age, the majority (69%) is under ten years of

548 age, 22% are between ten and nineteen years and only 9% are

549 more than 20 years of activities. The entry mode of Brazilian

550 subsidiaries abroad represents 77% acquisitions and 23% direct

551 investment or greenfield investment.

552 Evaluation of the measurement model

In measuring the constructs, the model was conducted by evaluating the convergent, discriminant and reliability validity. As presented in Table 1, the constructs (also called latent variables) were measured using reflective indicators to verify the adequate reliability of the Cronbach's alpha values. In addition, all latent variables achieved convergent validity, that is, they have an average variance extracted (AVE) higher than 0.5, and composite reliability higher than 0.7 (Hair et al., 2014; Henseler et al., 2009; Tenenhaus, Vinzi, Chatelin, & Lauro, 2005). However, three items of the scales had to be removed from the model so that the AVE reached the reference value (3.7; 4.2; 4.4 in Appendix 1). The discriminant validity is verified by the For-nell Larcker criterion and was evaluated through the cross-loads analysis. This facilitated to determine whether a construct is truly distinct from other constructs through empirical patterns. Based on this result, it was noted that all correlations amongst the

The structural model is able to specify the relationship patterns amongst the constructs. The model was assessed using five criteria: (i) path coefficients (ft); (ii) path significant (p-value); (iii) variance explain (R2); (iv) effect size (f2) and (v) predictive relevance (Q2). According to Hair et al. (2014), the main criteria for the structural model evaluation are the coefficient of determination (R2) and the level and significance of the path coefficients (ft). To calculate them, the path weighting scheme and a bootstrapping technique were used with 78 observations and 500 random samples to estimate the t-values in order to assess the significance. For social science researches, R2 values of 0.26, 0.13 and 0.02 are considered strong, moderate and weak, respectively (Cohen, 1998).

Continuing Fichman and Kemerer (1997), in addition to the full model, we have evaluated two nested models (control model and theoretical model). In total, these three models were accessed to evaluate the true impact and the additional explanatory power of the theoretical variables after the variance explained by the control. The full model includes all this study variables, the control model includes only the control variables, and the theoretical model includes the hypothesized relationships. Comparisons amongst the three models are summarized in Table 3.

The R2 value results for the full model (including control variables) indicate that the variance of 36% in RKT was explained by the model. This result is considered substantial and provides evidence that the model is capable of explaining the dependent variable (Cohen, 1998). When comparing the results of the adjusted R2 (33%) with the sensitivity analysis on statistical power, it is found a R2 value well above the minimum detectable by the model, which is 15%.

A comparison between the full model and control model (location and age) shows that the control model explains an incremental variance on R2 of 19% on the dependent variable (RKT). The delta between the control model and the full model was (AR2 = 0.17). This result suggests that, despite having presented a moderate result, control variables alone do not provide a solid basis through which one can understand and predict RKT patterns.

600 601 602

610 611 612

alllhi.e in press

F.F. Silveira et al. /Revista de Administraçao xxx (2016) xxx-xxx

Table 2

Evaluation of the measurement model.

Variables Mean S.D. AVE C.R. C.A. 1 2 3 4 5 6

1. Age 10.8 11.1 1.00

2. Country - - 0.13 1.00

3. Complexity 3.7 1.1 0.61 0.89 0.86 -0.18 0.19 0.78

4. Autonomy 3.2 1.0 0.55 0.88 0.90 0.07 0.17 0.03 0.74

5. Embeddedness 3.0 1.0 0.53 0.90 0.87 0.08 0.28 0.02 0.07 0.73

6. Reverse Transfer 3.0 1.0 0.54 0.87 0.83 -0.15 0.35 0.27 0.30 0.41 0.73

Note 1: In bold on the diagonal, there are values of the square root of the average variance extracted. Note 2: AVE, average variance extracted; C.R., composite reliability; C.A., Cronbachs alpha. Note 3: AVE benchmarks: 0.5; composite reliability: 0.7; Cronbach's alpha: 0.6.

Table 3

Significance test results of the structural model path coefficients.

H Path from To Full model Control model Theoretical model Effect size (f2)

ß p-Values ß p-Values ß p-Values

Age RKT -0.182 0.01** -0.212 0.00***

Country RKT 0.195 0.04* 0.408 0.00***

H1 Autonomy RKT 0.188 0.04* 0.252 0.01** 0.05

H2 Complexity RKT 0.246 0.01** 0.262 0.00*** 0.08

H3 Embeddednes RKT 0.351 0.00*** 0.395 0.00*** 0.15

Reverse transfer R2 0.36 0.19 0.30

AR2 0.17 0.06

Values of t were calculated through bootstrapping with 500 resamples and 78 cases per sample. * p < 0.05. ** p <0.01. *** p < 0.001

615 Comparing the full model and the theoretical model, the

616 incremental variance derived by the model is around 30% for

617 RKT. Results indicate that the theoretical model in this study is

618 substantive enough to explain the variance in the research model.

