Internationalization and firm performance: Meta-analytic review and future research directions

WINFRIED RUIGROK HARDY WAGNER Research Institute for International Management University of St. Gallen Bodanstrasse 6 9000 St. Gallen Switzerland [email protected] [email protected]

Internationalization and firm performance: meta-analytic review and future research directions

Abstract Despite three decades of management research on the relationship between internationalization and firm performance, no consensus exists about its direction and magnitude. By synthesizing and analyzing research findings from 62 studies (174 samples, N = 35,631), we find empirical support for a non-zero, positive impact at the aggregate level. Equally important, meta-analysis reveals that the form of the relationship is dependent on contextual settings (i.e., company size and company nationality). Therefore, we develop a parsimonious moderator set (scale, scope, mode, object, goal, and pace and rhythm of internationalization) that may be used by future researchers as a platform for programmatic contextual inquiry.

Keywords: Internationalization, firm performance, contextual inquiry.

Introduction Over the last three decades, the intriguing question of whether foreign direct investment affects firm performance has been addressed by multiple research fields (e.g., international management, strategic management, economics, finance and accounting, retail and marketing, and business ventures) and has become one of the most researched linkages in international management (Werner, 2002). Yet, after close to a hundred accumulated studies, researchers still perceive the findings to be inconsistent (Palich, Carini and Seaman, 2000; Riahi-Belkaoui, 1999), conflicting (Annavarjula and Beldona, 2000; Capar and Kotabe, 2003; Kotabe, Srinivasan and Aulakh, 2002; Qian, 2002), mixed (Doukas and Lang, 2003; Hitt, Hoskisson and Kim, 1997; Martinez, 2002; Vermeulen and Barkema, 2002), and contradictory (Contractor, Kundu and Hsu, 2003; Geringer, Tallman and Olsen, 2000; Ruigrok and Wagner, 2003) and the question therefore unresolved. Earlier narrative reviews of the research stream have provided method-focused reasons for these inconclusive findings (Annavarjula and Beldona, 2000; Ramaswamy, 1992; Sullivan, 1994). Reviewers have argued that the choice of variable operationalization has been largely convenience based rather than explicitly coordinated with authors conceptual logic. Consequently, supposed lack of content and construct validity caused spurious findings. Later studies did incorporate reviewer suggestions but apparently still failed to resolve the problem of inconclusive findings. This investigation is the first to use meta-analysis in an attempt to quantitatively consolidate the fragmented research findings and identify promising directions for future inquiry. Specifically, we aim at making three contributions to the literature. First, meta-analysis is a powerful tool for the detection of the true nature of the linkage between two variables (Dalton, Daily, Johnson and Ellstrand, 1999) because it allows for the elimination of key statistical artifacts that confound the results of primary research. Thus, by meta-analytically analyzing a representative sample of 62 studies published or reported 2

between 1974 and 2004, we intend to determine whether, and if so how, internationalization is related to firm performance. Second, we seek to identify new methodological as well as conceptual moderators of the internationalization-performance relationship. The opportunity to detect novel moderator variables by investigating a large set of findings reported by primary research is a unique strength of meta-analysis (Eden, 2002; Hunter, Schmidt and Jackson, 1982; Miller and Monge, 1986). This search for moderator variables also allows us to determine whether there indeed exists the widely assumed but to date empirically unconfirmed universalistic form of the internationalization-performance relationship or whether the phenomenon is context dependent, a hitherto largely omitted but central question in the research stream. Third, we aim at contributing to theory development by outlining potentially fruitful directions for future investigations. As noted by Eden (2000), meta-analysis can be used to show which kind of further research will be the most worthwhile Moreover, the findings of meta-analysis can raise new theoretical questions and frontiers for future replication research. In this way, meta-analysis is not necessarily the terminus in a stream of research; it can also point to the best direction for new theory development (pp. 843-844). The article is structured as follows: after delineating the underlying logic of the internationalization-performance research stream, we explain our meta-analytic methodology and then present our results. In the final sections, we discuss our findings and outline promising avenues for future research.

Conceptual logic of the research stream The conceptual logic of this research stream rests on two key explanatory factors: benefits and costs. Whereas early inquiries were confined to or accentuated the benefits of internationalization, later research addressed both benefits and costs together with their tradeoffs along the internationalization continuum.

Benefits of internationalization Theories of foreign direct investment (FDI) aimed at justifying foreign expansion have constituted the core conceptual foundations of internationalization-performance inquiry. Early FDI concepts focused on beneficial foreign factors located outside the organization, arguing particularly that imperfections in world factor and financial markets benefit the internationally expanding company. Whereas authors with backgrounds in international business and industrial economics emphasized cost advantages (transaction costs, labor costs, material costs) and economies of scale or scope (monopoly advantage and risk diversification, respectively) stemming from internationalization (e.g., Buckley and Casson, 1976; Caves, 1971; Dunning, 1979; Hymer, 1976; Porter, 1985; Vernon, 1966), scholars of financial economics underlined portfolio diversification theory (exploitation of national or regional divergence of factor price levels, tax systems, and capital sources) and its implications on risk-return performance (e.g., Errunza and Senbet, 1984; Hirsch and Lev, 1971; Lessard, 1973, 1976; Levy and Sarnat, 1970; Morck and Yeung, 1991). However, this focus on reactive exhaustion of external opportunities as the sole source of benefit shifted once FDI and international management researchers began to examine the organization itself and take into account the proactive development and exploitation of internal advantages. For example, Fayerweather (1978) suggested that the transfer of managerial and technical know-how on a worldwide scale can equip multinational companies (MNCs) with comparative advantages not available to firms operating purely domestically. Likewise, the resource-based view of the firm (Prahalad and Hamel, 1990; Wernerfelt, 1984) underlined intra-company global competencies as promoters of superior performance development. For instance, cultural heterogeneity in the workforce and at management level was argued to trigger innovation and technological advancement (Barkema and Vermeulen, 1998; Hitt et al., 1997). Finally, the theory of operational flexibility (Kogut, 1989) proposed

that firms operating in multiple environments are flexible enough to generate both arbitrage (e.g., information transfer) and leverage (e.g., cross-subsidization) opportunities.

Costs of internationalization In addition to conceptualizations of the benefits of internationalization, there exist multiple interdisciplinary rationales regarding the costs of corporate internationalization, particularly the stigma of being foreign (Hymer, 1976), the liability of foreignness and newness (Zaheer, 1995; Zaheer and Mosakowski, 1997; Coviello and McAuley, 1999), and the discount for global diversification (Click and Harrison, 2000; Denis, Denis and Yost, 2002; Fatemi, 1984). Costs, like benefits, can be viewed as stemming from either the external business or intra-company environment. External costs of internationalization emerge from political, financial, and market-related hazards. Political uncertainty refers to the home (e.g., taxes, business prohibition) and host (e.g., fund remittance control, boycotts, expropriation) governments differential treatment of internationalizing firms and native firms (Boddewyn, 1988; Brewer, 1992; Moran, 1974). Financial risk refers to exchange-rate fluctuations and inflation (Reeb, Kwok and Baek, 1998; Solnik, 1974) that may offset advantages of lower earnings volatility proclaimed in portfolio diversification research. Market-related hazards include such expansion liabilities as consumer ethnocentricity, consumer taste divergence, and low purchasing power. Even though these costs are not necessarily time steady, they can represent high fixed costs at the outset of foreign expansion. With respect to intra-company costs of internationalization, researchers have emphasized the increase in transaction (Jones and Hill, 1988) and agency costs (Doukas and Travlos, 1988; Morck and Yeung, 1992; Roth and ODonnell, 1996) that result as international expansion leads to the geographical and cultural scattering of employees globally. Not only can significant geographical distance from the home market cause high transportation and 5

traveling costs, but cultural and linguistic demands may produce immense informationprocessing requirements. In addition, as principals and agents of the firm are dispersed across nations, cultures, and institutional settings, the stress of coordinating, configuring, communicating, and motivating (Hofstede, 1980; Ramaswamy, 1992) may exhaust managerial capacity and make navigating the corporate complexity increasingly difficult (Qian, 2002; Siddhartan and Lall, 1982). Overall, transaction and governance costs are argued to mount rapidly as firms intensify their foreign market exposure.

