Does global value chain participation drive firms’ innovation in South Africa? Open Access

Innovation is widely recognized as a central driver of economic development and structural transformation. At the macroeconomic level, innovation promotes sustainable development by fostering productivity growth, diversification and long-term competitiveness (Gaglio et al., 2022; Naidoo et al., 2023). At the microeconomic level, innovation enables firms to improve productivity, profitability and competitiveness, thereby strengthening their position in domestic and international markets (Wadho and Azam, 2022; Wu and Wang, 2024). For firms operating in increasingly globalized environments, the adoption of innovation is therefore not optional but a critical source of competitive advantage (Odei et al., 2021).

Given these well-established benefits, the key challenge for economists and policymakers is no longer whether innovation matters, but rather what drives it. A growing body of literature shows that firms’ incentives and ability to innovate depend on a combination of internal characteristics (such as firm size, age, skills and research capabilities) and external conditions embedded in the business environment, including trade regulations, institutional quality and political and legal frameworks (Ayalew et al., 2020; Divella and Sterlacchini, 2021; Priyadarshini et al., 2024). Among these external factors, participation in global value chains (GVCs) has emerged as a particularly important and multifaceted determinant of firm behavior. GVC participation exposes firms to international competition and foreign technologies, creating pressure to upgrade and adapt to remain competitive.

This issue is especially salient in South Africa, where innovation performance remains a persistent concern despite sustained policy efforts^[1]. According to the Global Innovation Index, South Africa ranks 69^th out of 133 economies, reflecting a steady decline in its global innovation position over the past decade (WIPO, 2024). Although South Africa remains the second most innovative economy in sub-Saharan Africa after Mauritius, this relative standing masks deeper structural weaknesses. The decline has been driven by low innovation outputs and a deterioration in both innovation inputs and outputs (WIPO, 2024). At the firm level, private investment in research and development remains limited and innovation is consistently identified as a weakness in competitiveness assessments (Naidoo et al., 2023). World Bank Enterprise Survey data indicate that only 22% of South African firms invest in R&D, while just 4% report introducing a new product or service and only 2% report process innovation (World Bank, 2021).

At the same time, South Africa’s integration into global value chains lags behind that of other small open economies. The country has experienced deindustrialization, weak upgrading and limited structural transformation, particularly in manufacturing (Andreoni et al., 2021; Tregenna et al., 2021). Firm-level evidence shows that South African firms tend to specialize in non-customized intermediates and primary, unprocessed products, reflecting a pattern of comparative advantage based on natural resources and commodity exports (Mazzi et al., 2024). Although many firms participate in international trade, only a small share is integrated into GVCs, and participation varies substantially across sectors (Ndubuisi and Owusu, 2023). Medium- and high-technology sectors (such as pharmaceuticals, chemicals, electronics and machinery) display higher GVC participation rates, while large segments of manufacturing remain weakly connected to global production networks.

These stylized facts raise an important question: is South Africa’s weak innovation performance systematically linked to its limited integration into global value chains? From a theoretical perspective, GVC participation may influence innovation through knowledge-based mechanisms. Fragmentation of production across borders allows firms to access foreign knowledge embedded in imported intermediate inputs, stimulates upgrading through export quality requirements and exposes firms to new ideas through interactions with foreign buyers, suppliers and competitors (World Bank, 2020). Empirical studies highlight three main learning channels: learning-by-importing, upgrading through export participation and competitive pressure arising from international market exposure (Eissa and Zaki, 2023; Ndubuisi and Owusu, 2021; Aghion et al., 2023; Hruskova, 2024). However, these learning effects are not automatic and depend critically on firms’ absorptive capacity. Firms that invest in R&D or possess stronger internal capabilities are better positioned to absorb and exploit external knowledge generated through GVC participation (De Vries et al., 2021).

Against this backdrop, this paper investigates whether participation in global value chains contributes to firm-level innovation in South Africa. By focusing on a single country with relatively advanced industrial capabilities within the African context, the analysis provides a nuanced understanding of how GVC participation shapes innovation under specific institutional and structural conditions. Using firm-level data from the World Bank Enterprise Surveys, the study examines the relationship between GVC participation and innovation while accounting for firm heterogeneity and potential endogeneity.

This paper contributes to the literature in several ways. First, it provides firm-level evidence on the relationship between GVC participation and innovation in South Africa, a middle-income African economy with a relatively diversified manufacturing sector but persistently weak innovation performance. Second, the analysis explicitly addresses endogeneity concerns by using instrumental-variable techniques, complemented by extended probit and propensity score matching approaches, thereby strengthening the causal interpretation of the results. Third, the paper moves beyond average effects by examining heterogeneity across firms and highlighting the role of absorptive capacity (proxied by firm size, research and development engagement and foreign ownership) in moderating the innovation benefits of GVC participation.

By doing so, the study offers nuanced insights into the conditions under which GVC integration can foster innovation in developing-country contexts. Rather than treating GVC participation as an automatic pathway to upgrading, the results emphasize the importance of firm-level capabilities in transforming exposure to global markets into innovation outcomes. This perspective contributes to ongoing debates on trade-led development and industrial upgrading in Africa by demonstrating that the innovation gains from GVC participation are conditional rather than universal.