619 However, control variables were responsible for a considerable

620 proportion of the variance in the R2 value of RKT. As the pre-

621 dicted paths for the structural model, all the hypothesized were

622 statistically significant. The confidence level in the prediction

623 model was measured by the indicator Q2 which must be higher

624 than zero. The Q2 value to construct 'RKT' is 0.171 ensuring

625 the model predictive relevance (Hair et al., 2014; Henseler et al.,

626 2009).

627 The effect size (f2) measures the magnitude of an indepen-

628 dent variable on a dependent variable (Tabachnick & Fidell,

629 2007). The exogenous constructs omission of the model can

630 be used to assess in which case these omitted constructs have

631 substantial impact on the endogenous constructs. Cohen (1998)

632 provided values of 0.02, 0.15 and 0.35 considered weak, mod-

633 erate and strong, respectively. The f2 is also calculated by R2

634 included = f2 - R2 excluded/1 - R2 included (Hair et al., 2014).

635 Following, Table 3 shows the significance results of each path

636 amongst the latent variables and the effect size.

637 The results support two of the three hypotheses statements.

638 Hypothesis H1 shows that autonomy has a positive and sig-

639 nificant effect on reverse transfer (^ = 0.19, p <0.05). The

640 effect size (f2) of 0.05 indicates that the construct subsidiary

641 autonomy has a weak effect on the endogenous latent variable

642 RKT (Cohen, 1998). Hypothesis H2 states that the lower the

complexity of subsidiary's R&D knowledge, the larger the rate of reverse technology transfer to headquarters. Surprisingly, this study's results showed that knowledge complexity has a significant, but positive effect on reverse transfer (^ = 0.25, p <0.01). This relationship is characterized by a weak effect (0.08) on the endogenous latent variable 'RKT' (Cohen, 1998). Finally, the results showed that subsidiary embeddedness has a significant and positive effect (0.15) on RKT, which confirms H3 hypothesis (^ = 0.35,p < 0.001). This relationship is characterized by a moderate to strong effect on the endogenous latent variable 'RKT' (Cohen, 1998).

With regard to the control variables, the localization effect was positive and significant (^ = 0.19, p = 0.05) for RKT, indicating that subsidiaries located in developing countries are more likely to transfer knowledge in reverse direction. Also for the subsidiary age variable the coefficient is significant (j3 = -0.18, p = 0.01) but the negative sign indicates that RKT is more likely to occur from young subsidiaries, confirming the findings of Bezerra et al. (2015).


Despite ambiguous evidence about RKT in Brazil (Fleury & Fleury, 2011), this study found that Brazilian subsidiaries with a high autonomy degree are more capable of transferring knowledge back to headquarters, confirming our hypothesis H1. An argument on the positive effect of autonomy for RKT is based on the idea that the subsidiaries independence provides

660 661

F.F. Silveira et al. /Revista de Administrafao xxx (2016) xxx—xxx 9

669 greater access to local knowledge databases, knowledge from With regard to the first control variable (location), the results 726

670 local partners and possibilities to innovate (Andersson et al., indicated that subsidiaries located in developed markets, such 727

671 2002; Ciabuschi & Martin, 2012; Gupta & Govindarajan, 1991; as North America and Europe, are probably the ones that most 728

672 Mudambi & Cantwell, 2005). Hence, subsidiary autonomy is transfer knowledge to their headquarters. This result is in line 729

673 recognized as an important predictor of reverse knowledge trans- with several contributions in the literature which state that the 730

674 fer in the context of EMNEs. Autonomy empowers subsidiaries innovation capacity of subsidiaries largely depends on the host 731

675 to explore their own business and market opportunities so that countries advantages (Gupta & Govindarajan, 2000; Mudambi 732

676 they can make use of external sources to their competitive & Cantwell, 2008; Yang et al., 2008). 733

677 advantage. Taking into account that Brazilian multinationals Based on previous findings about EMNEs, two perspectives 734

678 are still at an early stage of internationalization, it is a new can be presented. The first perspective, led by Cuervo-Cazurra 735

679 phenomenon the fact that their subsidiaries have been seek- andGenc(2008),Ramamurti(2008),KhannaandPalepu(2011), 736

680 ing for autonomy and independence from their headquarters' Cuervo-Cazurra (2012) and Ramamurti (2012), argues that 737

681 decisions. EMNEs have a new type of capability, unlike the traditional 738

682 This paper identified that knowledge characteristics and sub- MNEs capabilities, which is related to the ability of coping with 739

683 sidiary characteristics determine the rate of reverse knowledge the institutional deficiencies to which they are exposed. This 740

