A Bird in the hand is worth two in the bush: How SMEs leverage different types of knowledge and networks during effectual internationalization

Internationalization literature uncritically treats different networks and knowledge as having the same impact on strategy and performance. Recognizing the need to approach these concepts in a more sophisticated way and inspired by Effectuation ’ s Bird-in-hand principle, we posit that internationalization knowledge, customer network knowledge, and supplier network knowledge influence performance while mediated by a non-predictive strategy. We combine six hypotheses in a structural model that we test on a sample of 851 SMEs from Brazil, China, Italy, Poland, and Sweden. The analysis supports the idea that non-predictive strategy mediates positive relationships between internationalization knowledge and supplier network knowledge, on the one side, and performance, on the other. The indirect relationship involving internationalization knowledge is stronger for SMEs in developed countries, but supplier network knowledge does not differ across groups. The analysis does not support the importance of customer network knowledge. We discuss theoretical and managerial implications of these findings.


Introduction
Despite the role of business plans in business practice, firms often adopt a non-predictive strategy (NPS) by emphasizing what can be done in the face of uncertainty rather than focusing on goals and plans (Wiltbank et al., 2006).NPS allows small and medium-sized enterprises (SMEs) to leverage limited resources into international opportunities (Cowden et al., 2024;Tolstoy et al., 2021;Vasconcellos et al., 2019).It encourages entrepreneurial behavior in new market entries and facilitates rapid foreign expansion (Chetty et al., 2024b;Gabrielsson et al., 2022), with relevant performance outcomes for the internationalizing firm (Bai et al., 2022).NPS embodies what Sarasvathy et al. (2014) label Lemonade Contingencies and Pilot-in-the-plane principle in Effectuation theory: it acknowledges that unforeseen events can reveal opportunities rather than obstacles (Lemonade contingencies) and asserts that firms can shape their future by concentrating on things under their control (Pilot-in-the-plane).
Another crucial aspect of Effectuation applicable to internationalization studies is the Bird-in-hand principle (Karami et al., 2020;Prashantham et al., 2019), whith outlines available means that firms can transform into opportunities.It addresses critical questions like Who I am, Whom do I know, and What do I know, thereby capturing knowledge and network elements that are pivotal in the success of foreign expansion (Johanson & Vahlne, 2009).Detailed examinations of the Bird-inhand principle in international entrepreneurship (IE) studies remain scant, with the few recent efforts limited to qualitative approaches or missing an analysis of performance implications (e.g., González & Massieu, 2021;Lassalle et al., 2020).Current scholarship mostly offers conceptual discussions (e.g., Prashantham et al., 2019;Sarasvathy et al., 2014) and general views on knowledge's role in internationalization, overlooking the nuanced impact of specific knowledge types on this process (Stoian et al., 2018).Addressing this topic is critical since the Bird-in-hand is the "taken-for-granted starting point in the effectuation literature" (Alsos et al., 2020, p. 611) and provides strategies for managing constraints, risks and costs under uncertainty (Lassalle et al., 2020).A deeper understanding could therefore clarify the dynamics of opportunity development in both entrepreneurial and international venture contexts.
Although the contingency effects of knowledge and the network of stakeholders are often threated as general concepts, internationalization literature has over the years indicated that knowledge and networks are neither general nor homogeneous but rather sophisticated concepts (Bai & Johanson, 2018;Bai et al., 2021;Chandra & Wilkinson, 2017;Stoian et al., 2018).Emerging in various contexts, they may have different influence on firm strategy and performance.We maintain that this is a major gap in the internationalization literature, which we address in this study.Network has been previously divided into business and social network where the first mentioned is based on relationships between firms and organizations and the second on relations between individuals (Agndal & Axelsson, 2002;Slotte-Kock & Coviello, 2010).We extend these attempts to apply a more sophisticated perspective by dividing the knowledge emerging from two other types of networks, namely the one being developed with the firm's suppliers and the other one evolved through business relationships with customers.In addition, a second gap in the literature is that we know little about how these various contingencies influence strategy and performance in international markets.Thus, at the same time as knowledge and network are key concepts in the understanding of internationalization, the literature offers limited explanation based on a sophisticated approach of the concepts.This study contributes to the internationalization literature by bridging these gaps.
We argue that the knowledge possessed by the firm can be of various types and emerge based on whom the firm knows, demonstrating different importance and applications.In this sense, this study investigates how knowledge about suppliers and customers affect performance.Our approach differs from most network studies by not treating all network partners equally, but considering their specific roles in the network.This distinction is crucial for recognizing the varied knowledge assets firms can acquire through their networks.Considering that the firm does not exist in isolation from the market, and acknowledging SMEs' heavy reliance on home government support and the concentration of their assets in the domestic market (Paul et al., 2017), we investigate the research question: How do supplier and customer-related knowledge affect the international market performance of emerging and developed country SMEs?
Based on survey data on 851 SMEs from Brazil, China, Poland, Italy and Sweden, our conceptual model compares the international performance of non-predictive decisions based on different types of knowledge and under distinct home institutional conditions.Our results show that emerging and developed market SMEs seem to differ in how effectively they can use internationalization knowledge (IK), but no such differences seem to affect their use of supplier network knowledge (SNK).Customer network knowledge (CNK) seems to be the least relevant knowledge type for the international market performance of our sampled SMEs.
Our study contributes to bringing the Bird-in-hand principle more formally into the scope of effectuation principles investigated within international entrepreneurship, exploring its performance implications through the effects of different knowledge types.Most internationalization studies applying effectuation lenses tend to focus on the resource scarcity characterizing SMEs at the expense of complementary resources accessible through their networks (Karami et al., 2020).That can be understood as a reflex of a logic that takes available means as the starting point of the effectual process (Alsos et al., 2020) instead of concentrating on them in the first place.When discussing the role of knowledge in these firms' foreign expansion, IE studies tend to consider knowledge as opportunity material in a general sense, without delving into its potentially different types and applications (Stoian et al., 2018).In this sense, our study also add to a knowledge-based view of SME internationalization (Casillas et al., 2009;Prashantham, 2005;Stoian et al., 2018), as SMEs seem to perform and be affected differently when using different forms of knowledge.Finally, our study offers a contextualized view of decision-making (Elbanna et al., 2020) by highlighting the relevance of home-country effects.This appears in our findings and discussion about how SMEs coming from emerging and developed countries leverage different types of knowledge, revealing similarities (like with SNK) and differences (with IK).
After this introduction, we present our theoretical background and hypotheses.Methodological issues come in the sequence, followed by our results and the discussion section.We conclude by addressing implications and limitations of this study, together with future research directions.