Benefit-cost trade-off Explicit consideration of the benefit-cost trade-off along firms foreign expansion paths is absent from the early studies (e.g., Aggarwal, 1979; Hirschey, 1982; Severn and Laurence, 1974; Siddharthan and Lall, 1982). Implicitly, researchers assumed that either the benefits or costs of international expansion dominate throughout the internationalization continuum (from low to high degrees of firm internationalization), resulting in either a positive or negative linear internationalization-performance nexus. In contrast, during the last decade, many researchers have paid explicit attention to the benefit-cost trade-off, drawing on differing scenarios for explanation, and hypothesizing and identifying two primary non-linear curve types: quadratic (i.e., J, U, inverted-J, and inverted-U) and cubic (i.e., horizontal-S). Two polar logics underlie the quadratic curves J and U and their inversions. To support the former, researchers have argued that over time firms can learn to minimize the significant costs associated with foreign expansions (e.g., Capar and Kotabe, 2003; Dragun, 2002; Eden and Thomas, 2000; Lu and Beamish, 2001; Mathur, Singh and Gleason, 2001; Qian, 1997; Ruigrok and Wagner, 2003), meaning internationalization costs outweigh benefits only until firms gain experience and learn to deal with them. Consequently, firms will reach an inflexion point along the expansion path at which incremental benefits start to outweigh incremental costs (visualized as a J- or U-curve). This scenario implies that firms may have to 6

undergo a period of performance deterioration before experiential knowledge can lead to high performance levels. Researchers identifying the opposite quadratic curve types (i.e., inverted-J and invertedU) have contended that firms do not need to explicitly address initial internationalization costs through organizational learning (e.g., Daniels and Bracker, 1989; Geringer, Beamish and daCosta, 1989; Gomes and Ramaswamy, 1999; Ramaswamy, 1995) but can, as noted by Gomes and Ramaswamy (1999), deploy their home based skills and resources to achieve economies of scale and/or scope without huge cost increases (p. 176). Thus, at the outset of foreign expansion, the incremental benefits of internationalization are suggested to outweigh the incremental costs. However, as firms intensify their foreign expansion, not only may coordination and monitoring costs increase exponentially and become difficult to address through organizational learning, but learning costs may outweigh value generated. In sum, an internationalization threshold is identified at the point where incremental costs of internationalization start to outweigh incremental benefits, implying that firms should not overstep this expansion peak. Finally, researchers who advocate a cubic curve type (i.e., horizontal-S) have proposed a three-stage conceptual logic for the internationalization-performance link (e.g., Contractor et al., 2003; Riahi-Belkaoui, 1998; Sullivan, 1994b) that aims at reconciling the two conflicting two-stage quadratic curve logics. The basic tool of synthesis has been differentiation between two types of internationalization costs: fixed and modest costs at low DOIs (Type 1) and continuous and immense costs at high DOIs (Type 2). Type 1 costs stem from the liabilities of foreignness and newness, e.g., unfamiliarity with trade laws, consumer ethnocentricity, new consumer tastes, and cross-cultural communication problems. Type 2 costs stem from the significant coordination and monitoring demands caused by intense market complexity, dynamism, and uncertainty.

Based on this cost differentiation, advocates of the cubic curve reconcile the seemingly opposite quadratic curves by refocusing the learning issue. Whereas the proponents of the standard J and U curves contend that organizational learning is required and cost-effective along the complete internationalization continuum, and advocates of the inverted-J and U curves suggest that organizational learning on firms expansion paths is unnecessary (at low DOIs) and not cost-effective (at high DOIs); the horizontal-S curve logic argues that learning to address the liabilities of foreignness and newness at low DOIs (Type 1 costs) is necessary and cost-effective, but learning to successfully manage extreme levels of internationalization (Type 2 costs) is not. This final benefit-cost trade-off logic implies that firms should aim at learning to deal with initial costs of foreign expansion but avoid extreme levels of foreign market dependence. Despite this latter attempt to theoretically reconcile opposing views and the multiple conceptual rationales suggesting a link between internationalization and firm performance, no empirical consensus has yet been forthcoming about the direction and magnitude, i.e. the form of this link. By synthesizing and analyzing empirical research findings from 64 studies, we aimed at clarifying whether there indeed exists one universalistic form (largely assumed in the research stream to date) or whether the performance impact of internationalization is context dependent. An affirmative outcome for either question would have different implications with respect to fruitful directions for future inquiry.

Methodology As is customary in business meta-analyses, we began by extracting and filing the references of the most recent studies on the internationalization-performance relationship, moving backward in an iterative process. Then, to locate unpublished research, we searched ProQuest Dissertation Abstracts and requested relevant working papers from researchers at

academic conferences. As explained below, these unpublished studies were necessary to addressing the so-called file-drawer problem (Rosenthal, 1979). For a study to be ultimately included in the meta-analysis, the internationalizationperformance relationship did not have to be its key focus; rather, the authors had to have reported the following five elements: 1) a Pearson product-moment correlation r between internationalization and performance, or other metrics that could be converted into it1; 2) the operationalization method for internationalization and performance; 3) the studys sample size; 4) the nationality of investigated firms; and 5) the time period covered by the study data. If only one of the required criteria was unreported, the respective authors were contacted and asked to supply the information. Overall, this extensive literature search resulted in 89 studies that explicitly or implicitly addressed the internationalization-performance relationship. Of these, 70% (62 studies) met the above-noted five criteria and were included in our database. Table 1 lists these studies and their key study characteristics (author[s], year of publication, published or unpublished, sample source, and variable conceptualization). Our final study population of 48 publications from 18 academic journals representing multiple business research fields2, 7 conference papers, and 7 dissertations yielded 174 usable samples (N = 35,631 companies) in all. The relatively large sample-to-study ratio (2.81) is due to authors normally relying on multiple operationalization modes for both internationalization and firm performance.

Insert Table 1 about here Although approximately 75% of documents in our sample were published or reported during the last decade, the median year of investigation is 1988. With respect to company nationality, 66% of the examined MNCs were headquartered in the U.S., 14% in ten Western European countries (Belgium, France, Germany, the Netherlands, Italy, Norway, Spain,

Sweden, Switzerland, and the United Kingdom), and 10% in Japan, while the residual 10% had been set up through merging of firms from multiple country clusters. The effect size measure was the Pearson product-moment correlation coefficient r between internationalization and firm performance, widely used in meta-analyses (Hunter et al., 1982; Miller and Monge, 1986) because of its ability to determine direction and strength of relationship between two variables in an easily apprehensible metric ranging from -1.0 to +1.0. The effect sizes r extracted from individual studies were statistically analyzed using MetaDos 2.0 (Martinussen and Fjukstad, 1997), a software package that employs formulas provided by Hunter and Schmidt (1990). This meta-analysis technique divides the observed variance in effect sizes into variance resulting from statistical artifacts (random sampling error, measurement error, and range restriction) and variance attributable to true variance. Whereas in our study we could account for random sampling error, because of the lack of information on data reliability and validity in the identified studies, a correction for measurement error and range restriction was not feasible. Nevertheless, we do not consider this limitation serious because the measures for internationalization and firm performance applied in the research stream are (1) largely based on identical metric ranges with similar variance, and (2) sourced from archival and standardized databases. Moreover, as noted by McEvoy and Cascio (1987), corrections for artifacts beyond sampling error generally account for very little additional variability in correlation coefficients in meta-analyses (p. 749). Our meta-analysis proceeded in three major steps. First, we calculated a sample-size weighted mean effect size corrected for sampling error and a 95% confidence interval. This step allowed us to determine the direction and magnitude of the effect of internationalization on firm performance at the aggregate level. Second, we estimated a 95% credibility interval to be tested for the existence of moderator variables. Our three general criteria for population homogeneity were residual standard deviation size, observed effect size variance accounted 10

for by sampling error, and the chi-square test of homogeneity (cf. Hunter and Schmidt, 1990; Sagie and Koslowsky, 1993; Whitener, 1990). Although these tests can suggest the existence of moderator variables, they do not precisely pinpoint them. Therefore, to define individual moderator variables, we then needed to separate our data into multiple two-subgroup sets given the characteristics of the proposed moderators. Z-statistics (cf. Hunter and Schmidt, 1990) were calculated to indicate whether effect sizes significantly differed between the two subgroups in each moderator set (p. 438). (For the statistical formulas and a detailed outline of the applied meta-analytic procedure, see the Appendix). Finally, for our test of nonlinearity, we regressed effect sizes on average sample DOIs to statistically identify the curve type linear, quadratic, or cubic that best reflects the internationalization-performance relationship at the meta-level. To measure the average DOI of individual samples, we used foreign sales to total sales (FSTS), a ratio already provided for 82 of the 174 samples included in the meta-analysis. The application of other indicators for degree of company internationalization (e.g., foreign assets to total assets, foreign subsidiaries to total subsidiaries) would have resulted in insufficient data points. Below, we present the underlying conceptual logics for potential moderator variables and depict our respective coding definitions for the data on each of the five moderator variables elicited from the individual studies. Publication The term file-drawer problem, coined by Rosenthal (1979), refers to the likelihood of publication being greater for significant rather than insignificant results, meaning authors are more inclined to hide the latter in file drawers. Hence, published research may exhibit a bias toward significant findings. To address this publication bias error, researchers must investigate whether effect-size differences are due to the particular source of the studies (i.e., published or not). Our inclusion of 14 unpublished studies on the internationalization-performance relationship allowed us to determine whether effect-size differences between published and unpublished research and the potential for biased results 11