684 transfer from subsidiaries to emerging market MNEs. First, current advocates that emerging MNEs, for having operated 741

685 from the knowledge characteristics viewpoint, it was possible to in environments presenting difficult conditions, such as under- 742

686 show that the knowledge complexity level has a positive impact developed premises, corrupt bureaucracies, poor educational 743

687 on the extent of RKT. This finding is contrary to this study's institutions and unstable governments, have the "advantages 744

688 hypothesis (H2), which suggested that the lower the subsidiary of adversity." The second perspective, led by authors such as 745

689 knowledge complexity, the greater the RKT. It is suspected that Mathews (2006) and Child and Rodriguez (2005), argue that 746

690 one of the reasons for this intriguing, but interesting result, MNEs place their subsidiaries in developed countries as a way 747

691 may be related to the knowledge complexity paradox, because, to leverage their productive, technological and marketing effi- 748

692 while knowledge transfer encounters higher costs problems, it is ciency, following an asset-seeking strategy, looking for their 749

693 the most compensatory type of knowledge to the headquarters. competitive advantages increase. Therefore, the preferences of 750

694 Thus, it is suspected that the Brazilian multinationals try to trans- emerging MNEs for developing markets exemplify their ten- 751

695 fer the most complex knowledge developed in their subsidiaries, dency to explore the "institutional voids". However when it 752

696 regardless of the complexity levels associated, which includes comes to subsidiaries that transfer knowledge in the reverse 753

697 the involvement to a greater extent, of the headquarters so that direction, they are more likely to be in countries where there 754

698 this type of transfer actually materializes (Nair et al., 2015). are better infrastructure conditions, business support institutions 755

699 Such a result is also in line with the framework of learning and and favorable legal environment. 756

700 effective leverage (Mathews, 2006) of the EMNE's resources Regarding the second control variable (age), the results 757

701 and networks abroad (LLL chart). Other possible explanation is surprisingly indicated that there was a significant correlation, 758

702 the effect of subsidiary's role. For example, more innovative sub- though negative, between age and RKT. Thus, the younger the 759

703 sidiaries might transfer more complex R&D knowledge, which subsidiary, the more likely the existence of RKT. A possible 760

704 suggests that implementer and contributor subsidiaries may not explanation for this unexpected result is the fact that experience 761

705 transfer (or transfer to a lesser extent) complex type knowledge. leads to efficiency gains, but on the other hand, in environ- 762

706 In summary, although the initial H2 was not supported, this result ments where changes occur very rapidly, the adjustment between 763

707 provides an opportunity to suggest that innovative subsidiaries organizational capabilities and market demands declines, as 764

708 may engage in complex knowledge transfer and thus become a the subsidiaries grow older, having in view that more mature 765

709 competitive player. companies take longer to incorporate the most current techno- 766

710 Our results also support the hypothesis H3 which proposes logical developments (S0rensen & Stuart, 2000). It is in this 767

711 that local embeddedness impacts the rate of RKT. It was found perspective that age and accumulated skills can become disad- 768

712 that embeddedness with suppliers and customers, in other words, vantages when compared to younger subsidiaries. Particularly, 769

713 localbusinessnetworksincreasethepossibilityofgainingaccess this occurs with regard to the company's ability to adapt or 770

714 to new knowledge, which can subsequently be transferred to develop major technological changes (S0rensen & Stuart, 2000; 771

715 EMNEs. This paper confirms that subsidiaries from emerg- Tushman & Anderson, 1986). With respect to the group of 772

716 ing market multinationals become internationalized in order emerging MNEs, younger subsidiaries may be more influen- 773

717 to explore knowledge and existing capabilities in foreign mar- tial in the headquarters' knowledge exactly because they are 774

718 kets as well as to develop new knowledge and capabilities able to be more agile and dynamic in relation to technological 775

719 through knowledge available in the subsidiaries' host environ- developments. 776

720 ment (Narula, 2012). For subsidiaries is essential to be embedded

721 in local business networks to obtain distinctive knowledge devel- Conclusions, limitations and further research 777

722 opment. New connections with local networks allow subsidiaries

723 to perform innovative tasks for headquarters, instead of tasks This paper explained reverse knowledge flows in subsidiaries 778

724 limited to adaptation of products and processes to the local of emerging market multinationals and tested the impact of 779

725 market (Borini & Fleury, 2011). three determinants in Brazilian multinationals (Govindarajan & 780