The non-predictive strategy of SMEs
Various theories embody the concept of non-predictive strategy (NPS) by highlighting the unstructured nature of decision-making, as well as the importance of creativity and flexibility in transforming available means into opportunities amid ambiguity and uncertainty (Johanson & Oliveira, 2024).For example, Hayes' (1985) concept of Reverse planning suggests that organizations identify potential objectives by exploring alternatives with their available resources.Brown and Eisenhardt's (1997) Continuous change posits that firms can achieve success by granting freedom for experimentation and improvisation.An earlier model, the Garbage Can (Cohen et al., 1972), align with this perspective by indicating that decision-makers only comprehend what they are doing in the process of doing it.
A recent addition to this lineage is Sarasvathy's (2001) Effectuation, a theory originally introduced as an explanation of entrepreneurial activity in contrast with the planning-based approach derived from the work of strategy scholars like Ansoff (1979) and Porter (1980).With the growing interest in effectuation within International Entrepreneurship, the literature has explored the nature of various heuristics associated with effectuation (Deligianni, 2023;Tolstoy et al., 2021;Vuorio & Torkkeli, 2023) and debated whether they must all be applied simultaneously (Kerr & Coviello, 2019;Welter et al., 2016).Nonetheless, it is often the non-predictive nature of effectuation, rather than its specific heuristics, that renders it particularly beneficial for firms facing uncertainties such as those encountered during internationalization (Bai et al., 2021;Karami et al., 2020).Considering non-prediction as a fundamental principle of organizational behavior with extensive theoretical relevance, we concentrate on it instead of developing a comprehensice analysis of effectuation theory.
Non-predictive strategies assume strategic flexibility to change action paths and defend a set of heuristics that guarantee firms a focus on what they can do given their resources (Conz et al., 2023).Such heuristics include short-term experiments, allowing firms to explore different decision routes with low resource commitment (Chandler et al., 2011).They may also encompass improvisation efforts (Alsos et al., 2020), which deliberately build new resource bases instead of reorienting or reinterpreting pre-existing reference models (Crick et al., 2023).By improvising and experimenting, firms are more likely to face unanticipated paths and expand their networks in an unstructured way, which potentially give them access to new resources and opportunities.
In line with the extant literature, the concept of Non-predictive Strategy (NPS) entails three main elements.The first is flexibility, which implies deviation from directions, plans, and formalized strategies, in order to develop unpredicted opportunities while leveraging preexistent resources (Mauer et al., 2018;Vuorio & Torkkeli, 2023).NPS also incorporates creativity and purposeful creation of alternatives, which implies thinking with the feet and going out of the box during the internationalization (Scazziota et al., 2023;Vasconcellos et al., 2019).Finally, as internationalization is a process over time and unexpected opportunities and problems are likely to emerge, quickly finding solutions and acting on opportunities are necessary attitudes.From this follows that improvisation, where plans and execution co-exist in time and take place simultaneously (Alsos et al., 2020;Chetty Gabrielsson et al, 2024).
We contend that NPS influences performance positively in at least three ways.First, it enhances learning as it is capitalizes on a process of learning by doing and is related to feedback dynamics focused on improving how decisions are made (Deligianni, 2023;Khurana et al., 2022).Second, it implies flexible, fast, and cheap problem-solving, which has a positive effect on outcomes and market performance (Racat et al., 2023;Wu et al., 2020).Third, as NPS welcomes doing things in new ways, using existing resources for new purposes and finding new solutions to old problems, the firm has to be creative and think "outside the box" (Ko et al., 2022).Novel experiences may then result in innovations that increase efficiency or produce more value than previous solutions (Williams et al., 2021).

Knowledge as a means for firm internationalization
The Bird-in-hand principle emphasizes leveraging existing and accessible resources as opportunity material (Karami et al., 2020).In the context of firm internationalization, a significant portion of these resources pertains to knowledge, which is among the most critical assets SMEs can use to create opportunities abroad (Casillas et al., 2009;Prashantham, 2005;Stoian et al., 2018).Both Johanson and Vahlne (2009) and Sarasvathy et al. (2014) highlight the importance of the knowledge a firm gathers about its network for shaping its strategic decisions.This knowledge influences the firm's entry strategy and guides future opportunity-seeking efforts, enabling firms to reduce uncertainty in their search for opportunities (Hilmersson & Jansson, 2012;Hohenthal et al., 2003).In line with Johanson and Johanson (2021), the extent of a firm's embeddedness is in the network-that is, the more indirect and inter-connected relationships the firm has-enhances its knowledge of the products and technologies circulating within the network, as well as of the specific characteristics of the market network and its participants.
Business network knowledge conveys not only the firm's knowledge itself, but also how the network is structured (Hohenthal et al., 2014;Magni et al., 2022).That happens as knowledge is a result of cumulative processes that take time (Blomstermo et al., 2004;Mejri et al., 2018) and emerges from the interaction between the firm and its customers and suppliers (Bullinger et al., 2004;Michaelides et al., 2013).Moreover, it depends on the nature of the interactions and exchanges between the firm and these actors (Anderson et al., 1994).At the core of this process is experience, gained through various activities conducted with specific customers and suppliers.Knowledge flows not only between firms that are directly connected in the network but also from those outside these direct conncetions (Hohenthal et al., 2014;Sammarra & Biggiero, 2008), facilitating a deep understanding among firms.This comprehension enables firms to use this information to craft their strategies in international markets.Additionally, the knowledge gained from specific subsets of a firm's business relationships can be leveraged to refine strategies pertinent to those relationships (Eng, 2005).
For example, SNK covers their products and services, technologies, capabilities, constraints, past performance, and strategic initiatives, as well as the analogous knowledge about the suppliers' own suppliers (Esper et al., 2010;Furr & Snow, 2015).Conversely, CNK includes understanding customers' requirements, preferences, and perceptions of the available value propositions, in addition to market trends and opportunities for the firm's products and services (Magni et al., 2022).These knowledge pools can be leveraged in strategies that are specifically aimed at the supply or the customer side of the network (Marcos-Cuevas et al., 2016;Priem & Swink, 2012;Sheth, 2020;Storbacka et al., 2016).Therefore, we argue that a firm's network of customers and customers' customers, as well as its network of suppliers and suppliers' suppliers, can have distinct properties and serve different roles for the firm in terms of their respective knowledge dynamics.
In addition to SNK and CNK, which are tailored to specific network segments and thus represent manifestations of market-specific knowledge (Makino & Delios, 1996), existing literature also highlights the importance of IK for firms' foreign expansion (Adomako et al., 2022;Stoian et al., 2018).This form of knowledge steems from the codification of context-specific knowledge, derived from doing business in particular network relationships, into procedures and routines that make it applicable in other market settings (Wiklund & Shepherd, 2003).Such a process originates IK, which covers a broader spectrum of general considerations including insights on customers, suppliers, cultural practices, institutional frameworks, etc (Eriksson et al., 1997;Fletcher et al., 2013).Because of that, it plays a crucial role in navigating the intricacies of global expansion, offering firms a comprehensive understanding essential for successful international ventures.
IK is associated with the development, implementation, and management of international operations in general (Eriksson et al., 1997) and is usually accumulated by processing procedural know-how or learning from consultants and advisors (Fletcher et al., 2013;Mejri et al., 2018).It concerns knowledge about how to adapt products and services for foreign markets, as well as how to adapt the organization to international operations (Adomako et al., 2022).Therefore, IK is more general and transferable across opportunities in the network.While SNK and CNK reveal to the firm specific views of the networked platform from which they internationalize, IK feed them with general routines and procedures about how to act abroad (Stoian et al., 2018).