in the research stream did indeed exist. Table 1 (Column 2) shows the coding of individual studies with respect to this moderator variable. Investigative time frame As already noted, FDI theories propose global market

asymmetries as core sources of advantage for companies expanding internationally. However, benefits stemming from these market imperfections can be conditional upon the particular time period investigated (Qian, 1996; Siegel, Omer and Leavins, 1995). For example, imperfections in factor and financial markets around the world are likely to have had greater impact during the 1960s and 70s than during the 1990s after increasing economic integration of world markets through trade links and the establishment of large trade regions like NAFTA, the EU, MERCOSUR, and ASEAN gradually diminished the arbitrage and leverage advantages of international diversification (Levitt, 1983; Yip, 1989). Therefore, we expect the strength of the impact of internationalization on corporate performance to be time variant. In particular, we anticipate that the effect sizes produced by relationship studies carried out after the median year investigated (i.e., 1988) will be significantly lower than those produced by studies examining the pre-1988 linkage.3 Company size As noted by Shuman and Seeger (1986), small-sized firms are not smaller versions of big business (p. 8). Rather, they are fundamentally different from large firms with respect to organizational design, management style, and ownership structures (Lu and Beamish, 2001). Therefore, as regards international expansion, small-sized companies tend to be more disadvantaged than large firms for the following three reasons:

(1)

Small firms frequently have difficulty maximizing economies of scale, a key benefit of internationalization (Qian, 2002), because their product volume in the home market has not yet reached the critical mass that allows scale momentum in the foreign market (Vernon, 1966).

12

(2)

The limited financial capital available to small enterprises may not sufficiently support international expansion (Coviello and McAuley, 1999), a resource constraint that is particularly hazardous at the outset of foreign expansion. The liabilities of foreignness and newness come along with fixed costs that must be financed for a difficult-to-predict time period before learning and market legitimacy occur.

(3)

Managerial constraints resulting from small firms lack of access to managerial talent with international experience (Bloodgood, Sapienza and Almeida, 1996) can cause early exhaustion of managers absorptive capacity, leading to significant implementation costs.

Because small-sized firms may face more difficulties in generating key benefits and addressing the costs of internationalization, we expect the effect sizes produced by studies of small-sized firms to be lower than those generated by researchers examining large MNCs. With respect to variable coding, we were able to identify 5 studies (11 company samples) that explicitly investigated the internationalization-performance relationship for small-sized firms (i.e., Lu and Beamish, 2001; Majocchi and Zucchella, 2001; Qian, 2002; Qian and Li, 2003; Zahra, Ireland and Hitt, 2000).4 Company nationality We offer four conceptual rationales that point to a moderating impact of company nationality on the internationalization-performance relationship: experience of internationalization, scope of internationalization, mode of internationalization, and goal of internationalization. Compared to U.S. firms, who can draw on a large home market to generate economies of scale, companies headquartered in small countries must look outside their home country early on in order to take advantage of these benefits. As a result, these firms now operate at high degrees of internationalization. To illustrate, on average, the largest European firms have

13

recently surpassed the 50% foreign sales-to-total sales ratio (UNCTAD, 2000). Even more impressive, for many decades a significant share of large Dutch, Swedish, and Swiss MNCs has generated and employed more than 80% of sales and employees, respectively, outside the home markets. Consequently, European firms can draw on profound internationalization tenure (Tallmann and Li, 1996; Ruigrok and Wagner, 2003), i.e., accumulated internationalization experience. Such strong experiential knowledge may facilitate the organizational learning that eventually allows for successful foreign expansion. Thus, given their differing internationalization tenures and experience, companies from different countries are likely to encounter divergent internationalization-performance relationships. Second, firms from different nations differ in scope of international expansion (i.e., culturally related versus culturally unrelated). For example, several studies have indicated that U.S. MNCs pursue culturally related strategies at the outset of foreign expansion (Johansson and Yip, 1994; Makhija, Kwongsoo and Williamson, 1997; Tung and Miller, 1990). Given that the average U.S. firm, predominantly due to its large home market, exhibits a comparatively low scale or degree of internationalization (Harveston, Kedia and Francis, 1999; Ruigrok and Wagner, 2003), this finding suggests that, to date, the U.S. organization has pursued a somewhat culturally related internationalization scope. In contrast, given their small home-market sizes, most large European and Japanese companies are compelled to expand early into culturally unrelated business environments (Kogut and Singh, 1988; Cho and Padmanabhan, 1995). For example, Ronen and Shenkar (1985) identified Japan as a fairly insular nation with respect to culture. Therefore, no matter which foreign markets they expand into, Japanese firms move into culturally unrelated markets. Since different scopes of internationalization are related to divergent benefits and costs (Gomes and Ramaswamy, 1999; Gomez-Mejia and Palich, 1997; Ruigrok and Wagner, 2003; Vachani, 1991), they can have a moderating effect on the internationalization-performance relationship.

14

Third, previous inquiry has shown that company nationality is related to organizational mode of entry into foreign markets (Anand and Delios, 1997; Erramilli, 1991). For example, Japanese firms were found to use wholly-owned entry modes (Cho and Padmanabhan, 1995) that allow more rapid transfer of competitive advantages because, in contrast to partiallyowned entry modes (e.g., equity joint ventures), firms need not invest costly and timeconsuming effort into mutually agreeable communication and interaction rules to satisfy the foreign business partner (Barkema and Vermeulen, 1998). This entry mode represents a particular advantage for firms moving into culturally unrelated market environments (Weber, Shenkar and Raveh, 1996. In sum, company nationality can induce entry modes having distinctive benefits and costs that in turn idiosyncratically moderate the internationalizationperformance link. Fourth, company nationality can come along with specific performance goals for internationalization. For example, Nonaka and Okumura (1984) and Harveston et al. (1999) revealed that European and U.S. firms have different performance preferences than their Japanese counterparts. Whereas the former prioritize accounting or profitability ratios (i.e., financial performance), the latter focus on market share, which represents operational performance. Hence, firms from different nations may pursue international expansion to satisfy divergent performance goals. Such dissimilar performance priorities are likely to produce different internationalization-performance relationships. Overall, company nationality is likely to be related to idiosyncratic scales (and thus experience), scopes, modes, and performance goals of internationalization. Therefore, this variable is expected to represent an important contextual setting and macro-level moderator of the internationalization-performance nexus. Based on this assumption, we used the nation in which the companies were headquartered for moderator coding. Variable conceptualization The independent variable internationalization has been

conceptualized in the research stream in three distinct modes: structural, financial, and index. 15

The structural mode defines internationalization in terms of asset, subsidiary, and employee distribution between a firms home country and its host nation(s). Thus, it may be argued that it reflects a companys dependence upon foreign production (Ramaswamy, 1992; Sullivan, 1994). Operationalization ratios applied in the identified studies included foreign assets-tototal assets; regional scope of subsidiaries; number of foreign subsidiaries; foreign subsidiaries-to-total subsidiaries; foreign employees-to-total employees; and Herfindahl or entropy measures for assets, subsidiary, and employee diversification across pre-defined country clusters. The financial mode has been measured using foreign sales-to-total sales, foreign profits-to-total profits, foreign taxes-to-total taxes, and Herfindahl or entropy composites that measure the geographical diversification of these monetary factors. Therefore, it may be interpreted to reflect firm dependence on foreign market sales and profits (Sullivan, 1994). Finally, the index mode refers to multi-item composites that encompass the other two modes. These composites are believed to have lower measurement error and therefore be capable of achieving higher construct validity (Gomes and Ramaswamy, 1999; Sullivan, 1994). Thus, we expect studies applying the index mode to generate higher effect sizes than studies using one of the single-item modes. The dependent variable corporate performance has also been conceptualized in the internationalization-performance literature in three different modes: accounting, market based, and operational. Whereas the accounting mode reflects historical performance, the market-based mode reflects investor expectations of future performance (Venkatraman and Ramanujam, 1986). Accounting-based return indicators applied in the identified studies included sales and asset growth, net profit, profit margin, cash flow, and several profitability ratios (e.g., return on assets, return on sales, return on equity, return on investment, return on value-added). With respect to market-based measures, researchers have chosen market-tobook and P/E-ratios, risk-adjusted market return (e.g., Jensens ) and Tobins q. Aiming to grasp operational performance (i.e., using the operational mode), researchers have made use 16

of company market share, cost efficiency, innovation capability, and employee productivity. In contrast to accounting and market-based performance measures, operational performance indicators do not assess the company-level financial outcome of business operations but quantify the latent competitive advantages that ultimately determine financial performance (Annavarjula and Beldona, 2000; Gomes and Ramaswamy, 1999; Venkatraman and Ramanujam, 1986). Therefore, we expect studies applying the operational mode to generate higher effect sizes than studies using one of the other two performance modes. Figure 1 illustrates the conceptualization skeleton of the internationalization-performance research stream, and Table 1 (Columns 4 and 5) provides our coding of individual studies in terms of conceptualization mode(s).