F.F. Silveira et al. /Revista de Administraçâo xxx (2016) xxx-xxx

800 801 802

808 809

Ramamurti, 2011). Hence, several contributions to the knowledge flow of RKT in Brazilian MNEs are suggested. First, in comparison with traditional MNEs, Brazilian MNEs have a higher interest in reverse technology transfer, due to the higher importance of subsidiaries for headquarters. Second, subsidiaries of Brazilian MNEs will transfer products' knowledge just with a basic and intermediate level of technological complexity (Ariffin & Figueiredo, 2003; Iammarino et al., 2008). Third, on the process of RKT in foreign subsidiaries of Brazilian MNEs, this work explored the impact of knowledge complexity characteristics as well subsidiary characteristics, i.e. autonomy and embeddedness. The results showed that RKT is positively affected by knowledge complexity, subsidiary autonomy and embeddedness of foreign subsidiaries with customers and suppliers. Fourth, it was assessed the effect of the subsidiary's location and age on the RKT. The results indicate that subsidiaries located in developed countries are more likely to transfer knowledge in reverse direction as well as younger subsidiaries. This paper's empirical implications suggest that subsidiaries with higher access to local knowledge will be better positioned to acquire new knowledge and consequently transfer it back to headquarters. The external embeddedness has been indicated as an important determinant of RKT. From the viewpoint of practical implications, it is necessary that subsidiaries invest in mechanisms of relationship and knowledge exchange to establish strong collaborations with local partners. These findings may also be useful for policy makers in as much as understanding the innovation transfer pattern is a key component of a country's innovation system.

An important limitation of this study is that this research is 8io

limited to the narrow context of Brazilian subsidiaries, which 8ii

therefore imposes limits to the results generalization. Second, 812

the sample size and sample composition turn it difficult to make 813

far-reaching generalizations of its results. Third, the survey 8i4

method provides a snapshot that reduces the information source 815

credibility, the access to the right people, the responses control, 8i6

and the utilization of only one respondent by company. Fourth, 8i?

its choice of control variables, which could have covered other 8i8

aspects, possibly stakeholders in the achieved result. Finally, 8i9

it is assumed some restrictions related to the unit of analysis 820

and the information from headquarters. Further researches could 82i

explore the autonomy and integration degree of subsidiaries 822

from emerging markets multinationals. 823

Conflicts of interest 824

The authors declare no conflicts of interest. Q14 825

Uncited references Q15 826

Breschi and Lissoni (2001), Dimaggio and Powell (1983), 827

Lall (1983), Minbaeva (2007), Reddy (2011), and Ringle, 828

Wende, and Will (2005). 829

Appendix A. Operational definition of model variables Q16 8



Dependent variables Reverse knowledge transfer (RKT)

Independent variables Complexity



embeddedness (with customers, suppliers)

Moderating variables Subsidiary's location

Subsidiary's age

1.1 Development of new production process; 1.2 Development of new equipment and/or tools; 1.3 Development of new products; 1.4 Know-how and expertise in the form of plans, models, instructions, guides, formulas, specifications, designs, plans, technical drawings, and/or prototypes to design new products; 1.5 Results of research into new materials and specifications; 1.6 Results of research and development (R&D) into new product generations.

2.1 Its understanding requires prior learning from other related technological knowledge; 2.2 Its understanding requires a large amount of information; 2.3 It is the product of many interdependent routines, individuals and resources; 2.4 It includes many different skills or competencies; 2.5 It is technologically sophisticated and difficult to deploy; 2.6 It is complex (vs. simple) 3.1 Implementation of changes in products and services; 3.2 Development of new products and services; 3.3 Implementation of changes in production processes; 3.4 Entry into new markets in the country; 3.5 Procurement and supply chain management; 3.6. Management of Purchasing and Supply Chain; 3.7 Hiring and firing of the subsidiary workforce.

4.1 Customers/suppliers has fully participated in the development of technological knowledge in the subsidiary; 4.2 Customers/suppliers showed important initiatives for the development of technological knowledge in the subsidiary; 4.3 Customers/suppliers satisfied the requirements in developing technological knowledge in the subsidiary; 4.4 The technological subsidiary knowledge was partially developed within this Customers/suppliers' premises; 4.5 The cooperation with customers/suppliers has been characterized by frequent interactions.

5.1 Low-income countries (0); 5.2 High-income countries (1)

6.1 subsidiaries under 10 years old (0); 6.2 Subsidiaries with over 10 years old (1)

Ariffin and Figueiredo (2003); Bell and Pavitt (1995); Iammarino et al. (2008); Lall (1992); Yang et al. (2008)

Simonin (2004); Zander and Kogut (1995)

Ghoshal and Nohria (1989); Birkinshaw et al. (1998); Rabiosi (2011)

Lane and Lubatkin (1998), Andersson et al. (2005) and Najafi-Tavani et al. (2013)

Mudambi and Cantwell (2005)

Ambos and Schlegelmilch (2007); Rabiosi (2011)

F.F. Silveira et al. /Revista de Administraçâo xxx (2016) xxx—xxx


Ambos, T., Ambos, B., & Schlegelmilch, B. (2006). Learning from foreign subsidiaries: An empirical investigation of headquarters benefits from reverse knowledge transfers. International Business Review, 15(3), 294-312.