The remainder of the paper is structured as follows. Section 2 reviews the related literature and presents the conceptual framework. Section 3 describes the data and econometric methodology. Section 4 presents and discusses the empirical results and robustness checks. Section 5 concludes and outlines theoretical and policy implications for strengthening innovation performance in South Africa.

A substantial body of literature has examined the relationship between global value chain (GVC) participation and firm-level innovation, particularly in the context of globalization and industrial upgrading. Early contributions emphasize that integration into international markets can expose firms to new knowledge, technologies and competitive pressures, potentially stimulating innovation through learning-by-exporting and learning-by-importing mechanisms (Clerides, Lach, and Tybout, 1998; Wagner, 2007). By engaging in exporting activities, firms may be compelled to improve product quality, adapt designs to foreign market requirements and upgrade production processes in response to international competition.

In addition to exporting, access to imported intermediate inputs has been identified as a critical channel through which GVC participation can influence innovation. Imported inputs often embody advanced foreign technologies and superior quality, enabling firms to enhance productivity and develop new or improved products and processes (Amiti and Konings, 2007; Goldberg et al., 2010). Empirical studies provide robust evidence that firms using imported intermediates tend to exhibit higher productivity and innovation performance, particularly in developing-country contexts where domestic input quality may be limited. More contributions emphasize the role of vertical linkages within GVCs. Interactions with foreign buyers and lead firms can transmit technical knowledge, managerial practices and compliance requirements related to quality standards and certification (Gereffi, Humphrey, and Sturgeon, 2005). Such vertical relationships may facilitate process upgrading and incremental innovation, even in the absence of formal research and development activities. From this perspective, GVCs are viewed not merely as trade relationships but as organizational structures through which knowledge is diffused across borders.

More recent conceptual work reframes GVCs not simply as trade relationships but as structured learning environments, where innovation outcomes depend on governance structures, power relations and firms’ positions within value chains (Gereffi and Lee, 2016). Building on this perspective, contemporary conceptual studies argue that innovation upgrading depends less on participation per se and more on firms’ ability to move across functional and technological segments of the value chain (Ponte, Gereffi, and Raj-Reichert, 2019; Lema et al., 2018).

Empirical evidence generally supports a positive relationship between GVC participation and innovation. Using large cross-country data sets, Eissa and Zaki (2025) and Elshaarawy and Ezzat (2023) find that firms engaged in GVCs are more likely to innovate, although the magnitude of this effect varies across countries and firm types. In the African context, Ndubuisi et al. (2025) report similar findings, highlighting the potential of GVC participation to foster innovation. Moreover, Lema et al. (2015) document innovation gains from GVC integration in Brazil and India, while Brancati et al. (2017) and Ito et al. (2023) show that governance structures and firms’ positions within GVCs shape innovation outcomes in Italy and Japan, respectively.

Despite this generally positive evidence, the literature also points to important limitations. Financial constraints, weak domestic capabilities and shallow integration into value chains can reduce or offset the innovation benefits of GVC participation (Elshaarawy and Ezzat, 2023). Recent conceptual contributions stress that GVCs should be viewed as conditional learning platforms, where innovation outcomes depend on complementary firm capabilities and supportive institutional environments rather than automatic spillovers (Cirera and Maloney, 2017; Kaplinsky, 2021).

The concept of absorptive capacity has played a central role in explaining why firms differ in their ability to benefit from external knowledge sources. Originally introduced by Cohen and Levinthal (1990), absorptive capacity refers to a firm’s ability to recognize the value of new external information, assimilate it and apply it for commercial purposes. In the context of developing countries, absorptive capacity has been identified as a critical determinant of firms’ innovation performance, particularly when learning opportunities arise from international trade and foreign linkages. More recent conceptual work emphasizes that absorptive capacity is not a static attribute but a cumulative and evolutionary capability, shaped by firms’ prior learning, organizational routines and institutional context (Cirera and Maloney, 2017). From this perspective, exposure to global markets can only translate into innovation when firms possess the internal capabilities required to internalize and adapt external knowledge.

Empirical studies commonly proxy absorptive capacity using firm-level characteristics such as size, engagement in research and development (R&D), workforce skills and ownership structure. Larger firms may possess more diversified human capital, better managerial capabilities and greater access to finance, enabling them to invest in innovation-related activities and respond more effectively to external knowledge shocks (Cirera and Muzi, 2020). Similarly, firms that engage in R&D are more likely to have the internal knowledge base required to adapt, modify and improve upon technologies acquired from external sources (Cohen and Levinthal, 1990).

Foreign ownership has also been highlighted as an important dimension of absorptive capacity in developing economies. Foreign-owned firms may benefit from direct access to parent-firm technologies, global innovation networks and advanced managerial practices, which can enhance their ability to absorb and apply knowledge transmitted through trade and GVC linkages (Javorcik, 2004). At the same time, the presence of foreign firms can generate spillovers to domestic firms through backward and forward linkages, although the extent of such spillovers depends on local firms’ absorptive capacity.