Country as context and the role of institutions
Given the concentration of SMEs' operations and assets domestically (Paul et al., 2017), it is natural for these firms' strategies and resource management to be significantly influenced by their home market environments.Specifically, the institutions prevailing in the home market set the "rules of the game" that affect the efficiency of these firms' decisions and activities (North, 1990).Institutions in developed markets typically provide stability and reduce the costs of economic activity through better resource and information availability and greater security for contracts.In contrast, emerging markets often face challenges due to ineffective or absent institutional support, manifesting excessive bureaucracy, legal barriers, financial constraints, a shortage of skilled labor, and competition issues (Khanna et al., 2005).To navigate these institutional voids (Khanna & Palepu, 1997), companies often need to internalize such inefficiencies (Doh et al., 2017) or engage in informal practices centered on trust and relationship-building to secure better operating conditions (Puffer et al., 2010).Moreover, the speed and unpredictable nature of regulatory changes and crises in emerging markets introduce significant contextual uncertainty and affect managers' ability to meet market requirements and commit to long-term investments (Nicholls-Nixon et al., 2011).
Firms from emerging and developed markets are likely to vary in their capacity to leverage their knowledge assets for decision-making.The distinct characteristics of these environments suggest that firms in emerging market face particular challenges (Rottig, 2016).In emerging markets, the support needed for local firms to develop advanced capabilities or translate these capabilities into performance outcomes is often lacking due to the absence of a well-regulated and politically-stable business environment (Wu & Chen, 2014).Addititonally, access to opportunities and financial resources necessary for deploying assets may be hampered by the interference of public actors and interest groups, making room for lobbying and corruption (Cuervo-Cazurra, 2016).Furthermore, the lack of effective legal protection for knowledgeintensive initiatives exposes companies in emerging market to opportunistic behavior and unfair competition (Amankwah-Amoah et al., 2022).Consequently, these firms may need extra efforts to safeguard their operations against market inefficiencies or risk being overtaken by stronger rivals.
Despite its significance, research on the effects of home country on internationalization remains limited (Cuervo-Cazurra et al., 2018).It is only recently that scholars have begun to integrate these effects in their analyses of foreign expansion (Becker-Ritterspach et al., 2022).Traditionally, research has recognized the role of the home country in providing firms with location-specific assets such as natural resources, a local workforce, and incentives (Rugman, 1981).It has also noted that firms from emerging markets may encounter prejudice and barriers abroad due to their origin (Cuervo-Cazurra et al., 2007;Madhok & Keyhani, 2012).Most studies that consider home country institutions have focused on the concept of distance faced by firms when entering foreign markets, including those stemming from cultural differences (Kogut & Singh, 1988) and differing institutional profiles (Kostova, 1999).This emphasis on distance has hindered a thorough exploration of the specific effects of home market contexts and their impact on SME internationalization.

Hypotheses development
Our model hypothesizes that NPS mediates relationships between SMEs' knowledge assets and international market performance (Fig. 1).Assuming that firms can combine different decision logics during internationalization (Karami et al., 2020), the model also admits performance effects of knowledge not mediated by non-prediction.

The mediated performance of SMEs' knowledge
IK equips firms with the resources necessary to craft various internationalization strategies and navigate different market contexts effectivelly (Fletcher et al., 2013).This includes procedures and routines applicable that are applicable and beneficial across multiple markets.This utility extends to SMEs, for whom IK can aid designing management structures and business models tailored for international operations, organizing distribution systems, and fostering business relationships with the networks of foreign markets (Fletcher & Harris, 2012).As firms often gravitate towards non-predictive strategies (NPS) due to the inherent uncertainties of entering foreign markets (Karami et al., 2020), it is worth considering that this strategy is particularly suited to capitalize on the possibilities enabled by IK.
NPS offers the flexibility and creativity needed to embrace the broad possibilities that general-purpose knowledge types like IK can provide (Mauer et al., 2018;Vasconcellos et al., 2019).The high transferability of IK across market contexts means that it represents a rich source for alternative action paths to be identified and leveraged with the adoption of non-prediction.Decision-making frameworks that neglect such opportunities might fail to fully realize IK's potential in aiding SMEs' international expansion.In that vein, NPS embodies a readiness to explore new and unforeseen alternatives (Wiltbank et al., 2006), enabling firms to move beyond outdated structures and business models towards innovative ones they develop.Furthermore, non-predictive strategies encourage SMEs to first assess the resources at their disposal for generating new market opportunities (Hayes, 1985;Sarasvathy, 2001).This approach is crucial when leveraging IK, which, being more comprehensive, presents a wider range of options than other knowledge forms (Hilmersson & Johanson, 2016).
SMEs that opt for decision-making approaches other than nonpredictive strategies may not fully leverage their IK, applying it in strategies that struggle to withstand the uncertainties of international environments or only acting on opportunities in a reactive manner.Given SMEs' inherent resource limitations, a failed strategy in such volatile scenarios could, in the worst-case scenario, threaten the survival of these businesses (Ropega, 2011).By embracing NPS for decisionmaking, firms can harness improvisation and experimentation as means to uncover the hidden advantages of IK, presenting themselves in the form of unexpected opportunities and serendipitous discoveries Fig. 1.Theoretical Model.(Crick & Spence, 2005;Kalinic et al., 2014).These innovative actions are crucial in helping firms avoid the pitfalls of path dependence, which can confine them to outdated strategic paths due to past decisions and institutionalized practices (Sydow et al., 2009).This underscores the role of NPS in mediating SMEs' capacity to achieve success in international markets by effectively utilizing their IK assets.Hence, we hypothesize: H1. SMEs' level of non-predictive strategy adoption mediates a positive relationship between internationalization knowledge and international market performance.Supplier Network Knowledge provides firms with valuable insights into the resources available through their supplier networks, enabling SMEs to expand their resource base and generate new opportunities (La Rocca et al., 2019).SNK facilitates the deployment of flexible strategies in uncertain environments (Wathne & Heide, 2004) and identifies resources that can contribute to innovation efforts (Lee & Schmidt, 2017).This knowledge configures for the firm an insidership position in the network, as a result of having built multiple relationships and thereby accumulating knowledge which is specific for this network and difficult to apply in other networks (Forsgren, 2016).Because the firms' cooperation with network partners to a large extent contain trust, commitment and interdependence, firms can be expected to adopt a nonpredictive strategy where they have strong SNK.
The application of non-predictive strategies (NPS) when leveraging SNK with supplier networks can benefit SMEs for two main reasons: it allows firms to effectively mobilize their supplier networks and supports the use of non-predictive reasoning in innovation processes.While SMEs can engage with suppliers and suppliers' suppliers in different ways, employing NPS sets the stage for collaboration and opportunity creation (Storbacka et al., 2016).That comes with the understanding that, from a supply chain perspective, the network of suppliers represents means available for the creation of goods and services that will later on be offered to markets (Qrunfleh & Tarafdar, 2013).Collaboration with suppliers towards opportunity creation often translates into co-creation efforts, whose outcomes include products and services that are not bounded by strategic plans (Priem & Swink, 2012).Reactive adaptation would probably not do better, given that relying on delayed information flows can easily disrupt relationships with the suppliers or the suppliers' suppliers (Lee et al., 1997).In this sense, NPS seems able to help SMEs leverage SNK to tap on their network of suppliers for the creation of new value on an ongoing basis.
Besides, NPS can be expected to be useful when SMEs use SNK specifically to implement formal innovation initiatives involving their supplier networks.According to Brown and Eisenhardt (1997), neither planning nor reacting is effective in delivering successful multi-product innovation.They stress rather the need for initiatives that promote innovation in a continuous and intentional way, leveraging experiments and strategic alliances.Other scholars similarly identified the importance of improvising and experimenting with contingencies to promote the continuous efforts that may result in successful product innovations (e.g., Orlikowski, 1996;Weick & Quinn, 1999).These are the kind of heuristics that NPS offers firms.Altogether, when SMEs exploit their SNK by engaging with their supplier network in regular operations or in the context of formal innovation initiatives to serve the international market, positive results should come from the use of NPS.Hence:

H2. The level of non-predictive strategy adoption mediates a positive relationship between the levels of supplier network knowledge and international market performance.
One could expect that NPS should also help SMEs find performance outcomes when exploiting customer network knowledge (CNK).Accordingly, both buyers and final customers may represent pools of network resources that those firms can activate on a continuous basis or mobilize through specific initiatives (Chetty & Blankenburg Holm, 2000).Prior research has explored the value that customer commitment can offer firms in the context of open innovation and co-creation (Marcos-Cuevas et al., 2016;Sheth, 2020).Although such initiatives seem to require some structure, like reward programs and knowledgesharing platforms (e.g., Foss et al., 2011), their outcomes are usually exploratory and unpredictable and may require equally non-predictive decision approaches.Hence, non-predictive decision-making should positively impact the firms' performance when they combine resources with and leverage customers' resources and ideas.
However, simply assuming that customers represent network resources to be leveraged, like with suppliers, does not properly distinguish the role of these groups in the supply chain.While suppliers effectively make resources available to firms for the creation of goods and services, making up the cost structure of the business, relationships with customers help mainly in understanding the upside of opportunities with their associated market potential, purchasing behavior, and revenue streams (Ryals & Payne, 2001).Thus, knowledge acquired about customers in the market is likely to make firms turn to cost-return analyses and structured decision-making at the expense of nonprediction.Decisions based on this knowledge are, at least, evaluated against expectation of relationship continuity (Lee et al., 2008), where current investments are implicitly or explicitly assumed to pay off over time.
In this sense, the focus of CNK on customers and market trends may lead one to believe that such a knowledge makes firms less likely to employ NPS.This happens because market-related knowledge tends to reduce uncertainty in internationalization (Alvarez et al., 2018;Johanson & Vahlne, 2009), minimizing firms' need for creativity and out-of-box thinking.Moreover, CNK is a result of interaction with the customer network, where joint planning and prediction is essential for distribution, payment, product improvements, spare parts, etc.It also tends to center firms' attention on building competitive advantages and creating value for customers' needs (Priem et al., 2018).Customerdriven strategies may additionally compromise flexibility, given that customer-related investments tend to be more locational specific than those related to suppliers (Autio & Zander, 2016).These effects should be particularly salient for SMEs, whose resource constraints may lead them to focus on safer, existing opportunities at the expense of others yet to be created.Moreover, if perceived as strategic, CNK may foster competition over existing market opportunities instead of collaboration and co-creation (Ma, 2004).
The resulting scenario suggests that high levels of CNK may reduce the adoption of NPS by SMEs, which nonetheless seems to be an important driver of international market performance.The combined outcome is a negative indirect effect of CNK on international market performance.Since "negative" mediations are rarely examined but accept the same procedures used with "conventional" mediations (Cheung & Lau, 2008), we hypothesize: H3.Customer network knowledge has a negative indirect effect on international market performance, through a negative effect on the level of non-predictive strategy adoption.

Home country effects on the non-predictive internationalization
There is no obvious reason to believe that the way SMEs employ knowledge differs according to their home market.Once acquired, IK represents generic and procedural knowledge that any firm can in principle use for strategizing.The roles of SNK and CNK, in turn, seem to depend rather on the structural relevance of the business network for firms' strategies: SNK offers knowledge about the supplier network and what can be done with it; CNK offers knowledge about the customer network and potential market opportunities.However, the performance outcomes of strategies derived from these knowledge stocks may be likely to differ between SMEs from emerging and developed countries.
Developed market SMEs experience more mature and complete institutions at home, meaning that they face simpler obstacles to operate abroad than emerging market firms (Tesfom & Lutz, 2006).Developed contexts feature more sophisticated product, labor, and capital markets, including a superior promotion of private sector development and more confidence in the enforcement of contracts and property rights (Khanna et al., 2005;Narula, 2012).Limitations in emerging markets' institutions, in turn, create room for inefficiencies, bureaucracy, corruption, and other symptoms of weak institutional configurations.In these countries, such obstacles may require specific investments that can be particularly challenging for resource-constrained SMEs (Doh et al., 2017).These firms see their potential profits consumed by the mentioned investments or have to accept being exposed to the risks and liabilities of operating in ineffective markets without protecting their assets.
Countries with poorly-developed institutions may limit local firms in their capability development (Narula, 2012).This means that emerging market SMEs may still progress in their internationalization efforts when processing knowledge through NPS, but liabilities at home may compromise their capacity of developing functional operations and serving foreign customers properly.Additionally, emerging market firms may find problems in profiting from the products and services they sell abroad.Such firms are likely to need additional efforts for disguising their liabilities, or accept lower prices while overcompensating liabilities with investments in quality (Cuervo-Cazurra et al., 2018).Considering these arguments, one can expect that the performance effects of non-predictive strategies will be stronger for developed market SMEs than for their counterparts in emerging markets.
If the performance derived from decisions that follow NPS is conditional on the development level of the SMEs' home country, this should in principle be valid regardless the knowledge that is leveraged in this process-be it IK, SNK, or CNK.As put by Preacher, Rucker, and Hayes (2007), any systematic variation in indirect effects that is conditional on another variable can be called a "moderated mediation".In this scenario, home country institutional development seems to be a likely moderator for the indirect relationships described in the previous hypotheses.Hence, we consider: H4a.The indirect relationship between internationalization knowledge and international market performance, mediated by non-predictive strategy, is stronger for developed market SMEs than emerging market SMEs.H4b.The indirect relationship between supplier network knowledge and international market performance, mediated by non-predictive strategy, is stronger for developed market SMEs than emerging market SMEs.H4c.The indirect relationship between customer network knowledge and international market performance, mediated by non-predictive strategy, is stronger for developed market SMEs than emerging market SMEs.