Ambos, T. C. (2015). Reverse knowledge transfer. The Palgrave encyclopedia of strategic management. Palgrave Macmillan.

Andersson, U., Forsgren, M., & Holm, U. (2002). The strategic impact of external networks: Subsidiary performance and competence development in the multinational corporation. Strategic Management Journal, 23(11), 979-996.

Andersson, U., Bjorkman, I., & Forsgren, M. (2005). Managing subsidiary knowledge creation: The effect of control mechanisms on subsidiary local embeddedness. International Business Review, 14, 521-538.

Andersson, U., Forsgren, M., &Holm, U. (2007). Balancing subsidiary influence in the federative MNC: A business network view. Journal of International Business Studies, 38(4), 802-818.

Ariffin, N., & Figueiredo, P. (2003). Internationalization of innovative capabilities: Counter-evidence from the electronics industry in Malaysia and Brazil. In Paper for DRUID summer conference (pp. 12-14).

Bartlett, C. A., & Ghoshal, S. (1989). Managing across borders. The transnational solution. Boston, MA: Harvard Business School Press.

Bartlett, C. A., & Ghoshal, S. (2000). Transnational management (3rd ed.). New York: McGraw Hill.

Bell, M., & Pavitt, K. (1995). The development of technological capabilities. In Trade, technology and international competitiveness (1st ed.). Washington, DC: World Bank.

Bezerra, M. A., & Borini, F. M. (2015). The impact of social and relational contexts on innovation transfer in foreign subsidiaries. International Journal of Learning and Intellectual Capital, 12.

Birkinshaw, J., & Hood, N. (1998). Multinational subsidiary evolution: Capability and charter change in foreign-owed subsidiaries companies. Academy of Management Review, 23(4), 773-795.

Birkinshaw, J., Hood, N., & Jonsson, S. (1998). Building firm-specific advantages in multinational corporations: The role of subsidiary initiative. Strategic Management Journal, 19, 221-241.

Birkinshaw, J. (1997). Entrepreneurship in multinational corporations: The characteristics of subsidiary initiatives. Strategic Management Journal, 18(3), 207-229.

Borini, F. M., & Fleury, M. T. L. (2011). Development of non-local competences in foreign subsidiaries of Brazilian multinationals. European Business Review, 23, 106-119.

Borini, F., Oliveira, M. M., Jr., Silveira, F. S.,& Concer, R. O. (2012). Thereverse transfer of innovation of foreign subsidiaries of Brazilian multinationals. European Management Journal, 30(3), 219-231.

Borini, F. M., Costa, S., Bezerra, M. A. B., & Oliveira, M. M., Jr. (2014). Reverse innovation as an inducer of centres of excellence in foreign subsidiaries of emerging markets. International Journal of Business and Emerging Markets, 6(2), 163-182.

Breschi, S., & Lissoni, F. (2001). Knowledge spillovers and local innovation systems: A critical survey. Industrial and Corporate Change, 10(4), 975-1005.

Cantwell, J., & Mudambi, R. (2005). MNE competence-creating subsidiary mandates. Strategic Management Journal, 26, 1109-1128.

Cantwell, J., & Piscitello, L. (1999). The emergence of corporate international networks for the accumulation of dispersed technological competences. Management International Review, 39, 123-147.

Ciabuschi, F., & Martín, M. O. (2012). Knowledge ambiguity, innovation and subsidiary performance. Baltic Journal of Management, 7(2), 143-166.

Child, J., & Rodriguez, S. B. (2005). The internationalization of Chinese firms: A case for theoretical extension? Management and Organization Review, 1(3), 381-410.

Chin, W. W., & Newsted, P. R. (1999). Structural equation modeling analysis with small samples using partial least squares. In R. H Hoyle (Ed.), Statistical strategies for small sample research (pp. 307-341). Thousand Oaks, CA: Sage Publications.

Chu, R. A., & Wood, T., Jr. (2008). Cultura organizacional brasileira pós-globalizacao: global ou local? Revista de Administrando Pública, 42(5), 969-991.

Cohen, J. (1998). Statistical power analysis for the behavioral sciences. Hills-dale, NJ: Lawrence Erlbaum Associates.

Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. ASQ, 35, 128-152.

Cooper, D. R., & Schindler, P. S. (2003). Business research methods (8th ed.). Boston: McGraw-Hill Irwin.

Criscuolo, P. (2005). Inter-firm reverse technology transfer: The home country effect of R&D internationalization. Industrial and Corporate Change, 18(5), 869-899.