In developing-country contexts, limited absorptive capacity has frequently been cited as a key constraint preventing firms from translating exposure to global markets into sustained innovation and upgrading. Even when firms participate in GVCs, weak internal capabilities may confine them to low-value-added segments, with limited scope for learning and innovation (Pietrobelli and Rabellotti, 2011; Taglioni and Winkler, 2016). This perspective underscores the importance of examining not only whether firms participate in GVCs, but also the conditions under which such participation leads to innovation.

The relationship between global value chain (GVC) participation and firm-level innovation has been extensively examined in the literature on international trade, industrial upgrading and innovation in developing economies. A central argument in this literature is that participation in GVCs can facilitate access to external knowledge, technologies and production practices that are not readily available in domestic markets (Gereffi, Humphrey, and Sturgeon, 2005; Taglioni and Winkler, 2016). Through exporting, importing intermediate inputs and engaging with foreign buyers and suppliers, firms may be exposed to new product specifications, quality standards and managerial practices that can stimulate innovation.

Several learning mechanisms underpin this relationship. First, learning-by-exporting suggests that firms improve their capabilities through exposure to international markets and competition, which may induce process improvements or product adaptations (Clerides, Lach, and Tybout, 1998; Wagner, 2007). Second, learning-by-importing highlights the role of imported intermediate inputs as carriers of embodied foreign technology, which can enhance firms’ production efficiency and innovation potential (Amiti and Konings, 2007; Goldberg et al., 2010). Third, vertical linkages within GVCs (particularly interactions with lead firms) can transmit knowledge related to production techniques, quality control and compliance with international standards (Gereffi et al., 2005; Pietrobelli and Rabellotti, 2011).

However, the literature also emphasizes that the innovation benefits of GVC participation are not automatic. Firms differ markedly in their ability to internalize and exploit external knowledge, a concept commonly captured by absorptive capacity. Originally defined by Cohen and Levinthal (1990), absorptive capacity refers to a firm’s ability to recognize the value of new information, assimilate it and apply it to commercial ends. In developing-country contexts, limited absorptive capacity has been identified as a key constraint preventing firms from translating GVC participation into sustained innovation and upgrading (Pietrobelli and Rabellotti, 2011; Taglioni and Winkler, 2016).

Empirical studies typically proxy absorptive capacity using firm-level characteristics such as size, engagement in research and development (R&D) and foreign ownership. Larger firms may benefit from scale economies, more diversified skill bases and better access to finance, which can facilitate learning and innovation (Cirera and Muzi, 2020). Firms that invest in R&D are more likely to possess the internal knowledge base necessary to adapt and improve upon external technologies (Cohen and Levinthal, 1990). Similarly, foreign-owned firms may have privileged access to parent-firm technologies and global knowledge networks, enhancing their capacity to absorb and apply knowledge obtained through GVC participation (Javorcik, 2004).

Based on this literature, the present study adopts a conceptual framework in which GVC participation influences firm-level innovation through learning and knowledge transmission channels, while the magnitude of this effect depends on firms’ absorptive capacity. Figure 1 summarizes this framework. GVC participation (captured through exporting, importing intermediate inputs or both) provides opportunities for learning through exposure to international markets, foreign technologies and vertical relationships with global partners. These learning opportunities can foster product and process innovation. At the same time, absorptive capacity conditions the extent to which firms are able to transform GVC-related exposure into observable innovation outcomes.

This framework directly informs the empirical analysis. It motivates the baseline estimation of the relationship between GVC participation and innovation and underpins the heterogeneity analysis examining whether the innovation effects of GVC participation vary systematically with firm-level absorptive capacity.

We use the World Bank Enterprise Survey (WBES) 2020 data for South Africa, to examine the effect of GVC participation on firm innovation. The 2020 data are the latest released by the World Bank at the time of this research. The World Bank conducts face-to-face interviews with firm managers or owners and collects information on firm characteristics across various dimensions such as international trade, innovation, cost of inputs, access to finance, number of workers, bribery, competition, taxation, sales, informality, business-government relations and performance measures. The WBES dataset covers firms representing nonagricultural, formal, private firms, and the survey uses stratified random sampling by location, size and sector with replacement techniques. In South Africa, the survey was conducted in four regions: Eastern Cape, Gauteng, KwaZulu-Natal and Western Cape. Moreover, the survey consists of firms with at least five full-time employees. As a part of our data filtering process, we exclude firms with missing information on innovation, trade and sales. Our final sample includes a total of 1,097 firms in 2020 in South Africa. We use these data to construct all the variables in our model.

It is important to acknowledge that the 2020 wave of the World Bank Enterprise Survey coincides with the COVID-19 pandemic, an exceptional period characterized by significant disruptions to economic activity, trade flows and firm operations. The pandemic may have affected both firms’ participation in global value chains and their innovation behavior through multiple and potentially opposing channels. On the one hand, supply chain disruptions, demand uncertainty and financial constraints may have reduced firms’ incentives or ability to invest in innovation. On the other hand, the crisis may have induced firms to adapt production processes, reorganize supply chains or introduce incremental innovations as part of resilience and survival strategies.