Insert Figure 1 about here Results In this section, we first present and discuss descriptive findings (cf. Table 2) indicating developments over time with respect to study characteristics in the research stream. Then we proceed with the presentation of results obtained from the meta-analysis (cf. Table 3).

Study characteristics over time In regard to the conceptualization of internationalization, the data indicate that the structural and financial modes have been and still are the most frequently used in the realm of inquiry. However, whereas use of the structural mode has remained constant over the last decade or so, the financial mode has fallen significantly and appears to have been replaced by the index mode, which moved up from 10% of all applied internationalization conceptualization modes (before 1988) to 26% (after 1988). Obviously, Sullivans (1994) plea for the use of index modes has been heeded.

17

Insert Table 2 about here In contrast, conceptualization of performance appears still to be dominated by the accounting mode (74%), while the use of market-based and operational modes remains very low. Further, over the last decade, the market-based mode for performance has been replaced by the operational mode. Yet, the operational mode still only represents about 13% of all modes applied in studies conducted during the last fifteen years. With respect to company nationality, the data suggest that U.S. firms have played and still play a dominant role in the research stream. Although this is not unusual for empirical inquiry in business disciplines (Hoskisson, Hitt, Wan and Yiu, 1999), it is important to note that the dominance of inquiry into U.S. firms has not decreased over the years. Moreover, over the years, research into European firms has decreased (down from 19% to 11%) in favour of investigations into Japanese MNCs (up from 5% to 14%). In terms of sample DOIs (i.e., the average degree of internationalization for individual company samples), the overall population exhibited a foreign sales-to-total sales (FSTS) ratio of 38%5. A sample segregation by the median year of investigation shows that multinational companies have increased their internationalization scales over the last three decades (pre1988 = 30% FSTS; post-1988 = 46% FSTS). In addition, distinct differences exist between companies from different nations and regions. Whereas the average European firm in the sample operated at a 50%6 FSTS, the average U.S. firm exhibited 35%, and the average Japanese company, 29%. At the meta-level, these descriptive statistics empirically confirm findings by previous primary research indicating that European firms operate at significantly higher internationalization scales than their U.S. and Japanese counterparts (e.g., Haar, 1989; Harveston et al., 1999; Ruigrok and Wagner, 2003; Tallman and Li, 1996).

18

Meta-analytic findings Figure 2 displays a stem-and-leaf diagram (Tukey, 1977) showing the first (stem) and second digit (leaf) of the Pearson product-moment correlation coefficients ranked from largest to smallest. It should be noted that the samples and not the studies have been our unit of analysis. Also provided to facilitate interpretation of the findings is a summary of core statistics (largest and smallest correlation, mean, quartiles, standard deviation, and sign proportion). The stem-and-leaf plot and the meta-analytic results obtained by applying Hunter and Schmidts (1990) method indicate the existence of a positive relationship between internationalization and firm performance. Internationalization exhibits effect sizes having a mean of 0.07 (median 0.05) with a range from -0.31 to 0.60 and a standard deviation of 0.15. The proportion of positive signs for effect size is 63%.6 The meta-analysis (cf. Table 3) shows a sample-size weighted r of 0.04 with a 95% confidence interval that does not include zero (0.02 : 0.06), which, as noted by Dalton et al. (1999), is an indicator that there is a true relationship between the variables of interest (p. 676). Overall, our results suggest that at the aggregate level, the benefits of internationalization marginally but significantly outweigh the costs of internationalization.

Insert Figure 2 and Table 3 about here The tests for effect-size heterogeneity met all three statistical criteria (i.e., residual standard deviation size, observed effect size variance accounted for by sampling error, and chi-square test of homogeneity) and thus suggest the existence of moderator variables. First, the residual standard deviation after controlling for sampling error is larger than 25% of the population effect size. Second, less than 75% of the observed effect-size variance is explained by sampling error (i.e., 26%). Third, the null hypothesis of the chi-square test of homogeneity is significantly rejected (2 = 633.83, p < 0.001). 19

In the next step, we identified individual subgroups by testing for the moderator variables defined in the methods section: publication, investigative time frame, company size, company nationality, and variable conceptualization. First, the suggested moderator variable publication did not pass the required significance level (Z = 0.04). That is, the effect sizes produced by published and unpublished studies were not significantly different.7 Thus, there appear to have been no biased results in the research stream due to the file-drawer problem. Based on this observation, during our further analyses, the published and unpublished studies remained in one sample. Similarly, investigative time frame was also not found to act as a moderator variable (Z = 1.16). Although the critical value exhibited the expected direction, the validity generalization test indicated that the effect sizes generated by studies investigating the internationalizationperformance relationship in the 1970s and 1980s are not significantly different to those obtained by studies examining the link in the 1990s. Additionally, when, in an ex-post analysis, we fine-tuned the method and divided our sample population into four clusters separated by decades (1960s, 1970s, 1980s, and 1990s), ANOVA revealed no significant effect-size differences.8 In contrast, company size was found to be a relevant moderator of the internationalizationperformance relationship (Z = -1.98). As expected, the data indicate that studies of smallsized firms tend to generate significantly lower effect sizes than do studies of large firms. Most important, data analysis confirmed our assumption that company nationality represents a key contextual moderator of the internationalization-performance relationship. European firms generated significantly higher effect sizes than their U.S. (Z = -2.62) and Japanese (Z = -4.75) counterparts. Moreover, U.S. firms achieved significantly higher effect sizes than did their Japanese counterparts (Z = -2.94). In addition, we found variable conceptualization, illustrated in Figure 3 by boxplots of r correlations, to represent an important moderator of the internationalization-performance 20

relationship. With respect to conceptualization of internationalization, the meta-analysis only partially confirmed our expectation that studies applying index modes generate significantly higher effect sizes than studies using single-item conceptualization modes. The index modes were found to produce higher effect sizes than the single-item structural mode (Z = -2.44) but not the financial mode (Z = 0.60). Further, the application of financial internationalization modes resulted in significantly higher effect sizes than the structural mode (Z = -4.01). Concerning conceptualization of performance, we found strong support for the expectation that researchers using accounting (Z = -4.46) and market-based (Z = -4.16) modes produce significantly lower effect sizes than those employing the operational mode. Corroborating the strength of this last finding, further analysis showed that the operational mode for performance resulted in significantly higher effect sizes no matter with which conceptualization mode of internationalization it was combined (i.e., structural, financial, or index; cf. Table 4). Finally, our tests for non-linearity revealed that a positive linear curve type achieves the strongest, although still not significant, result (p = 0.25; cf. Table 5). Moreover, a cubic horizontal-S form, although also not significant (p = 0.38), was found to explain the highest degree of performance variance (R-square = 0.038).

Insert Figure 3 and Tables 4 and 5 about here Discussion Positive performance impact of internationalization The key objective of this study was to answer a previously unresolved but fundamental question in the international management field: Does internationalization affect firm performance? Given the obtained sample-size weighted mean effect size of r = 0.04 for the overall sample population, the answer must be that a true effect does indeed exist that is positively and statistically significant but at a low magnitude. Nevertheless, as demonstrated 21

in the simple utility analysis (Breaugh, 2003; Premack and Wanous, 1985) detailed in Table 6, even such a presumably low effect size can have a significant impact in practice. Taking as an example a large MNC currently operating at a DOI of 40% foreign sales-tototal sales, our meta-analytic findings would suggest that by expanding its international operations to a 60% DOI, the firm could increase its operating income by U.S. $ 40 million. When we apply the identified mean effect size of 0.22 for operational performance (e.g., reflecting cost efficiency), the performance implications of international expansion start to become impressive. Here, the meta-analytic results would suggest that, by increasing the DOI to 60%, the same corporation could reduce its operating costs by a notable U.S. $ 876 million. These findings corroborate Breaughs (2003) recent comment on valid effect size communication: [An effect size] even accounting for [only] 1% of the variance can result in considerable cost savings (p. 92).