Criscuolo, P., & Narula, R. (2007). Using multi-hub structures for international R&D: Organisational inertia and the challenges of implementation. Management International Review, 47(5), 639-660.

Cuervo-Cazurra, A., & Genc, M. (2008). Transforming disadvantages into advantages: Developing-country MNEs in the least developed countries. Journal ofInternational Business Studies, 39(7), 957-979.

Cuervo-Cazurra, A. (2012). Business relationships between Latin America and Asia: The next research Frontier. Globalization, Competitiveness, and Gov-ernability, 6(1), 16-22.

Dimaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147-160.

Dunning, J. H. (1993). Multinational enterprises and the global economy. Wokingham, England/Reading, MA: Addison Wesley.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A. G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149-1160.

Fichman, R. G., & Kemerer, C. F. (1997). The assimilation of software process innovations: An organizational learning perspective. Management Science, 43(10), 1345-1363.

Figueiredo, P. N. (2005). Acumulacao tecnológica e Inovacao industrial: con-ceitos, mensuracao e evidencias no Brasil. Sao Paulo em Perspectiva, 19(1),


Figueiredo, P. N. (2011). The role of dual embeddedness in the innovative performance of MNE subsidiaries: Evidence from Brazil. Journal ofManagement Studies, 48(2), 417-440.

Figueiredo, P. N., & Brito, K. (2011). The innovation performance of MNE subsidiaries and local embeddedness: Evidence from an emerging economy. Journal ofEvolutionary Economics, 21(1), 141-165.

Fleury, A. (Ed.). (2010). Gestao Empresarial para a internacionalizando das empresas brasileiras. Ed. Atlas.

Fleury, A., & Fleury, M. (2011). Brazilian multinationals. Competences for internationalization Cambridge.

Foss, N. J., & Pedersen, T. (2002). Transferring knowledge in EMNs: The role of sources of subsidiary knowledge and organization context. Journal of International Management, 8, 49-67.

Fornell, C., & Larcker, D. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39-50.

Frost, T. S., & Zhou, C. (2005). R&D co-practice and reverse knowledge integration in multinational firms. Journal ofInternational Business Studies, 36, 676-687.

Frost, T., Birkinshaw, J., & Ensign, P. (2002). Centers of excellence in multinational corporations. Strategic Management Journal, 23, 997-1018.

Galina, S. V. R., & Moura, P. G. D. (2013). Internationalization of R&D by Brazilian multinational companies. International Business Research, 6(8),


Gammelgaard, J., Holm, U., & Pedersen, T. (2004). The dilemmas of MNC subsidiary knowledge transfer. In V. Mahnke, & T. Pedersen (Eds.), Knowledge flows, governance and the multinational enterprise. New York: Palgrave Macmillan.

Ghoshal, S., & Bartlett, C. A. (1988). Creation, adoption, and diffusion of innovations by subsidiaries of multinational corporations. Journal ofInternational Business Studies, 19, 365-388.

Ghoshal, S., & Nohria, N. (1989). Internal differentiation within multinational corporations. Strategic Management Journal, 10, 323-337.

Govindarajan, V., & Trimble, C. (2012). Reverse innovation: A global growth strategy that could pre-empt disruption at home. Strategy & Leadership, 40(5), 5-11.

12 F.F. Silveira et al. /Revista de Administraçâo xxx (2016) xxx—xxx

Govindarajan, V., & Ramamurti, R. (2011). Reverse innovation, emerging markets, and global strategy. Global Strategy Journal, 1(3-4), 191-205.

Guillén, M. F., & García-Canal, E. (2009). The American model of the multinational firm and the new multinationals from emerging economies. Academy of Management Perspectives, 23, 23-35.

Grant, R. (1996). Toward a knowledge-based theory of the firm. Strategic Management Journal, 17, 109-122.

Grabher, G. (1993). The weakness of strong ties: The lock-in of regional networks? In G. Grabher (Ed.), The embedded firm — On the socioeconomics of industrial networks (pp. 255-277). London: Routledge.

Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91(3), 481-510.

Gupta, A. K., & Govindarajan, V. (2000). Knowledge flows within the multinational corporation. Strategic Management Journal, 21, 473-496.

Gupta, A., & Govindarajan, V. (1991). Knowledge flows and the structure of control within multinational corporations. Academy of Management Review, 16(4), 768-792.

Hair, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. A. (2014). Primer on partial least squares structural equation modeling (PLS-SEM). Sage: Thousand Oaks.

Hákanson, L., & Nobel, R. (2001). Organizational characteristics and reverse technology transfer. Management International Review, 41, 395-420.

Henseler, J., Ringle, C. M., & Sinkovics, R. R. (2009). The use of partial least squares path modeling in international marketing. In R. R. Sinkovics, & P. N. (Org. ) Ghauri (Eds.), New challenges to international marketing (pp. 277-319).