While the WBES 2020 remains the most recent and comprehensive firm-level data set available for South Africa at the time of this study, the results should therefore be interpreted as reflecting firm behavior during a period of heightened uncertainty. Importantly, the direction of the pandemic’s net effect on innovation and GVC participation is theoretically ambiguous, which limits concerns that the estimated relationship is mechanically driven by COVID-19 conditions alone. Nevertheless, this context represents a limitation of the analysis, and future research using panel data or post-pandemic surveys would be valuable to assess whether the observed effects persist in more stable economic conditions.

We construct firms’ GVC participation by following the definitions provided by Dovis and Zaki (2021) and used by Gopalan et al. (2022), Reddy et al. (2021) and Elshaarawy and Ezzat (2023). First, the least strict measure entails that firms are engaged in exporting or importing activities (one-way trader). The second measure entails that firms are engaged in exporting and importing activities (two-way trader). The third measure entails that firms are two-way traders and have either international quality certification or a share of their capital is owned by a foreign firm. The fourth measure is the strictest one, encompassing all four dimensions, namely exporting, importing, having a quality certification and foreign ownership. Thus, four dummy variables measuring the degree of participation in the GVC are shown in our model, taking the value of 1 if the corresponding measure is respected, and 0 otherwise. However, given that our sample is not very large, we use the less strict definitions in this study: one-way trader and two-way trader. Indeed, only 1% of the companies in the sample correspond to the third measure, and none to the fourth measure (see Table A1 in the Appendix).

To measure the firms’ innovation, this paper follows the approach used by Reddy et al. (2021) and Elshaarawy and Ezzat (2023). Innovation is measured by the ability of the firm to introduce a new product or process in the market. Thus, a dummy variable is used, taking the value of 1 for firms that have introduced a new product/service or process in the market in the last three years, and 0 otherwise.

Guided by the literature on the determinants of innovation (see Elshaarawy and Ezzat, 2023; Fritsch and Görg, 2015; Ito et al., 2023; Reddy et al., 2021), our empirical specification controls for size, age, FDI, website adoption, R&D, access to finance, skill of employees, training and some characteristics of the top manager (gender and experience). The description of these variables, and all other variables used in our analysis, are presented in Table A1 (in the Appendix) and Table 1, respectively. Table A2 (in the Appendix) displays the correlation matrix. The highest correlation value between independent variables in this table is 0.46, suggesting that multicollinearity is not a serious threat to our model.

Table 1 shows, among other things, the percentage of innovative firms and those participating in CVGs according to different definitions. In the sample, around 16% of firms are one-way traders (export or import), which represents the highest percentage of GVC participation. Around 5% of firms export and import simultaneously (two-way traders). One percent of firms are two-way traders with an international certification or foreign ownership, and almost no firm meets at least three criteria. Thus, the share of firms integrating in GVC decreases as the definition becomes stricter. Extending the analysis by industry, Figure 2 shows that GVC integration is concentrated in the manufacturing sector. These firms in the manufacturing sector have developed their production systems, acquired new technologies and started to have strong global integration of the production structure. Thus, firms become more efficient and highly productive, and their production process becomes the most fragmented across the country. Moreover, firm size is an important determinant of whether the firm is financially constrained or not.

Figure 3 confirms that large firms are more incentivized to participate in GVCs because they are more productive and they include offshoring and outsourcing as important goals in their trading strategy (Elshaarawy and Ezzat, 2023). Large firms have a higher ability to afford the fixed costs of exporting and importing compared to small- and medium-sized firms. However, Figure 3 shows that output innovation is not a function of firm size. Innovation is not a function of company size. On the contrary, it seems that small firms are as innovative as large ones and more innovative than medium-sized ones. Although larger firms in the sample are more likely to participate in global value chains, the descriptive statistics indicate relatively low unconditional innovation rates across firm size categories. This apparent discrepancy should be interpreted with caution. First, innovation is measured using binary indicators of product and process innovation, which may not fully capture incremental or informal innovation activities that are more prevalent among firms in developing-country contexts. Second, during the 2020 survey period, firms (particularly larger ones) may have prioritized short-term survival and operational continuity over formal innovation investments in response to the COVID-19 shock. As a result, unconditional innovation rates may understate the role of firm size in shaping innovation outcomes. Moreover, small firms have greater flexibility, faster decision-making and stronger entrepreneurial culture (Acs and Audretsch, 1988; Freel, 2000). Small firms focus on niche markets and rely on innovation to survive.

The Table 1 shows that only 5% of firms in the survey innovate, while around 95% do not. Indeed, innovative investments are costly and are related to uncertainty because they do not yield instantaneous returns. However, they yield long-term returns (Kerr and Nanda, 2015). Figure 4, which extends the analysis by industry, shows that there are twice as many innovative firms in the manufacturing sector compared to service firms. Manufacturing firms that are generally more productive are more likely to have innovative investments (Morris, 2018).

Data display an association between different GVC participation measures and innovation. Figure 4 presents the proportion of innovative firms by each GVC measure. The figure demonstrates a higher proportion of innovative firms within the CVG compared to those outside. We see that the gap is 14 percentage points when the CVG measure is the least strict and 17 percentage points when participation in the CVG is defined as importing and exporting simultaneously. These figures also show that, on average, the more firms are integrated into the CVG, the more they innovate (17% if GVC1 and 22% if GVC).