Insert Table 6 about here Mediating role of operational performance Besides having aimed at answering the question of internationalizations direct effect on firm performance, we have tested for new methodological and conceptual moderator variables. With respect to methodological moderators, our most important finding is that studies using operational performance variables exhibit a significantly higher effect size than studies applying accounting and market-based conceptualization modes. That is, operational performance measures seem to be purer than financial indicators (accounting based and market based): they directly identify internationalization benefits at their place of origin and appear to be less susceptible to dilution and manipulation. This result suggests that operational performance acts as an important mediating variable between internationalization and accounting and market-based performance.9

Internationalization appears first to induce operational effectiveness as reflected, for example,

22

in cost efficiency, labour productivity, or innovation capability, which in turn positively impact firm-level profitability measures. If studies using operational performance measures have indeed investigated the internationalization-performance relationship at the most valid and reliable level, then given the identified sample-size weighted mean effect size of r = 0.22 for operational performance modes we can claim the existence of a significant positive and high-in-magnitude effect of international expansion on firms key competitive advantages that may transform into financial accounting and market-based performance at a later stage. By applying operational performance measures, future researchers may be able to open up the hitherto acknowledged but largely omitted black box of the internationalizationperformance relationship.

Effective variable operationalization The second key finding with respect to methodological moderators is that both index and financial modes of conceptualization for internationalization produce significantly higher effect sizes than the structural mode. However, we found no significant differences in effect sizes between the index and financial modes themselves. Therefore, our results do not provide a definite corroboration of previous claims that the application of multi-item composites is better than the use of one-item measures. Our results also beg the question of why the financial mode outperforms the structural mode, given that both are one-item indicators. In response, we note that the two one-item modes may be interpreted to reflect different objects of international expansion, i.e., sales and manufacturing, respectively. That is, whereas the financial mode of conceptualizing internationalization reflects foreign sales, the structural component reflects foreign manufacturing (Ramaswamy, 1992). Foreign manufacturing is associated with higher costs than are foreign sales because of needed construction or acquisition of suitable production facilities, costly knowledge transfer, intense governance and coordination, and the potential 23

for costs resulting from low labour productivity (Cooke, 2001; Larudee and Koechlin, 1999). Consequently, structural internationalization may result in lower effect sizes than financial internationalization. Most important for future research, this finding has implications for the highly debated topic of effective variable operationalization (e.g., Annavarjula and Beldona, 2000; Hoopes, 1999; Ramaswamy, 1992; Ramaswamy, Kroeck and Renforth, 1996; Sullivan, 1994a,b; Vachani, 1991). Our finding that, as implicitly assumed, differing conceptualization modes are indeed likely to reflect different conceptual logics (e.g., financial conceptualization reflects sales-driven expansion, whereas structural conceptualization reflects manufacturingcentred expansion) reinforces the argument that researchers benefit from selecting conceptualization modes that best reflect their studys conceptual logic. In doing so, they achieve content validity. In a second step, within the selected content-appropriate conceptualization mode, researchers may benefit from operationalizing their chosen concept with multi-item composites. By following this two-step approach, researchers may achieve both content and construct validity.10

Conceptual moderators Investigative time frame Studies of the internationalization-performance link during different time periods did not produce different effect sizes. Thus, our findings failed to support our expectation that the benefits of internationalization related to market imperfections may subside as time progresses. From this, one may conclude that, to this day, internationalization benefits stemming from external market imperfections are largely available around the world. Alternatively, this finding could be interpreted to mean that while advantages of internationalization emanating from global-market imperfections have indeed diminished with time, MNCs have simultaneously learned how to compensate by deriving a major share of their competitive advantage from the proactive development and worldwide 24

exploitation of intra-company resources. The latter interpretation suggests that during the 1980s and 90s, the main source of advantage from internationalization changed from benefits found externally to those generated internally. Company size Company size was identified as a conceptual moderator of the nexus. Therefore, our findings tend to confirm the conceptual claims of resource and managerial constraints faced by small-sized firms. More important, the finding of an overall negative mean effect size (r = - 0.03) indicates that small firms are not only at a minor disadvantage compared to large firms (i.e., enjoy marginally lower effect sizes than do large firms), but that small-sized MNCs face costs that generally outweigh, even though only marginally, the benefits of internationalization throughout the internationalization continuum. Nevertheless, as only eleven company samples could be tested for the moderator effect of company size, this finding should be interpreted with caution. Notwithstanding, it clearly suggests that company size should be increasingly perceived as a moderator variable rather than solely as a control variable in our research domain. Company nationality Company nationality was also found to represent an important conceptual moderator of the internationalization-performance relationship. Effect sizes were significantly higher in studies investigating European companies (r = 0.12) than in studies examining U.S. (r = 0.05) and Japanese companies (r = -0.02). This interesting observation, together with our descriptive finding that compared to U.S. and Japanese firms European MNCs have the highest average DOI, points to a potentially important contingency factor of the linkage. Assuming that the average DOI of a company sample from a specific nation reflects the internationalization tenure of the typical company in that nation, this result suggests that a firms capacity or competence in managing internationalization is largely dependent on its internationalization experience. It should also be noted that the overall negative effect size for Japanese firms may be partially due to different performance goals.

25

Non-linearity question Lastly, we tested for non-linearity. At the meta-level, we were able to perform regression analysis testing for both quadratic and cubic curves. Even though not statistically significant, a cubic curve (horizontal-S) explained the highest degree of performance variance and therefore may arguably best reflect the form of internationalization-performance relationship at the aggregate level. However, this proposition should be treated with caution. First, this curve type was not statistically supported at a significant level. Second, as already mentioned, our study sample was biased towards U.S. companies. Unfortunately, due to the rather small number of European and Japanese companies in our dataset, we were unable to run a separate regression analysis for them. However, given our finding of significantly higher effect sizes for the European companies sampled than for the U.S. companies studied, there appear to be strong boundary conditions that question the existence of one universalistic form of internationalization-performance relationship across company nationalities.

Future research directions Our meta-analytic results indicate that the form of the internationalization-performance relationship is context dependent, an issue barely addressed in the literature to date. In particular, the findings suggest that firms located in different contextual settings (in terms of company size and company nationality) are likely to choose or be obliged to pursue idiosyncratic strategic actions that, in turn, moderate the internationalization-performance relationship.11 For example, organizations execute divergent scales (low vs. high), scopes (culturally related vs. culturally unrelated), modes (wholly owned vs. partially owned) and goals of internationalization (financial vs. operational). Further, our meta-analytic results can be interpreted to indicate that firms choosing differing objects of internationalization (e.g., sales vs. manufacturing) experience differing internationalization-performance relationships. Finally, recent inquiry (i.e., Vermeulen and Barkema, 2002) has revealed that organizations

26

pursuing divergent paces and rhythms of international expansion (evolutionary vs. revolutionary; stable vs. erratic) encounter different internationalization-performance nexus. Below, we integrate these potentially relevant moderator elements (i.e., scale, scope, mode, object, goal, and pace and rhythm) into an exploratory framework for future inquiry (cf. Figure 4). However, it is important to note that the proposed moderators are not exhaustive; our purpose is to be parsimonious, illustrative, and stimulating. Yet we also hope that the components will not be perceived as a mundane checklist. Rather, our goal was to develop a parsimonious variable system, largely deduced from the results of our metaanalysis, that may help to induce new theoretical development. We believe that the selected elements are likely to be requisite components for any overall explanatory model of the organizational phenomenon under investigation.

Insert Figure 4 about here Scale of internationalization Researchers throughout the last decade have intensely debated the moderating impact of firm scale or degree of internationalization on the internationalization-performance relationship and constructed differing forms of the link (the linear, quadratic, and cubic curves already discussed). However, despite first attempts at reconciliation, consensus about a universalistic form for the link is as yet nonexistent. As noted above, our meta-analysis provides only weak support for such a form. Our results do indicate that a horizontal-S form may best reflect the link at an aggregate level i.e., across time, firm size, and variable operationalization; however, our meta-analytic database was biased towards U.S. firms and, given our finding of significantly different effect sizes for studies investigating different company nationalities, this universalistic form of the DOIperformance relationship at the aggregate level appears to break down in particular national settings.