Empirical context
We assess our model with data collected in a survey of manufacturing SMEs in Brazil, China, Italy, Poland, and Sweden.This selection facilitates a comprehensive examination of differing market dynamics and provides valuable insights into the varying challenges and opportunities faced by businesses within the economic environments of emerging and developed markets.
Brazil, China, and Poland represent the most significant emerging markets in Latin America, Asia, and Central Europe, respectively (PwC, 2017).Through unique characteristics, these countries are united in their shared uncertainties and challenges as emerging nations.Brazil and China exemplify the distinct idiosyncrasies within the group of emerging countries: Brazil has strayed from a consistent growth path due to severe economic and political turmoil in recent years, whereas China continues to expand its global competitive presence, experiencing both positive and negative impacts due to the Chinese state's significant role in the economy (The Economist, 2017).Poland, an European Union (EU) member and Central Europe's largest economy, exhibits an industrial competitiveness that rivals that of Asian countries (Euronews, 2018;Sharma, 2017).However, despite recent growth, Poland's stability is compromised by ongoing tensions between its main institutional bodies (Sharma, 2017).
Italy and Sweden, in turn, are developed countries with mature SME ecosystems that have received significant attention in internationalization studies (Ciravegna et al., 2014;Oparaocha, 2015).Yet, each of them offer particular environments to local SMEs.Since the mid-1990 s, Italy's performance has lagged behind its European counterparts and other countries in the Organisation for Economic Co-operation and Development (OECD) due to structural problems such as excessive bureaucracy, underdeveloped financial markets, and educational gaps (Jones & Mackenzie, 2014).These challenges hinder Italy's potential as Europe's second-largest manufacturer (Pinelli et al., 2015), which however does not prevent it from exhibiting strengths across market size, innovation capabilities and infrastructure (WEF, 2018).Moreover, Italy's EU membership affords its businesses favorable market and governance conditions, a stark contrast to those of emerging nations.Sweden, having joined the EU in 1995, ranks highly across various international indices (Gray, 2017).By blending a capitalist economy with a comprehensive social security system, Sweden fosters an environment conducive to innovation and entrepreneurship (Goodman, 2017).
Altogether, this sample provides the opportunity to observe the selected empirical phenomenon under a complex gradient of environmental conditions, where the chosen countries blend similarities and differences.While homogeneity in multi-country samples allows for the control of external factors, the deliberate introduction of heterogeneity in the sample can enhance the internal validity of studies and contribute to the generalization of results (Reynolds et al., 2003;Yang et al., 2006).This strategy enables a balance between the specificity of insights derived from similar environments and the broader applicability of findings across diverse contexts.

Data collection
Due to the lack of databases for identifying SMEs, especially in emerging markets (Batjargal et al., 2013), we drew on various sources to compile a list of firms that could fit our study, including government datasets, consultancy reports, and online search engines.We followed the European criterion for SMEs based on firm size (up to 250 employees) (European Commission, 2005), except for China where we considered firms with up to 300 people as SMEs following local legislation (Xiangfeng, 2008).We also sampled manufacturing firms featuring international sales within the last three years, with domestic ownership and independent operations.
Data collection in emerging markets happened in China in 2015, in Poland in 2016, and in Brazil between 2017 and 2018.Data for developed markets was collected in Sweden in 2015 and in Italy between 2017 and 2018.In Brazil, we gathered information about firms across the national territory from the Center for Studies on Foreign Trade and the Brazilian Ministry of Industry, Foreign Trade, and Services.In China, we focused on five major cities where private SMEs are dominant and relied on information from the Foreign Economic Trade Commission of each city, the China Commodity Net database run by MOFCOM, and the customs authorities in each city.In Poland, we identified ten major cities and their surroundings by relying on information from the Polish Central Statistical Office and Market Research Company.As for Italy, we focused on SMEs in the northern and central regions of the country with the help of the AIDA database.In Sweden, we focused on a concentration of manufacturing SMEs in the country's middle part and derived a list of potential SMEs from Business Sweden.
We randomly selected firms from a compiled list of candidates for the implementation of survey interviews in each country.The response rate varied between 3.6 % in Italy and 34 % in Poland, which can still be considered appropriate in an international research context (Harzing et al., 2013).These variations can primarily be attributed to differences in the geographic spread of the data collection efforts in each country, which had to observe relevant locations for firms targeted in this study.Interviews lasted for 40 min on average, with respondents holding managerial roles.The questionnaire was first developed in English and then translated with the revision of experienced native scholars not involved in this research.We also employed back translation to ensure cross-country equivalence of questions.
Our final sample included 851 SMEs, with 133 Brazilian firms, 215 Chinese, 183 Italian, 152 Polish, and 168 Swedish (Table1).In light of our efforts to ensure sampling equivalence among countries, differences reported in Table 1 can be believed to reflect variations in the profile of the SME population in each location.We found no evidence of nonrespondent bias after comparing early and late respondents within each country group with t-tests across both firm age and size (Armstrong & Overton, 1977).Considering that SMEs are marked by concentrated decision power (Lefebvre et al., 1997), and that we identified knowledgeable informants by asking to talk to the person responsible for the firms' foreign operations, we can assume single response bias to be limited.
We followed several approaches to remedy the possibility of common method bias (CMB) (Podsakoff et al., 2003).To get honest answers, for instance, we guaranteed anonymity to respondents and assured them that there were no right or wrong answers.We also distributed the items for our latent variables in the questionnaire to limit artificial correlations.The implementation of interviews by different researchers additionally avoids biasing the data.Post-hoc, a Harman's single-factor test showed that the largest factor of the un-rotated solution accounted for only 31.7 % of the variance in the data.We also conducted a full collinearity test that is recommended for assessing CMB in PLS-SEM studies: the maximum variance inflation factor (VIFs) among our latent variables was 1.347, satisfying the threshold of 3.3 and indicating limited risk of CMB (Kock, 2015).

Measures
We operationalized our constructs with multi-item measures developed from relevant literature, represented as sentences evaluated in a seven-point scale ranging from 1 = "Completely disagree" to 7 = "Totally agree."In line with previous work on internationalization knowledge (IK) (Eriksson et al., 1997;Fletcher et al., 2013), for instance, we considered the manifestation of this knowledge in items representing the firms' generic experience of foreign market operations.Following Hilmersson (2014), we asked whether respondents agree that their firms have experience with: (1) supplying foreign customers (IK1); (2) adapting our products and services to meet the needs and wants of foreign customers (IK2); (3) adapting our organization to meet the needs and wants of foreign customers (IK3); (4) marketing and sales our products and services abroad (IK4).
Considering the importance that knowledge about markets and technologies may have for SMEs (Prashantham & Young, 2011;Wiklund & Shepherd, 2003), we use these dimensions to assess the knowledge that such firms may have about their business network.The business network comprises not only immediate business contacts, but also their suppliers' suppliers and their customers' customers (Hohenthal et al., 2014) and consequently we assessed Supplier network knowledge (SNK) by asking respondents whether firms in their market usually have knowledge about their suppliers' suppliers': (1) products and technology (SNK1); and (2) markets and customers (SNK2).Similarly, we assessed Customer network knowledge (CNK) by asking them whether firms in their market usually have knowledge about their customers' customers': (1) products and technology (CNK1); and (2) markets and customers (CNK2).
Non-predictive Strategy (NPS) involves the development of unpredicted opportunities while leveraging pre-existent resources (Wiltbank et al., 2006).Such logic implies the incorporation of changes in a firm's strategy, which leads to assuming flexibility as a major aspect of this concept (Mauer et al., 2018), which corresponds with the definition of the concept's three elements.Therefore, we asked respondents to what extent: (1) Entering foreign markets often means that we have to be flexible (NPS1).Because non-predictive strategies entail creativity and purposeful creation of alternatives from the resources available to the firm (Sarasvathy, 2001;Wiltbank et al., 2006), we also asked them to what extent: (2) Entering foreign markets often means that we have to go out of the box (NPS2).Finally, experimenting with new solutions and improvising can also be perceived as a manifestation of non-predictive strategies (Alsos et al., 2020;Evers & O'Gorman, 2011), hence we asked them to what extent: (3) Entering foreign markets often means that we improvised solutions to problems (NPS3).
We measured International Market Performance (IMP) with three items reflecting the extent to which the respondents perceive the overall success of their operations in foreign markets (Katsikeas et al., 2000).In fact, access to objective and reliable performance data from privatelyheld SMEs is usually severely restricted, given that it is often not publicly available and also because managers tend to hesitate in revealing sensitive information to researchers (Dess & Robinson, 1984).For this reason, SME performance is recurrently assessed with managers' subjective perceptions through different kinds of measures (Madsen & Moen, 2018).Previous research supports this practice by showing that perceptual performance measures tend to be highly correlated with objective measures (Andersson et al., 2001;Madsen & Moen, 2018).In this sense, we asked whether interviewees agree with the following: (1) We are pleased with how the profitability of our international operations have developed over the last three years (IMP1); (2) We are pleased with our market-share over the last three years in the international markets where we operate (IMP2).To establish a reference for this construct and avoid vague assessments, we also asked interviewees to consider whether: (3) We have over the last three years had a leading position in the international markets where we operate (IMP3).
We control our model for firms' age (AGE, in years) and size (SIZE, in number of employees), which may imply differences in organizational and environmental characteristics experienced by companies (Wiklund & Shepherd, 2005).Another control is interviewees' experience in their companies (RESPEXP, in years), which can lead to variations in how they process risk and in their openness to new information.Due to potential effects of foreign market uncertainty on firms' decision strategy and performance, we also control our model for the institutional development level of each firm's major host country (identified through our questionnaire).Following previous research (Hernández & Nieto, 2015 ;Miller et al., 2016), we assessed such information with the World Bank's governance indicators (World Bank, 2019) evaluated in a factor analysis with varimax rotation.This analysis produced a single factor responding for 88.4 % of the variance in the data, with loads larger than 0.85, which we used as our control variable (INSTDEV).