Our empirical strategy focuses on using a variety of firm-level measures representing both GVC participation and innovation and check the importance of GVC as a possible determinant of innovation. To that end, we estimate variants of the following parsimonious model of firm-level innovation:

We hypothesize a positive association between our measures of GVC and firm innovation. Considering that our firm-level innovation indicators are binary in nature, we estimate equation (1) using a probit model. The subscript i identifies the firm. Innovation is product or process innovation which is dummy variable. GVC participation is an independent dummy variable. As we defined above, this variable takes into consideration different dimensions of GVC by using two different dummy variables: GVC1 (one-way trader) and GVC2 (two-way trader). Z is the vector of control variables (size, age, R&D, ICT use, skill intensity, training, foreign capital participation, access to finance, gender and experience of top manager). It also includes full sets of industry and region dummies. In our firm-level model (1), $∅$ denotes the standard normal cumulative distribution. γ is firm specific effect and ε is the error term.

A key empirical challenge in estimating the relationship between global value chain (GVC) participation and firm-level innovation is potential endogeneity. Endogeneity may arise from reverse causality if more innovative firms are more likely to self-select into GVCs, as well as from omitted firm-level characteristics (such as managerial quality or unobserved capabilities) that simultaneously influence both innovation and GVC participation. Ignoring these issues could lead to biased estimates of the effect of GVC participation on innovation.

To address these concerns, the empirical strategy adopts an instrumental variable (IV) approach, combined with additional robustness checks. The IV strategy exploits exogenous variation in firms’ likelihood of participating in GVCs arising from external business environment constraints and industry–region characteristics, which affect firms’ engagement in international production networks but are plausibly unrelated to their innovation outcomes except through GVC participation. Specifically, the analysis instruments GVC participation using firms’ reported obstacles related to trade regulations and tax administration, as well as the average level of GVC participation at the industry–region level. These instruments are expected to influence firms’ access to and participation in global markets, while not directly affecting firms’ innovation decisions once standard firm- and industry-level controls are included (Oudgou, 2021; Elshaarawy and Ezzat, 2023).

In addition to the IV approach, the analysis uses extended probit estimations and propensity score matching to further mitigate concerns related to self-selection into GVCs. While these methods do not fully eliminate endogeneity, the consistency of results across alternative estimation techniques provides additional confidence that the observed relationship between GVC participation and innovation is not driven solely by reverse causality or unobserved heterogeneity.

We report the results of the effect of GVC participation on innovation output in a threefold step. First, Table 2 presents the baseline results of the effect of GVC on product and process innovation. Second, by using the extended probit model and propensity score matching, we present the results of the robustness check of the baseline results in Tables 3 and 4. Third, to explore the heterogeneity of the results, we include the interaction between GVCs and some firm characteristics in Tables 5 and 6. These characteristics can be considered both as firm innovation capabilities and absorptive capacity.

Table 2 reports the baseline estimates of the relationship between global value chain (GVC) participation and firm-level innovation in South Africa. Columns (1) and (2) present marginal effects from probit estimates using alternative measures of GVC participation, while columns (3) and (4) report instrumental-variable (IV) results to account for potential endogeneity.

The marginal effects of probit estimates indicate a strong and positive association between GVC participation and innovation. Firms engaged in GVCs exhibit a significantly higher probability of introducing product and/or process innovations compared to nonparticipating firms. This result holds for both one-way GVC participation (GVC1) and two-way participation (GVC2), suggesting that integration into international production networks (whether through exporting, importing or both) is associated with greater innovative activity.

To address the endogeneity issue arising from reverse causality and omitted firm characteristics, columns (3) and (4) of Table 2 report IV estimates using external business environment constraints and industry–region GVC participation as instruments. The null hypothesis of the Durbin-Wu-Haussman endogeneity test is rejected in models (3) and (4), indicating the existence of endogeneity and the necessity of using the instrumental approach. The Sargan test statistic indicates that the hypothesis of instrument validity cannot be rejected. This test confirms that the chosen instruments are correlated with the endogenous variable (GVC), but not correlated with the residuals of the structural model (Innovation). Furthermore, the Stock-Yogo statistic allows us to reject the null hypothesis that the instruments are weak at a threshold of 5% (i.e. poorly correlated with the right-hand endogenous variable-GVC) ^[2]. These various tests demonstrate the relevance of the selected instruments (valid and powerful instruments) and thus allow us to isolate the impact of GVC on innovation in South African firms. The instrumental variable estimates confirm the positive effect of GVC participation on innovation.

Among the control variables, firm-level capabilities play an important role. The descriptive finding that larger firms exhibit higher rates of GVC participation but not necessarily higher unconditional innovation rates does not contradict the econometric results. Rather, it highlights the conditional nature of innovation in developing-country contexts. Larger firms may possess greater resources and capabilities, but these advantages translate into innovation primarily when combined with exposure to external knowledge and competitive pressures, such as those arising from GVC participation. In this sense, firm size alone is insufficient to generate innovation; it becomes effective when it enhances firms’ absorptive capacity to internalize and exploit learning opportunities embedded in global value chains.