27

Yet the horizontal-S form obtained for a company population biased towards U.S. firms may be a starting point from which to proceed with future inquiry. This particular curve type may indeed represent the form of the internationalization-performance nexus for firms operating in a particular contextual setting, particularly considering that scale of internationalization may be a good proxy for a firms internationalization experience (Herrmann and Datta, 2002). Given our finding that the average U.S. firm exhibits a rather moderate internationalization scale, the horizontal-S form may reflect the

internationalization-performance relationship typical of firms whose moderate expansion experience allows them to successfully handle the initial entrance costs of internationalization (Type 1) but not the Type 2 costs emerging at high scales of foreign expansion. This logic then raises the following question: Do the different curve types identified by previous research represent differing learning curves? Future researchers may address this question by investigating the moderating impact of internationalization experience (e.g., operationalized with company samples average scale of internationalization) on the internationalizationperformance relationship. Scope of internationalization With respect to scope of internationalization, we refer to culturally related versus culturally unrelated expansion strategies. Firms may either choose or be strongly bound to follow one option. For example, large MNCs headquartered in smallsized European nations (e.g., Sweden, Switzerland, the Netherlands) and Japan were forced to pursue culturally unrelated internationalization early in their existence (Ruigrok and Wagner, 2003). In contrast, firms headquartered in a country with a large home market or who are able to expand into large markets in culturally related nations can choose to follow one of the two scope alternatives. If choice is possible, the psychic distance paradigm (Johanson and Vahlne, 1977) suggests that firms initially follow culturally related strategies and then, after having gained experience, widen their scope towards culturally unrelated markets.

28

Firms forced to pursue a culturally unrelated internationalization strategy at the outset may be at a disadvantage compared to those able to consciously pick a suitable expansion scope. However, previous research findings on the inefficiency of culturally unrelated expansion are inconclusive (e.g., Evans and Mavondo, 2002; Gomez-Mejia and Palich, 1997) because high costs of unrelated expansion (e.g., coordination, monitoring, and motivation costs) may be outweighed, or at least off-set, by unique benefits of culturally unrelated markets (e.g., knowledge development through workforce cultural heterogeneity). Therefore, future researchers may aim at elaborating a potential moderating effect of cultural distance on the internationalization-performance relationship and, in doing so, help to decipher this psychic distance paradox. Mode of internationalization Companies expanding into foreign markets must choose an entry mode. The research stream being restricted to foreign direct investment, modes of entry addressed here are confined to greenfields and acquisitions that can be either wholly or partially owned. This choice of whole or partial ownership (i.e., control mode) can be largely driven by company nationality and firm size. Japanese firms, for example, were found to favour whollyowned greenfield investments (Cho and Padmanabhan, 1995) because wholly-owned new ventures allow firms to instantly take charge of their own competitive advantages. In the case of partially-owned entry modes, merging the two partners differing but strongly implanted behavioural routines, cultures, and infrastructures is time consuming and comes along with considerable coordination and establishment costs (Barkema and Vermeulen, 1998). Such costs are likely to be particularly significant when the partner is located in a culturally distant business environment (Weber, Shenkar and Raveh, 1996). Thus, by starting wholly-owned subsidiaries, internationalizing firms can avoid these costs. In contrast, small-sized firms were found to engage predominantly in partially-owned modes of foreign expansion (i.e., equity joint ventures), perhaps because by setting up a local 29

firm with a local partner knowledgeable about the foreign social, business, and regulatory environments, small-sized firms can compensate for their financial and human capital constraints (Inkpen and Beamish, 1997). This advantage may hold particularly true in the case of expansion into a culturally unrelated market in which cooperation with a local firm allows small firms to bypass the high set-up costs of wholly-owned ventures. In addition, this collaboration with a local partner may hasten the expanding firms market legitimacy by reducing governmental and consumer aversion. Thus, overall, the divergent modes of internationalization that firms choose are associated with differing costs and benefits and therefore result in idiosyncratic internationalization-performance relationships. Object of internationalization By objects of internationalization, we mean primary and support organizational value-chain activities (Porter, 1985) procurement and logistics, manufacturing, sales and marketing, after-sales service, research and development, finance and controlling, human resources, and infrastructure one, several, or all of which firms may internationalize at different depths (Ramaswamy, 1992). As outlined earlier in this discussion, our findings can be interpreted to indicate that the two objects of internationalization manufacturing and sales result in differing effect sizes for the internationalizationperformance link. Such differentiation can be explained by the differing costs and benefits accompanying different objects of internationalization. For example, whereas sales expansion is primarily associated with benefits stemming from economies of scale and with costs originating from increasing marketing and promotion demands, manufacturing expansion is predominantly related to benefits derived from increasing operating efficiency (i.e., lower transportation, material, and labour costs) and to costs stemming from employee training and knowledge transfer. Such idiosyncratic benefits and costs of internationalization objects are likely to cause differing performance impacts of foreign expansion. Thus, future research may benefit from investigating the moderating impact of the object of internationalization on the internationalization-performance link. 30

Goal of internationalization Our meta-analysis also indicated that firms may pursue divergent goals of international expansion. In particular, our findings can be interpreted as corroborating previous research results suggesting that Japanese firms do not place priority on short-term accounting performance goals but rather follow sustainable operational performance objectives. Therefore, company nationality can be associated with distinctive objectives sought through international expansion that are likely to moderate the internationalization-performance relationship. Like companies national membership, their business membership may be considered a major reason for divergent performance objectives. For example, due to their industries technology intensiveness, a major goal of internationalization for companies operating in the pharmaceutical and automotive businesses has been the redemption of high research and development costs (Ramaswamy, 1995). In contrast, due to the labour intensiveness of the industrys product manufacturing, the primary objective of internationalization in the clothing business has been labour cost reduction (Porter, 1985, 1990). Therefore, future researchers should be aware that divergent performance goals may result in differing internationalizationperformance relationships. Pace and rhythm of internationalization Recent research has found that firms with varying paces and rhythms of internationalization encounter differing internationalizationperformance relationships. In particular, Vermeulen and Barkema (2002) hypothesized and empirically confirmed that a fast internationalization pace and unsystematic expansion pattern may overexert the absorptive capacity of those steering the company and therefore negatively impact firm performance. It was concluded that firms may benefit from applying evolutionary and stable international expansion but fail to apply revolutionary and irregular expansion ways. Nevertheless, it may be argued that this implication might not hold strongly in different contextual settings. For example, absorptive capacity constraints may constitute a less 31

threatening burden for firms with a high level of internationalization experience because experiential knowledge may enable such firms to effectively manage fast and erratic expansion. Hence, costs stemming from fast and irregular expansion patterns may be particularly incisive for firms in early stages of internationalization but not for those having decades of internationalization experience and established competence. An important implication of this last point is that the suggested building blocks of the research framework for future inquiry are not independent of each other. For example, certain scales and concomitant experience of internationalization are likely to spur certain modes, scopes, and paces and rhythms of foreign expansion. Thus, a firm with a high level of experience of internationalization may be more prone to succeed in wholly-owned greenfield venturing in culturally unrelated markets at a revolutionary pace than a firm with a low level of experience. Further, certain internationalization paces may be more successful with certain expansion modes. For example, a revolutionary pace may be more likely to succeed with acquisition than with a time-consuming start-up. Finally, certain internationalization goals may be related to certain modes of internationalization. A firm aiming at organizational learning (i.e., knowledge development) through a culturally heterogeneous workforce may be more likely to achieve this goal by pursuing a joint venture mode, enabling the firm to draw upon the local partners knowledge. On the other hand, MNCs aiming at redeeming high research and development expenditures through scale economies may benefit more from starting up wholly-owned ventures that help them to avoid unwanted diffusion of technological knowledge. Thus, by addressing the interrelatedness of moderator variables, researchers may be able to explore the complex internationalization-performance relationship at a more comprehensive, fine-tuned, and reflective level.