Data analysis
We analyzed our data with partial least squares modeling (PLS), a variance-based technique that emerged more strongly in recent years as an alternative to the traditional SEM models based on covariance (CB-SEM) and gained popularity rapidly in various fields (Guenther et al., 2023;Hair et al., 2017;Richter et al., 2016Richter et al., , 2022)).While CB-SEM seeks to estimate how much a theoretical model is close to the covariance matrix observed in a given sample, PLS-SEM operates multiple regression analyses that maximize the variance explained for endogenous variables (Hair et al., 2014a).Because of that, CB-SEM tends to be better suited for theory confirmation whereas PLS finds its main applications in prediction and theory development (Henseler et al., 2014;Sarstedt et al., 2016).The latter situation reflects the context of our research, which includes a scant attention to the specificities of knowledge types and their role in SME internationalization.
Moreover, PLS-SEM models mitigate the demands that methods like CB-SEM impose on measurement scales, sample size, and residuals' distributions, while remaining robust against inadequacies such as multicollinearity, skewness, and omitted variable (Cassel et al., 1999;Hair et al., 2014a).That is possible with PLS' distributional assumptions, which it is able to make less strict without compromising the concepts in the study, the model, or the estimation technique (Lohmöller, 1989).These aspects reinforce PLS' relevance for a study like ours, whose measures are derived from theory but include items developed for this research.Due to those reasons, we adopted PLS-SEM and implemented our analysis using the software SmartPLS (Ringle et al., 2015).

Results
The assessment of a PLS path model requires the verification of the validity and reliability of measurements and the assessment of the hypothesized relationships (Hair et al., 2014a).

Measurement model
According to Table 2, all constructs satisfy the recommended thresholds of 0.7 for both composite reliability (CR) and Cronbach's Alpha, and 0.5 for average variance extracted (AVE) (Hair et al., 2014a), indicating that our measures are internally consistent and exhibit convergent validity.Moreover, all of our items feature loadings above the suggested cut-off value of 0.7.Following Fornell and Larcker's (1981) criterion, we can establish discriminant validity by verifying that the square root of each construct's AVE is higher than its correlations with other constructs (Table 3).This diagnostic indicates that the model employs reliable and valid constructs.
A potential concern in cross-country studies is measurement invariance, which requires assessing the consistency of item measures across subgroups (Steenkamp & Baumgartner, 1998;Vandenberg & Lance, 2000).We addressed this issue using bootstrapping to compare the factor loadings of our items across countries in our sample at 95 % confidence level, using Bonferroni correction to adjust the confidence interval of the pairwise comparisons (Hair et al., 2014a).This method allows to compare group-specific confidence intervals without any assumption about the distribution of the data (Sarstedt et al., 2011).If the country-specific confidence intervals for a given indicator overlap, there are no significant cross-country differences in this factor loading.Following such a procedure, we identified no threats to measurement invariance.
With the evaluation of the measurement model, the PLS algorithm computes standardized latent variables scores that configure proxies for the observed indicators.Such scores can be used to provide a brief overview of how the country groups in the sample compare in terms of the main constructs in the model.This comparison is organized in Table 4, which compares the country samples using the Scheffe test in line with the results of Levene's test for homogeneity of variance.This analysis indicates that Chinese firms appear consistently in the group of firms with higher knowledge stocks.Polish firms are indistinct from them in terms of Internationalization Knowledge and Supplier network knowledge.In turn, Swedish firms appear consistently in the group with lower scores for the knowledge types, with Italian firms appearing twice with them-in the analysis of Customer network knowledge and Supplier network knowledge.Brazilian firms' scores for Internationalization knowledge and Customer network knowledge cannot be distinguished from the other firms, while the same happens with Polish firms' scores for Customer network knowledge.
Relative to the scores reflecting the host country institutional development, Brazilian firms exhibit values statistically lower than the firms in the remaining groups.Both Brazilian and Chinese firms adopt non-predictive strategy at higher levels than their peers, with Swedish firms scoring the lower in this aspect.In the end, the international market performance reported by Polish and Chinese firms is higher than that of Swedish, Brazilian and Italian groups.This overview conveys the uniqueness of the countries included in the sample but at the same time does not offer an obvious rationale for the relative position of the SME populations between countries.Such an outcome reinforces the need for analyzes looking at the proposed effects across countries.

Structural model
The analysis of the structural model starts with the estimation of the baseline path model with the pooled data from all five countries.Following recommendations from the literature, we built on a 5,000sample bootstrap to assess the direct paths together with their significance levels (Hair et al., 2011).The results for this estimation appear in Table 5. Path estimates show that NPS is positively related to IK (β = 0.234, p < 0.001), SNK (β = 0.123, p < 0.01), and IMP (β = 0.184, p < 0.001).These results happen together with positive direct paths connecting IMP directly to both IK (β = 0.233, p < 0.001) and SNK (β = 0.097, p < 0.05).No relationship involving CNK turned out to be significant.The Stone-Geisser Q 2 statistic is greater than zero for both NPS (0.122) and IMP (0.137), indicating that the model has predictive validity (Henseler et al., 2009).In terms of controls: firm age is negatively related to both NPS (β = -0.156,p < 0.001) and IMP (β = -0.144,p < 0.001), while firm size is positively related to NPS (β = 0.148, p < 0.001).Finally, INSTDEV is negatively related to NPS (β = − 0.120, p < 0,001) and positively related to IMP (β = 0.085, p < 0,01).
We used bootstrapping to examine the magnitude and significance of indirect effects (Aguinis et al., 2017).This procedure assesses indirect paths resulting from the product of individual direct relationships without specific assumptions regarding the distribution of the data.Results of this procedure also appear in Table 5, indicating that both indirect paths connecting IK and SNK to IMP through NPS are significant (β = 0.043, p < 0.001, and β = 0.023, p < 0.01, respectively).Table 6 features the result of group comparisons performed with PLS multigroup analysis (Henseler, 2007), contrasting the pooled data representing the emerging and developed countries in our sample.The only paths significantly different between developed and emerging countries is NPS→IMP (Δ = 0.258, p < 0.001) and IK→NPS→IMP (Δ = 0.062, p < 0.01), with path coefficients assuming higher values for the developed country group in both cases.As a form of robustness check, we repeated the estimation of our model leaving out one country at a time and found results consistent with our main analysis.
The results in Tables 5 and 6 provide support for H1 and H2, which claimed that NPS mediates positive relationships between IK and SNK, on the one side, and IMP, on the other.The indirect relationship   involving IK is stronger for developed country SMEs, which supports H4a, but the one involving SNK is not different across groups and therefore denies H4b.The analysis did not support the hypotheses involving CNK (H3 and H4c).We discuss these findings in the next section.