Engagement in research and development (R&D), provision of employee training, foreign ownership and managerial experience are all positively and significantly associated with innovation outcomes. In contrast, firm age exhibits a negative and statistically significant relationship with innovation, indicating that younger firms may be more flexible or responsive to learning opportunities embedded in GVC participation. Firm size does not display a robust direct effect on innovation once other characteristics are controlled for.

We perform two robustness checks of our main results in the paper. First, we estimate the non-linear extended probit model in Table 3. Unlike the instrumental variable probit model that relies on strong distributional assumptions or two-step procedures, the extended probit model offers a flexible joint estimation of both equations (the structural-innovation and first-stage-GVC). This estimation of the correlation between unobservables influencing both equations produces more efficient and consistent parameter estimates under endogeneity (StataCorp., 2021). Second, to address the “self-selection” of superstar innovating firms in GVC participation, we use the common support propensity score matching (PSM) technique in Table 4. Indeed, innovating “superstar” firms are more capable of both innovating and dominating foreign markets compared to smaller, newer and more fragile counterparts. In this respect, innovating firms can be self-selected in GVC participation. We match observable firm innovation capabilities and absorptive capacity covariates, including firm size, R&D, ICT use, training and foreign capital participation.

Tables 4 and 5 (in the appendix) present PSM first-stage results and the balancing table. Figure A1 (in the Appendix) illustrates the biases (standardized mean differences) for five covariates (SIZE, WEB, TRAIN, RD and FDI) that were analyzed following propensity score matching (PSM) using the treatment variables GVC1 and GVC2. All covariates apart from size have absolute standardized biases very close to zero. This indicates that the matching effectively balanced those covariates between the treated and control groups. The standardized bias for covariate size is about 8%, which is still acceptable. A normalized bias of less than 10% (|normalized bias| < 10) is often referred to as acceptable (Rosenbaum and Rubin, 1985; Austin, 2009). Broadly, for most covariates, the matching process depending on GVC1 and GVC2 was effective in lowering selection bias. Although the standardized bias for size stays somewhat above the optimal level, generally balance is satisfactory, which supports the matching’s resilience.

Results from extended probit estimates confirm those obtained with 2SLS in terms of sign and significance of our interest variable (GVC). Furthermore, PSM shows a significant expected difference between treated and control groups using either GVC 1 or GVC 2 as treatments. Thus, across all specifications (IV2SLS and PSM), the estimated coefficients of GVC are positive and statistically significant, and the control variables preserve their signs and significance. Hence, the results are consistent with the baseline results. The consistency of results between the extended probit model, PSM, and our main results guarantees that our results are not motivated by functional form assumptions or sample imbalances and helps confidence in the causal interpretation of our results. Therefore, the positive link between GVC and innovation is not driven by potentially confounding omitted factors, such as market size, whereby greater access to foreign markets through international trade leads to more innovation (Aghion et al., 2018; Ito et al., 2023).

Although the baseline results show that global value chain (GVC) participation increases firm-level innovation, its benefits are not uniform across firms. Drawing on the absorptive capacity framework, we examine whether internal capabilities condition the innovation gains from GVC integration using 2SLS estimations with interaction terms (Tables 5 and 6).

For firms engaged in one-way trade (GVC1), the heterogeneity results in Table 5 strongly support the absorptive capacity mechanism. The interaction between GVC participation and R&D is positive and statistically significant, indicating that firms investing in R&D derive larger innovation gains from GVC exposure. The interaction with foreign ownership is also positive and significant, suggesting that foreign-owned firms benefit more from global integration, likely due to stronger international linkages. Moreover, the positive interaction between GVC1 and firm size indicates that larger firms are better positioned to translate GVC exposure into innovation, even though size alone does not robustly predict innovation. However, for two-way traders (GVC2), the evidence in Table 6 is more selective. The interaction with R&D remains positive and significant, while interactions with size and foreign ownership are less precisely estimated.

Overall, GVC participation enhances innovation primarily for firms with stronger absorptive capacity, particularly those engaged in R&D.

This study provides firm-level evidence that participation in global value chains (GVCs) is positively associated with innovation outcomes among manufacturing firms in South Africa. This finding is robust across baseline probit estimates, instrumental-variable specifications addressing endogeneity concerns and alternative robustness checks, including extended probit and propensity score matching. The consistency of these results supports the argument that GVC participation can act as a channel for firm-level learning and innovation in developing and middle-income economies.

These results are consistent with theoretical and empirical work emphasizing GVCs as conduits for international knowledge diffusion through learning-by-exporting, learning-by-importing and vertical linkages with foreign buyers and suppliers (Piermartini and Rubínová, 2021; Rodrik, 2018). Through exposure to international standards, production requirements and competitive pressures, firms engaged in GVCs may be induced to adapt production processes or introduce new products, even when operating in environments characterized by weak domestic innovation systems.