Holm, U., & Pedersen, T. (2000). The dilemma of centres of excellence: Contextual creation of knowledge versus global transfer of knowledge. Copenhagen Business School Department of International Economics and Management, 3, 8-20.

Hui, B. S., & Wold, H. O. (1982). Consistency and consistency at large of partial least squares estimates. In K. G. Joreskog, & H. Wold (Eds.), Systems under indirect observations: Causality, structure, prediction, Part II (pp. 119-130). Netherlands: North-Holland Publishing Company.

Iammarino, S., Padilla-Pérez, R., & Von Tunzelmann, N. (2008). Technological capabilities and global-local interactions. The electronics industry in two Mexican regions. World Development, 36(10), 1980-2003.

Immelt, J. R., Govindarajan, V., & Trimble, C. (2009). How GE is disrupting itself. Harvard Business Review, 56-65.

Jansen, J., Van Den Bisch, F., & Volberda, H. (2005). Managing potential and realized absorptive capacity: How do organizational antecedents matter? Academy of Management Journal, 48(6), 999-1015.

Joreskog, K. G., & Wold, H. (1982). The ML and PLS techniques for modeling with latent variables: Historical and competitive aspects. In K. G. Joreskog, & H. Wold (Eds.), Systems under indirect observation, Part 1 (pp. 263-270). North-Holland: Amsterdam.

Khanna, T., & Palepu, K. G. (2011). Winning in emerging markets: A road map for strategy and execution. Boston, MA: Harvard Business Press.

Kogut, B., &Zander, U. (1992). Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science, 3, 3.

Lane, P. J., & Lubatkin, M. (1998). Relative absorptive capacity and interorga-nizational learning. Strategic Management Journal, 19, 461-478.

Lall, S. (Ed.). (1983). The new multinationals: The spread of third world enterprises. New York: John Wiley Sons.

Luo, Y., & Tung, R. (2007). International expansion of emerging market enterprises. Journal of International Business Studies, 38(4), 481-498.

Lohmoller, J.-B. (1989). Latent variable path modeling with partial least squares. pp. 1989. Heidelberger: Physica-Verlag.

Lyles, M. A., & Salk, J. E. (1996). Knowledge acquisition from foreign parents in international joint ventures: An empirical examination in the Hungarian context. Journal of International Business Studies, 27, 877-903.

Mathews, J. (2006). Dragon multinationals: New players in the 21st century. Asia-Pacific Journal of Management, 23, 2-27.

Meyer, K., Mudambi, R., & Narula, R. (2011). Multinational enterprises and local contexts: The opportunities and challenges of multiple embeddedness. Journal ofManagement Studies, 48(2), 235-252.

Michailova, S., & Mustaffa, Z. (2012). Subsidiary knowledge flows in multinational corporations: Research accomplishments, gaps, and opportunities. Journal of World Business, 47(3), 383-396.

Minbaeva, D. B. (2007). Knowledge transfer in multinational corporations. Management International Review, 47, 567-593.

Mudambi, R., Mudambi, S. M., & Navarra, P. (2007). Global innovation in MNEs: The effects of subsidiary self-determination and teamwork. Journal of Product Innovation Management, 24, 442-455.

Mudambi, R., & Navarra, P. (2004). Is knowledge power? Knowledge flows, subsidiary power and rent-seeking within EMNs. Journal of International Business Studies, 35(5), 385-406.

Mudambi, R., Piscitello, L., & Rabbiosi, L. (2014). Reverse knowledge transfer in MNEs: Subsidiary innovativeness and entry modes. Long Range Planning, 47(1-2), 49-63.

Najafi-Tavani, Z., Giroud, A., & Andersson, U. (2013). The interplay of networking activities and internal knowledge actions for subsidiary influence within MNCs. Journal of World Business, 4(1), 122-131.

Nair, S. R., Demirbag, M., & Mellahi, K. (2015). Reverse knowledge transfer from overseas acquisitions: A survey of Indian MNEs. Management International Review, 55(2), 277-301.

Narula, R. (2012). Do we need different frameworks to explain infant MNEs from developing countries? Global Strategy Journal, 2(3), 188-204.

Nohria, N., & Ghoshal, S. (1994). Differentiated fit and shared values: Alternatives for managing headquarters-subsidiary relations. Strategic Management Journal, 15, 491-502.

Nohria, N., & Ghoshal, S. (1997). The differentiated network: Organizing multinational corporations for value creation. San Francisco: Jossey-Bass Publishers.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill.

Powell, W. W., Koput, K. W., & Smith-Doerr, L. (1996). Interorganizational collaboration and the locus of innovation: Networks of learning in biotechnology. Administrative Science Quarterly, 41, 116-145.