32

Limitations In interpreting our findings, we must be aware of the limitations of meta-analysis. In particular, meta-analytic approaches cannot answer questions of causality. For instance, they cannot indicate whether the direction of causality for the internationalization-performance relationship of MNCs with outstanding performance and a large investment volume that may be used for further international expansion is two-sided rather than one-sided. Future research may increasingly apply longitudinal designs to resolve this issue. A further limitation of meta-analysis in general is related to sample interdependence. If studies undertaken by one author or group dominate the study population, findings can be biased. However, of the 62 studies in our meta-analysis, not more than six were undertaken by the same author or group of authors. Therefore, sample interdependence was not a serious problem for our meta-analysis. As regards our particular methodology, the use of multiple variable conceptualizations in single studies (with a sample-to-study ratio of 2.81) may have caused statistical bias. Yet restricting ourselves to a single effect size per study or averaging effect sizes across multiple variable conceptualizations to one composite per study would have prevented the empirical identification of methodological moderators. Our approach not only utilizes the available data to the greatest extent but has been widely applied by many other researchers (e.g., Palich, Cardinal and Miller, 2000; Dalton et al., 1999; Dalton, Daily, Certo and Roengpitya, 2003). In addition, our use of the FSTS measure (i.e., the financial mode of conceptualizing internationalization) to test for non-linearity at the aggregate level may have led to another bias. That is, our meta-analysis of non-linearity for the internationalization-performance relationship may be skewed towards sales-oriented expansion and may not strongly mirror production-oriented expansion (as reflected by the structural mode of conceptualizing internationalization). Future inquiry may explicitly test for divergent curve types for

33

companies pursuing a sales-oriented strategy as compared to those following a productionoriented expansion strategy. Our exploratory research framework for future inquiry, besides not being exhaustive, may also face a content limitation in that it assumes that managers can shape the internationalization-performance relationship by making certain strategic choices. Yet, in the presence of certain external inevitabilities (e.g., economic crises, social and industrial trends and cycles, or legal and political developments), this power of influence or latitude may be largely restricted. However, to further understand how managers can consciously and proactively shape the internationalization-performance relationship, we believe our focus should be on strategic action variables, particularly as previous researchers (cf. Hansen and Wernerfelt, 1989; Powell, 1996) have found that strategic action variables generally account for twice as much of a companys performance variance as do uncontrollable external environmental factors. Moreover, we are aware that it may be difficult or perhaps even impossible for individual studies to address the complete research framework we propose here. Nevertheless, we believe the mere recognition of an eclectic moderator frame to be of considerable value for researchers that are, as in our research domain, scattered across multiple and largely independently operating fields. As noted by Ramanujam and Varadarajan (1989), research conducted with a clear awareness of concepts and linkages excluded by conscious choice is more likely to be richer than ad-hoc examinations of narrowly circumscribed models tied to the disciplinary binds of a particular field (p. 545).

Conclusions This study aimed at making three important contributions to the literature on the internationalization-performance relationship. The first goal was to synthesize the voluminous and diffused study pool on the internationalization-performance relationship

34

accumulated across time, nations, and research disciplines. Specifically, our meta-analytic results help to resolve a long-lasting research controversy by showing that internationalization has a non-zero, positive effect on firm performance. In addition, even though the mean effect size appears to be marginal at the aggregate level, utility analysis shows how impressive profitability benefits and cost savings can be. Our second objective was to explore the existence of moderator variables. The results show that the performance impact of internationalization is indeed context dependent, with company size and company nationality, in particular, representing key macro-level moderators. Therefore, although meta-analytic synthesis of previous inquiry has allowed us to draw important mid-term conclusions, our findings clearly emphasize that we have only just arrived at a challenging but potentially fertile starting point for future inquiry. Our third and final objective was to sensitize extant and new members of the research stream to promising new research directions and so stimulate further investigation. By developing an exploratory framework, we integrated potentially important interrelated moderator variables of the internationalization-performance relationship (scale, scope, mode, object, goal, and pace and rhythm of internationalization) in the hope that application of such a programmatic contextual perspective will lead to an admittedly more complex but potentially more coherent, holistic, and managerially relevant understanding of the performance impact of corporate international expansion.

Acknowledgements The authors would like to thank Departmental Editor Tarun Khanna as well as three anonymous reviewers for their helpful comments and guidance. We also gratefully acknowledge constructive suggestions from Wolfgang Amann, James Breaugh, Devereaux Jennings, Hansueli Keller, Claas van der Linde, Simon Peck, Micala Raschle, and Robert Straw for earlier drafts of this article. Any remaining errors are our own. 35

Endnotes
1

Formulas necessary for conversion can be found in Cohen (1977), Fern and Monroe (1996), Hunter and Schmidt (1990), Rosenthal (1984), and Wolf (1986).
2

Articles were taken from the Journal of International Business Studies, Management International Review, International Business Review, Journal of International Management, Academy of Management Journal, Strategic Management Journal, Journal of Management, Journal of Business, Journal of Business Research, Journal of Finance, Journal of Financial and Quantitative Analysis, Managerial Finance, Journal of Business Finance and Accounting, Quarterly Review of Economics and Finance, Journal of Economic Studies, Review of Economics and Statistics, Oxford Bulletin of Economics and Statistics, and the Akron Business and Economic Review.
3

For this analysis, our sample population was split by the median year investigated by the 64 studies. The investigative time frame of individual studies that examined the internationalization-performance relationship for multiple years was estimated by the mean year of investigation. In the identified articles, the company characteristic size was predominantly defined in terms of employee number. Cut-off points separating small-sized from medium- and largesized firms ranged between 250 and 500 employees (cf. Lu and Beamish, 2001; Majocchi and Zucchella, 2001).
5 4

This average number can be calculated with the numbers quoted in Table 2, Section Degree of internationalization, Row Overall (the sum of 35, 50, and 29 divided by three equals 38).

This FSTS ratio for European firms is estimated conservatively since the European sample population includes, in contrast to the other two sample populations, investigations of the internationalization-performance relationship in the 1970s. When we exclude studies that investigated the link in question before 1980 across our complete study population, the DOI gap between European firms (then average FSTS = 63%) and its regional counterparts becomes even more significant (average DOIs for U.S. and Japanese firms remain the same, i.e., 35% and 29%, respectively). The stem-and-leaf display indicates that the effect-size distribution approximates a normality function. A Kolmogorov-Smirnov test for normality distribution strongly confirmed this impression (p < .001). This result suggests that the effect sizes extracted from our study population are indeed representative of the research streams findings.
8 7

Findings are available upon request.

We thank an anonymous reviewer for underlining the important role of operational performance (e.g., market share, cost efficiency, labour productivity, and innovation capability) as a mediator between internationalization and financial performance (i.e., firmlevel accounting performance and market-based performance). To date, researchers following the proposed two-step approach when developing multi-item composites are few. More commonly, investigators merge measures drawn from the structural and financial conceptualization modes into one index. As indicated, this approach may hide certain content-bound performance impacts of international expansion. 36
10

Besides company size and company nationality, a further key macro-level contextual gestalt may be company industry (e.g., manufacturing vs. service). However, because of a lack of studies that investigated the service sector (two notable exceptions being Capar and Kotabe, 2003 and Contractor et al., 2003) in the research stream, we were unable to test for this moderator effect in the meta-analysis.

11

37

References *Aggarwal, R. (1979) Multinationality and stock market valuation: an empirical study of U.S. markets and companies, Management International Review 19: 5-12. AlNajjar, F.K. and Riahi-Belkaoui, A. (1999) Multinationality, profitability and firm value, Managerial Finance 25: 31-41. *AlNajjar, F.K. and Riahi-Belkaoui, A. (2001) Empirical validation of a general model of growth opportunities, Managerial Finance 27: 72-90. *Annavarjula, M. G. (1998), An examination of factors influencing the relationship between the degree of multinationality and performance of multinational corporations Unpublished doctoral dissertation, Temple University. Annavarjula, M. G. and Beldona, S. (2000) Multinationality-performance relationship: a review and reconceptualization, The International Journal of Organizational Analysis 8: 48-67. *Athanassiou, N. and Nigh, D. (1998) The MNCs international strategy, tacit knowledge dimensions of its top management teams character, and international performance Paper presented at the Academy of International Business conference 1998, Vienna. Bacharach, S. B. (1989) Organizational theories: some criteria for evaluation, Academy of Management Review 14: 496-515. Bae, S. C. and Jain, V. (2003) The effects of RandD and international expansion on the performance of U.S. industrial firms, Corporate Finance Review 8: 28-39. *Barkema, H. G. and Vermeulen, F. (1998) International expansion through start-up or acquisition: a learning perspective, Academy of Management Journal 41: 7-26. Barney, J. B. (1991) Firm resources and sustained competitive advantage Journal of Management 17: 99-120. *Benvignati, A. M. (1987) Domestic profit advantages of multinational firms, Journal of Business 60: 449-461. Berry, H. (2001) When does multinationality increase firm value? Evidence from US and Japanese firms, 1974-1997 Working paper, University of Pennsylvania. *Best, A. D. (1997) International diversity versus international diversification: host country effects on multinational firm performance Unpublished doctoral dissertation, Texas AandM University. *Black, S. S. (1997) Top management team characteristics: a study of their impact on the magnitude of international operations and international performance Unpublished doctoral dissertation, Columbia University. Bloodgood, J. M., Sapienza, H. J. and Almeida, J. G. (1996) The internationalization of new high-potential U.S. ventures: antecedents and outcomes, Entrepreneurship Theory and Practice 20: 61-76. Boddewyn, J. J. (1988) Political aspects of MNE theory, Journal of International Business Studies 19: 341-363. Bodnar, G. M., Tang, C. and Weintrop, J. (2000) Both sides of corporate diversification: the value impacts of geographic and industrial diversification Working paper, Johns Hopkins University. Breaugh, J. A. (2003) Effect size estimation: factors to consider and mistakes to avoid, Journal of Management 29: 79-97. Brewer, T. L. (1992) An issue-area approach to the analysis of MNE-government relations, Journal of International Business Studies 23: 295-309. *Broaden, C. B. and Samii, M. V. (2001) Measuring the impact of globalization: an analysis of the risk and return of multinational firms Working paper, Southern New Hampshire University.