However, the relatively low overall incidence of innovation observed in the data suggests that the innovation effects associated with GVC participation in South Africa are likely to be incremental rather than radical. This interpretation is consistent with evidence from developing countries showing that GVC-related innovation often takes the form of process improvements or “new-to-the-firm” innovations rather than frontier technological breakthroughs (Cirera and Muzi, 2020).

The findings align with a growing empirical literature documenting a positive relationship between GVC participation and firm-level innovation in developing economies. Cross-country studies show that deeper integration into GVCs (particularly through imported intermediate inputs and export linkages) enhances firms’ innovation performance by facilitating access to embodied knowledge and foreign technologies (Piermartini and Rubínová, 2021; Tajoli and Felice, 2018).

Placing these findings in a broader African context, the results are broadly consistent with firm-level evidence showing that participation in global value chains is associated with a higher propensity to innovate among manufacturing firms in Sub-Saharan Africa, albeit with substantial heterogeneity across countries and firms (Ndubuisi et al., 2025). Previous studies using World Bank Enterprise Survey data document positive but uneven innovation effects of GVC participation in African economies, often emphasizing the role of firm capabilities, absorptive capacity and local conditions in shaping outcomes (Calatayud and Rochina-Barrachina, 2023; Pietrobelli and Rabellotti, 2011). The South African evidence presented in this study aligns with this literature, while also suggesting that relatively stronger industrial capabilities and institutional frameworks may allow firms to better leverage GVC participation for innovation compared to firms operating in less diversified African manufacturing contexts. At the same time, the modest overall innovation rates observed in the data underscore that GVC participation alone is insufficient to generate widespread innovation, a pattern frequently highlighted in the African upgrading literature (Pietrobelli and Rabellotti, 2011; Taglioni and Winkler, 2016).

Moreover, the South African results help contextualize mixed findings in the African literature, where GVC participation has not always translated into sustained upgrading or innovation. Studies emphasizing “lock-in” risks argue that firms integrated into low-value-added segments of GVCs may face limited opportunities for learning and capability accumulation (Andreoni and Tregenna, 2020). The positive effects observed in this study suggest that South Africa’s relatively more diversified industrial structure and institutional environment may enable firms to extract greater innovation benefits from GVC participation than firms operating in less developed industrial contexts.

A key contribution of this study lies in highlighting the conditional nature of the GVC-innovation relationship. The heterogeneity analysis shows that the innovation benefits of GVC participation are stronger for firms with higher absorptive capacity, proxied by firm size, engagement in R&D activities and foreign ownership. This finding is consistent with theoretical frameworks emphasizing that external knowledge flows generate innovation only when firms possess sufficient internal capabilities to absorb, adapt and apply that knowledge (Rodrik, 2018).

Empirical evidence supports this view. Eissa and Zaki (2025) show that GVC participation can substitute for formal R&D in stimulating innovation among firms, but only in contexts where firms possess a minimum level of organizational and human capital. Similarly, Ndubuisi et al. (2025) demonstrate that firm characteristics such as size and experience shape the extent to which African firms benefit from GVC-induced knowledge spillovers.

These results also help explain the apparent discrepancy between descriptive patterns and econometric findings in the data, whereby larger firms are more likely to participate in GVCs but do not necessarily exhibit higher unconditional innovation rates. Innovation gains appear to materialize when GVC participation interacts with firm-level capabilities, rather than through size alone.

The findings refine and extend global value chain (GVC) theories of innovation by highlighting the critical interplay between external knowledge linkages and internal firm capabilities. Classic GVC frameworks posit that participation in international production networks can facilitate learning and upgrading for developing-country firms by exposing them to foreign technologies, standards and competitive pressures (Gereffi et al., 2005; Taglioni and Winkler, 2016). Evidence from South Africa supports this argument, showing that GVC integration is associated with higher innovation outcomes and confirming that global production networks can provide important learning opportunities for firms in late-industrializing economies (Pietrobelli and Rabellotti, 2011).

At the same time, the results add nuance by demonstrating that not all firms benefit equally from GVC participation. Consistent with Gereffi et al. (2005), who emphasize supplier competence as a key determinant of knowledge transfer within value chains, the findings indicate that only firms with sufficient absorptive capacity are able to fully capitalize on GVC-facilitated learning. Larger firms, firms investing in research and development and firms with foreign ownership exhibit stronger innovation gains from GVC participation. This pattern aligns with Cohen and Levinthal’s (1990) concept of absorptive capacity, defined as a firm’s ability to recognize, assimilate and apply external knowledge for commercial purposes.

These insights bridge GVC theory with perspectives from the national innovation systems literature. Pietrobelli and Rabellotti (2011) argue that developing-country firms must leverage global knowledge pipelines while simultaneously building local capabilities, since the central challenge lies in absorbing technologies produced elsewhere and transforming them into innovation. The present study provides empirical support for this view by showing that GVC participation is a necessary but not sufficient condition for innovation. This finding extends existing theory by underscoring the co-evolution of global linkages and firm capabilities and by highlighting the importance of firm-level heterogeneity and institutional context in shaping innovation outcomes (Taglioni and Winkler, 2016; Andreoni and Tregenna, 2020).