Rabiosi, L. (2011). Subsidiary roles and reverse knowledge transfer: An investigation of the effects of coordination mechanisms. Journal ofInternational Management, 17, 97-113.

Rabiosi, L. (2008). The impact of subsidiary autonomy on reverse knowledge transfer: Movingfrom single to interdependent explanations. Working paper. Copenhagen Business School. Center for Strategic Management and Globalization.

Rabiosi, L., & Santangelo, G. D. (2011). Parent company benefits from reverse knowledge transfer: The role of the liability of newness in MNEs. In Pro-ceeedings of DRUID 2011.

Ramamurti, R. (2008). What have we learned about emerging-market MNEs? Insights from a multi-country research project. Emerging multinationals: Outward FDI from emerging and developing economies. Copenhagen, Denmark: Copenhagen Business School.

Ramamurti, R., & Singh, J. V. (Eds.). (2009). Emerging multinationals in emerging market. Cambridge University Press.

Ramamurti, R. (2012). What is really different about emerging market multinationals? Global Strategy Journal, 2(1), 41-47.

Reddy, P. (2011). Global innovation in emerging economies. New York: Rout-ledge.

Reed, R., & DeFillippi, R. (1990). Causal ambiguity, barriers to imitation, and sustainable competitive advantage. Academy of Management Review, 25, 88-102.

Ringle, C. M., Wende, S., & Will, A. (2005). SmartPLS 3.0, Hamburg.

Roth, K., & Morrison, A. J. (1992). Implementing global strategy. Characteristics of global subsidiary mandates. Journal of International Business Studies, 23(4), 715-735.

Simonin, B. L. (2004). An empirical investigation of the process of knowledge transfer in international strategic alliances. Journal of International Business Studies, 35, 407-427.

Simonin, B.L. (1999). Transfer of marketing know-how in international strategic alliances: An empirical investigation of the role and antecedents of knowledge ambiguity. Journal of International Business Studies, 30, 463-490.

F.F. Silveira et al. /Revista de Administraçäo xxx (2016) xxx—xxx 13

1101 Szulanski, G. (1996). Exploring internal stickiness: Impediments to the transfer Uzzi, B., à Gillespie, J. J. (2002). Knowledge spillover in corporate finan- 1121

1102 of best practice within the firm. Strategic Management Journal, 17, 27-43. cing networks: Embeddedness and the firm's debt performance. Strategic 1122

1103 S0rensen, J. B., & Stuart, T. E. (2000). Aging, obsolescence and organizational Management Journal, 23(7), 595-618. 1123

1104 innovation. Administrative Science Quarterly, 45, 81-112. Van Wijk, R., Jansen, J. J. P., à Lyles, M. (2008). A inter- and intra- 1124

1105 Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). organizational knowledge transfer: A meta-analytic review and assessment 1125

1106 Boston: Allyn and Bacon. of its antecedents and consequences. Journal ofManagement Studies, 45(4), 1126

1107 Tenenhaus, M., Vinzi, E. V., Chatelin, Y., & Lauro, C. (2005). PLS path model- 830-853. 1127

1108 ing. Computational statistics and data analysis, 48, 159-205. Vernon-Wortzel, H., à Wortzel, L. H. (1998). Globalizing strategies for multi- 1128

1109 Tsai, W., & Ghoshal, S. (1998). Social capital and value creation: The role of nationals from developing countries. Columbia Journal of World Business, 1129

1110 intrafirm networks. Academy of Management Journal, 41, 464-476. Spring, 23(1), 27-35. 1130

1111 Tsai, W. (2001). Knowledge transfer in inter-organizational networks: Effects Yamin, M., à Otto, J. (2004). Patterns of knowledge flows and MNE innovative 1131

1112 of network position and absorptive capacity on business unit innovation and performance. Journal of International Management, 10, 239-258. 1132

1113 performance. Academy of Management Journal, 44(5), 996-1004. Yang, Q., Mudambi, R., à Meyer, K. E. (2008). Convention and reverse knowl- 1133

1114 Tsai, W. (2002). Social structure of "coopetition" within a multiunit organiza- edge flows in multinational corporations. Journal of Management, 34(5), 1134

1115 tion: Coordination, competition, and intraorganizational knowledge sharing. 882-903. 1135

1116 Organization Science, 13, 179-190. Zander, U., à Kogut, B. (1995). Knowledge and the speed of the transfer 1136

1117 Tushman, M. L., & Anderson, P. (1986). Technological discontinuities and orga- and imitation of organizational capabilities: An empirical test. Organization 1137

1118 nizational environments. Administration Science Quarterly, 31, 439-465. Science, 6(1), 76-92. 1138

1119 Uzzi, B. (1996). The sources and consequences of embeddedness for the

1120 economic performance of organizations: The network effect. American Sociological Review, 61, 674-698.