Robustness analyses
Given our sampling strategy, focused on manufacturing firms with international sales within the last three years, it is possible that the presence of firms with reduced internationalization could have affected our results.Looking at our sample composition, more than 70 % of the firms obtain at least 25 % of their turnover from foreign sales.The country with less firms in such a group is Brazil (about 46 % of firms with foreign sales equal at least to 25 % of turnover) and the one with more firms in such a category is China, where 100 % of the firms had at least 25 % of their sales from abroad.This picture is not surprising, considering the extremely favorable market growth registered in China when we implemented our survey, driven by the economic incentives deployed by the Chinese government (Hsu, 2016), as well as the structural aspects of the Brazilian market and the political and economic turmoil experienced by local firms at the time of data collection (Romero, 2016).The average share of firms with relevant foreign sales in the other countries we analyzed is about 67 %.
To understand whether the composition of our sample regarding internationalization level has meaningful influence over our results, we ran an additional mode specification with a control for the level of foreign sales (with SALESLVL being a binary variable that is equal to 1 when foreign sales is at least 25 % and 0 otherwise).The results with this alternative model provide basically the same insights obtained with our original model (Tables 7 and 8 below).The main additional finding from this new analysis is a significant relationship between the level of foreign sales and the reported performance (β = 0.234, p < 0.001, from Table 7), which cannot be considered unexpected.We obtained very similar results when we repeated the analysis while using the level of foreign sales directly as a continuous control variable.Thus, we understand that the difference between firms with reduced and relevant levels of foreign sales does not affect our study.
In addition to considering whether firms with reduced internationalization levels could have impacted our results, we also verified the opposite extreme: if the presence of firms with extensive and early internationalization (also called Born Globals) mean anything in the interpretation of our findings.Considering Born Globals as firms that internationalize within three years of inception and obtain at least 25 % of their revenues from abroad (Knight & Cavusgil, 2004;Paul & Rosado-Serrano, 2019), we find that 374 firms in our sample (approximately 44 % of the sample) qualify as born global.We repeated our analyses, comparing Born Globals with non-Born Globals, to understand whether this perspective brings meaningful insights to our analyses.
Table 9 below indicates that the differences between Born Globals and non-Born Globals is limited to direct relationships involving IK.The path IK→NPS is weaker for Born Globals (Δ = -0.173p< 0.05), as well as the path IK→IMP (Δ = -0.190p< 0.05).That is possible attributed to the nature of this type of knowledge, which is based on procedural information regarding how to operate abroad.Since Born Global firms internationalize quickly but very early, it is not surprising that they did not have enough time to process and store general knowledge about running foreign operations in the same way that other firms.That could explain why the effects associated with their IK is weaker.Other than   that, the results for both groups are not statistically different.

Discussion
Our results outline nuances in how available knowledge assets feed into SME internationalization, highlighting the relevance of specifying the type of knowledge considered in this process.For instance, while the use of IK seems to vary according to the institutional conditions at SMEs' home market, no cross-country differences seem to exist with SNK.Two major arguments may help in explaining this finding, starting with possible differences in how SMEs apply IK and SNK.Due to IK's nature as generic input for strategizing and operating abroad, it holds potential to be used in initiatives that are much more complex and subject to constraints than those powered by SNK.Such initiatives range from structuring organizational systems for foreign operations to designing new market entries and developing business relationships in the foreign market's network (Fletcher et al., 2013;Fletcher & Harris, 2012).These opportunities may be subject to limitations even before firms conceive them, like when home country regulations constrain the way foreign operations should be financed, designed, and managed.Comparatively, SMEs can use SNK for relating to their supplier networks and leveraging the resources and creativity that exist there (Lee & Schmidt, 2017).These initiatives are likely to translate into product development efforts (Mena et al., 2013), which have a much more focused scope and are simpler than opportunities powered by IK.It is also likely that SMEs' supplier networks involve mostly domestic actors, given that these firms' limited resources constrain their ability to search for foreign suppliers and pay for the related logistics and fiscal expenses.These aspects further simplify opportunities derived from SNK, even for emerging market firms.
Besides that, there seem to be differences in how applications of IK and SNK extend SMEs' resources.On the one hand, SNK helps firms relate to actors to which they are already connected, either directly or indirectly.This means that the resources that firms can obtain with SNK are accessed more easily than those obtained from new relationships in the business network.Because SMEs and their suppliers tend to be in the same country, knowledge accessed through these network relationships could also be relevant for helping SMEs in dealing with the home country institutions.On the other hand, although IK helps firms in operating abroad, it is less likely to expand SMEs' available resources in the same way that SNK can.It is true that IK may help firms to connect with foreign partners, but relationship development in foreign business network is a resource-intensive and time-consuming effort (Johanson & Vahlne, 2009).Even if SMEs connect easily with foreign partners, the knowledge accessed in this way is less likely to help them handle challenges at home, particularly in emerging countries.Altogether, it seems that the idea of home country institutions as a general layer of inefficiency for SMEs' strategies is an oversimplification-the effects seem to depend on the type of knowledge that these firms leverage in their decisions.
The non-significant results for CNK may reflect limitations in the measurement items adopted for this knowledge type, which may depend on specific foreign markets or particular product lines.It is also possible that market and technology aspects of CNK exhibit distinct relationships with variables of interest like decision-making strategy or international market performance, requiring separate measurement and hypotheses testing (Prashantham & Young, 2011;Wiklund & Shepherd, 2003).Although such issues deserve attention, it is opportune to consider that a reduced relevance of CNK for SMEs' international market performance is not something entirely odd.After all, SMEs tend to operate in the direct and intermediate parts of their business networks (Buckley & Prashantham, 2016) and use exports or intermediaries to serve foreign markets (Tan et al., 2016).It is thus possible to speculate that IK and SNK in fact play a greater role in SME internationalization than CNK.
Regarding our controls, the negative effect of firm age on IMP can be associated with the strategic rigidity that firms tend to develop over time (Casillas et al., 2012).Similar arguments apply to the negative effect of firm age on NPS, as older firms tend to exhibit reduced openness to the heuristics associated with this decision model.The positive relationship between firm size and NPS may reflect the nature of this decision approach, benefited by resource availability (Wiltbank et al., 2006).The negative relationship between INSTDEV and NPS may result from our operationalization choices, given that the World Bank's governance indicators translate a more formal aspect of national institutions (Hernández & Nieto, 2015).It is possible that countries with mature formal institutions limit firms' possibilities for acting in a creative and improvisational way, with clear requirements for acting there.INSTDEV's positive effect on IMP, in turn, goes in line with the institutional view that clear and well-developed institutions facilitate firm activity (North, 1990).