The findings of this study carry important implications for industrial and innovation policy in developing economies, particularly where innovation performance is weak and integration into global value chains (GVCs) is uneven. At the macro level, the results highlight the need to create an enabling environment that facilitates firms’ participation in GVCs while strengthening their capacity to absorb external knowledge. Policies that reduce trade frictions (through better logistics, more efficient customs and transparent trade regulations) can increase firms’ exposure to international markets and learning opportunities.

However, participation in GVCs alone is unlikely to generate innovation without supportive domestic conditions. Effective GVC-led innovation requires complementary measures that improve regulatory certainty, strengthen skills formation and reinforce national innovation systems. Public investment in education, technical and vocational training and applied research infrastructure helps firms internalize and adapt knowledge encountered through global linkages. Active industrial policies are therefore important for building absorptive capacity. Support for firm-level R&D, technology adoption and workforce training can enhance firms’ ability to translate exposure to foreign technologies and standards into concrete product and process improvements.

At the micro level, firms should treat GVC participation not only as market access but also as a deliberate learning strategy. Investing in employee skills, adopting quality management systems and engaging in collaborative innovation can strengthen internal learning routines. Partnerships with multinational enterprises and participation in supplier development programs can further facilitate access to advanced technologies and managerial practices.

Finally, targeted support for small and medium-sized enterprises is essential to ensure inclusive innovation gains. Innovation vouchers, technical assistance, business development services and improved access to finance can help smaller firms overcome capability constraints and leverage global integration as a pathway to innovation-driven growth.

Several limitations of this study should be acknowledged. First, the analysis relies on cross-sectional data from the 2020 World Bank Enterprise Survey, which coincides with the COVID-19 pandemic. While this data set remains the most recent and comprehensive source of firm-level data for South Africa, pandemic-related disruptions may have affected firms’ innovation and trade behavior. Future research using panel data or post-pandemic surveys would be valuable to assess the persistence of the observed relationships.

Second, innovation is measured using binary indicators of product and process innovation, which may not fully capture the intensity or quality of innovative activities. Prior research has shown that such measures can obscure important differences between incremental and more substantive innovations, particularly in developing-country contexts (Cirera and Muzi, 2020). Future work could incorporate richer innovation metrics to deepen understanding of GVC-related upgrading.

Finally, comparative analyses across African countries with differing levels of industrial development would provide further insight into how institutional and structural factors mediate the innovation effects of GVC participation.

This paper examines whether participation in global value chains (GVCs) contributes to firm-level innovation in South Africa, a country characterized by relatively advanced industrial capabilities within the African context but persistently low innovation performance. Using firm-level data from the 2020 World Bank Enterprise Surveys, the analysis investigates the relationship between GVC participation and innovation outcomes while accounting for firm heterogeneity and potential endogeneity.

Using an instrumental-variable framework complemented by extended probit estimations and propensity score matching, the results consistently show that participation in global value chains is positively associated with firms’ likelihood of introducing product and process innovations. These findings suggest that integration into international production networks can support innovation by exposing firms to external knowledge, competitive pressure and upgrading requirements. Importantly, the analysis reveals that these innovation gains are not uniform across firms. The positive effect of GVC participation is stronger for larger firms and for firms that invest in research and development or attract foreign direct investment, highlighting the critical role of absorptive capacity in translating GVC exposure into innovation outcomes.

Taken together, the findings underscore that GVC participation alone is insufficient to generate widespread innovation. Rather, innovation benefits from GVC integration are conditional on firms’ internal capabilities and their ability to absorb and exploit external knowledge. This insight contributes to the literature by providing country-specific evidence from South Africa and by emphasizing the importance of firm heterogeneity in shaping the innovation effects of global integration.

From a policy perspective, the results point to several implications for fostering innovation through GVC participation in South Africa. First, policies aimed at facilitating firms’ integration into global value chains (such as improving tax administration, streamlining customs procedures and reducing trade-related regulatory barriers) can enhance firms’ exposure to international markets and learning opportunities. Second, improving access to external finance is crucial, particularly for smaller firms that face financial constraints limiting their ability to innovate and participate effectively in GVCs. Third, strengthening firms’ absorptive capacity is essential to maximize the innovation gains from GVC participation. This can be achieved by expanding incentives for research and development, supporting skills development and promoting foreign direct investment that facilitates knowledge transfer.

Overall, the paper highlights the potential of GVC participation as a channel for fostering firm-level innovation in South Africa, while emphasizing that such benefits depend on complementary policies that enhance firms’ capabilities. Future research could build on these findings by exploiting panel data or post-pandemic surveys to assess the persistence of GVC–innovation linkages over time and to further explore the dynamics of learning and upgrading in African manufacturing firms.

The authors would like to thank the Editor and the anonymous reviewers for their careful reading of the manuscript and for their constructive and insightful comments. The authors are also grateful to the participants of the Globelics International Conference 2025, held from 24-26 November 2025 in Pretoria, South Africa, for their valuable feedback and engaging discussions. Thanks to the DSI/NRF/Newton Fund Trilateral Chair in Transformative Innovation, the 4IR and Sustainable Development for its support.

All data generated or analyzed during this study are not included in this submission but can be found on the website of World Bank Enterprise Survey database.