Theoretical contributions
Our study contributes to extending the scope of effectuation principles investigated within international entrepreneurship.Although the specific principles making up Effectuation and their exact nature may vary slightly between publications (Reymen et al., 2015), the bird-inhand principle, representing the means directly or indirectly available at the starting point of foreign opportunity development, seems to be one of the least explored facets of effectuation within firm internationalization.That goes perhaps because most internationalization studies applying effectuation lenses tend to focus on the resource scarcity characterizing SMEs (Karami et al., 2020) and let untapped the value of discussing specific complementary resources that a firm can access through its network.CNK and SNK, as we identify, illustrate such complementary resources and show potential to develop firm internationalization in different ways.
The key contribution of the study comes from the support that we find for the effect of SNK and IK as contingencies on NPS, which the literature tends to link to effectual decision-making.Thus, when a firm is able to develop a body of knowledge containting general IK and more specific knowledge emerging from its network of suppliers, NPS becomes critical for performance.Our interpretation is that IK advises the firm to be agile and flexible, that is, to pursue a NPS as it has a positive effect on performance.SNK, in turn, captures the knowledge the firm possesses not only about direct suppliers but also about suppliers beyond them.We contend that, for most SMEs, internationalization is mainly an export and sales acitivity which, in turn, indicates that purchasing and supply chain management is more of an ad-hoc activity, that is, it has much in common with NPS.These findings support the main reasoning of effectuation theory, vieweing internationalization process as difficult to predict and taking place under uncertain conditions.Besides our indication that SMEs from different origins seem to diverge in how effectively they can deploy IK, our results about the importance of SNK deserve attention to the extent that the supply segment of business network tends to be ignored by researchers (e.g., Gassmann & Keupp, 2007;Martin et al., 2019;Prashantham & Young, 2011).The analysis does not support the hypotheses involving CNK, which signals to an interpretation built on the strength of business relationships.As relationships with customers, thus, selling and exporting, is the main reason for SME internationalization, and because these networks tend to be developed on strong and long-term relationships, they offer a platform for joint planning with the partners.Knowing a lot about the customer network is a result on long term interaction and information exchange, where a critical component is the planning and predicting what product or service to sell, deliveries, financial solutions, design, etc.This cannot be realized without joint prediction.Consequently, in order to develop and maintaining customer network, the firm needs to exchange information and to jointly plan the relationships, which does not aloud a strategy based on non-prediction.Our results, therefore, draw attention to the possibility that CNK hold less relevance for internationalization than the other knowledge types.By understanding the differences in how SMEs use different types of knowledge, researchers may be able to devise more specific propositions to advance theory.
Finally, our study contributes to a more contextualized view of decision-making in internationalization (Elbanna et al., 2020) by highlighting home-country effects to be acknowledged in this process.On the one side, such effects remain underexplored in studies of foreign expansion (Skorupski et al., 2019;Torkkeli et al., 2019).On the other side, authors have preferred to focus on the challenges that SMEs face abroad and how these firms can overcome them in an entrepreneurial way (e.g., Kujala & Törnroos, 2018;Sarasvathy et al., 2014;Vasconcellos et al., 2019).This happens despite the fact that home country institutions can sometimes be more relevant than internal resources and capabilities in influencing SMEs' foreign expansion (Williams & Spielmann, 2019), shaping their strategies and international performance (Lahiri et al., 2020;Marano et al., 2016).

Managerial relevance
Our results also seem to have important managerial implications.To tap knowledge from the network has different consequences depending on the type of knowledge, implying that knowledge and network are heterogeneous concepts of which managers have to be aware.The negative relation between CNK and NPS and the effect on performance entail a structured and analytical approach to the knowledge coming from customers and their customers and how it can be used.As it does not match with NPS, it should probably be better applied in prediction and in the planning of future objectives and activities.One the other hand, general IK and SNK denote that the firm should pursue a NPS in order to achieve positive performance outcomes.Altogether, this suggests that managers should have a sophisticated approach to networks and knowledge if they aim for high performance in international markets.
Considering that home-country contexts may be strong enough to compromise the performance of SMEs' non-predictive strategies, especially when they use IK, SMEs may need to take additional measures to overcome obstacles at home in order to find success in the international arena.By accumulating the right type of knowledge and capabilities to deal with the liabilities of their home location, for instance, they may end up turning their disadvantages into advantages as larger firms sometimes do (Cuervo-Cazurra & Genc, 2008).They may also focus on using non-predictive strategies when leveraging their stock of SNK, which is the knowledge type that seems to be to some extent immune to home country effects.

Limitations and future research
The limitations of our research start with the extensive and complex data collection process involving empirical work over several years.This complexity is due to the challenges of conducting a large-scale survey study with SMEs, including the absence of comprehensive databases, difficulties in accessing and interviewing SMEs, and observing countryspecific SME ecosystem nuances while ensuring cross-country comparability.However, these effects are likely limited as the data collection during the study period happened without interruptions between country markets and was completed prior to COVID-19 disruptions.The observation of local SME ecossystems also resulted in variations in response rates due to differences in the geographic spread of the data collection efforts in each country, but we understand that this enabled a sampling process with an improved validity.
Besides that, our analysis does not individualize specific decisions of SME internationalization like market entries and foreign relationship development.Aspects related to the firms' host markets and the structure of their foreign market networks may hold some effect on their performance.However, such effects should be minimal given that SMEs tend to concentrate their operations at home and interact with foreign markets mostly indirectly (Tan et al., 2016).Additionally, the subjective measurement that we adopted to assess performance limits our ability to make specific inferences about the firms' actual conditions.Although this choice is common in SME-related research (Madsen & Moen, 2018), and perceived and objective measures tend to be consistent (Andersson et al., 2001;Madsen & Moen, 2018), future studies could verify our findings with objective indicators.
Moreover, we have focused this study on exploring the Bird-in-hand principle besides principles indirectly approached through the concept of non-predictive strategy (namely, Lemonade and Pilot-in-the-plan).Still other principles of Effectuation could (and have been) discussed in the context of firm internationalization, like in the case of Affordable loss (Dew et al., 2009).The concept of Affordale loss helps firms to select among the available courses of action depending on their ability and willigness to proceed (Martina, 2020) and is a multidimensional phenomenon, requiring a consideration of financial, psychological, social, status, and reputation aspects for a proper analysis (Karami et al., 2020).It thus exceeds the scope of this study but certainly offers important research avenues.
Future research could additionally explore the characterization of home-market effects in firm internationalization and implications of the importance of SMEs' supplier network for internationalization.Given that our results reported SNK with an effect greater than that associated with IK, future studies should take that into account and consider an extended network perspective when investigating SME internationalization from a knowledge-based perspective.In the same way that Mudambi and Puck (2016) considered that studies of the strategy and performance of multinationals that ignore value generation and knowledge in the supplier-related network are biased, the same could apply to studies that fail to recognize the importance of supply-related knowledge for SMEs.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Table 1
Sample profile (standard deviations in parenthesis).

Table 2
Reliability and convergent validity.

Table 4
Cross-country comparison of latent variable scores (standard deviations in parentheses).

Table 5
Path coefficients and t values.

Table 7
Path coefficients and t values (alternative model with sales level control).

Table 8
Multi-group comparisons (alternative model with sales level control).