38

Buckley, P. J. and Casson, M. C. (1976) The future of the multinational enterprise, London: Macmillan. *Buehner, R. (1987) Assessing international diversification of West German corporations, Strategic Management Journal 8: 25-37. *Capar, N. (2001) An empirical examination of the form of the relationship between international diversification and performance: evidence from German multinational firms Paper presented at the annual meeting of the Academy of Management 2001, Washington. *Capar, N. and Kotabe, M. (2003) The relationship between international diversification and performance in service firms, Journal of International Business Studies 34: 345-355. *Carpenter, M. A. (1997) Top management team characteristics, global strategic change, and firm performance: the moderating role of organizational and environmental context Unpublished doctoral dissertation, University of Texas at Austin. *Carpenter, M. A. and Fredrickson, J. W. (2001) Top management teams, global strategic posture, and the moderating role of uncertainty, Academy of Management Journal 11: 533-545. *Carpenter, M. A., Sanders, WM. G. and Gregersen, H. B. (2001) Bundling human capital with organizational context: the impact of international assignment experience on multinational firm performance and CEO pay, Academy of Management Journal 11: 493-511. *Carpenter, M. A. and Wade, J. B. (2002) Microlevel opportunity structures as determinants of non-CEO executive pay, Academy of Management Journal 45: 1085-1103. *Carpenter, M. A. (2002) The implications of strategy and social context for the relationship between top management team heterogeneity and firm performance, Strategic Management Journal 23: 275-284. *Carpenter, M. A. and Sanders, WM. G. (2004) The effects of top management team pay and firm internationalization on MNC performance, Journal of Management, forthcoming. Caves, R. E. (1971) Industrial corporations: the industrial economics of foreign investment, Economica 38: 1-27. *Chang, Y. and Thomas, H. (1989) The impact of diversification strategy on risk-return performance, Strategic Management Journal 10: 271-284. Cho, K. R. and Padmanabhan, P. (1995) Acquisition versus new venture: The choice of foreign establishment mode by Japanese firms, Journal of International Management 1: 255-285. Christophe, S.E. (1997) Hysteresis and the value of the U.S. multinational corporation, Journal of Business 70: 435-462. Click, R. W. and Harrison, P. (2000) Does multinationality matter? Evidence of value destruction in U.S. multinational corporations Working paper, George Washington University. Cohen, J. (1977) Statistical power analysis for the behavioral sciences, New York: Academic Press. Contractor, F. J., Kundu, S. K. and Hsu, C. (2003) A three-stage theory of international expansion: the link between multinationality and performance in the service sector, Journal of International Business Studies 34: 5-18. Coviello, N. E. and McAuley, A. (1999) Internationalisation and the smaller firm: a review of contemporary empirical research, Management International Review 39: 223-256. Dalton, D. R., Daily, C. M., Johnson, J. L. and Ellstrand, A. E. (1999) Number of directors and financial performance: a meta-analysis, Academy of Management Journal 42: 674686.

39

Dalton, D. R., Daily, C. M., Certo, S. T. and Roengpitya, R. (2003) Meta-analyses of financial performance and equity: fusion or confusion?, Academy of Management Journal 46: 13-26. Daniels, J. D. and Bracker, J. (1989) Profit performance: do foreign operations make a difference?, Management International Review 29: 46-56. Davidson, W. H. (1980) The location of foreign investment activity, Journal of International Business Studies Fall: 9-23. Davidson, W. H. (1983) Market similarity and market selection: implications for international marketing strategy, Journal of Business Research 11: 439-456. *Delios, A. and Beamish, P. W. (1999) Geographic scope, product diversification, and the corporate performance of Japanese firms, Strategic Management Journal 8: 711-727. *Delios, A. and Beamish, P. W. (2000) Multinationality and performance in Japanese firms, in D. Tse, C. M. Lau, K. Law and C. S. Wong (eds.) Asian Business, Englewood Cliffs, NJ: Prentice-Hall, pp 61-78. Denis, D.J., Denis, D.K. and Yost, K. (2002) Global diversification, industrial diversification, and firm value, Journal of Finance 57: 1951-1979. Doukas, J.A and Lang, L.H.P. (2003) Foreign direct investment, diversification and firm performance, Journal of International Business Studies 34: 153-172. Dragun, D. (2002) Challenging the rhetoric: internationalisation, size and financial performance, European Retail Digest 36: 25-33. *Eden, L. and Thomas, D. E. (2000) Is the multinationality-performance relationship Ushaped? Paper presented at the annual meeting of the Academy of Management 2000, Toronto. Eden, D. (2002) Replication, meta-analysis, scientific progress, and AMJs publication policy, Academy of Management Journal 45: 841-846. *Ellstrand, A. E., Tihanyi, L. and Johnson, J. L. (2002) Board structure and international political risk, Academy of Management Journal 45: 769-777. Eriksson, V. K., Johanson, J., Majkgard, A. and Sharma, D. D. (2001) Time and experience in the internationalization process, Zeitschrift fr Betriebswirtschaft 71: 21-43. *Errunza, V. R. and Senbet, L. W. (1984) International corporate diversification, market valuation, and size-adjusted evidence, Journal of Finance 34: 727-745. Evans, J. and Mavondo, F. T. (2002) Psychic distance and organizational performance: an empirical examination of international retailing operations, Journal of International Business Studies 33: 515-532. Fatemi, A. M. (1984) Shareholder benefits from corporate international diversification, Journal of Finance 39: 1325-1344. Fayerweather, J. (1978) International Business Strategy and Administration, Cambridge, MA. Fern, E. F. and Monroe, K. B. (1996) Effect-size estimates: issues and problems in interpretation, Journal of Consumer Research 23: 89-105. Garrod, N. and Rees, W. (1998) International diversification and firm value, Journal of Business Finance and Accounting 25: 1255-1281. Geringer, M. J., Beamish, P. W. and daCosta, R. C. (1989) Diversification strategy and internationalization: implications for MNE performance, Strategic Management Journal, 10: 109-119. *Geringer, M. J., Tallman, S. and Olsen, D. M. (2000) Product and international diversification among Japanese multinational firms, Strategic Management Journal 21: 51-80. Gomes, L. (1997) Multinationality and performance: the role of cultural dissimilarity Unpublished doctoral dissertation, Florida International University.

40

*Gomes, L. and Ramaswamy, K. (1999) An empirical examination of the form of the relationship between multinationality and performance, Journal of International Business Studies 30: 173-188. Gomez-Mejia, L. R. and Palich, L. E. (1997) Cultural diversity and the performance of multinational firms, Journal of International Business Studies 2: 309-335. *Grant, R. M. (1987) Multinationality and performance among British manufacturing companies, Journal of International Business Studies 18: 79-89. *Grant, R. M., Jammine, A. and Thomas, H. (1988) Diversity, diversification, and profitability among British manufacturing companies, 1972-84, Academy of Management Journal 31: 771-801. *Haar, J. (1989) A comparative analysis of the profitability performance of the largest U.S., European and Japanese multinational enterprises, Management International Review, 29: 5-18. Han, K.C., Lee, S.H. and Suk, D.Y. (1998) Multinationality and firm performance, Multinational Business Review 6: 63-70. Hansen, G. S. and Wernerfelt, B. (1989) Determinants of firm performance: the relative importance of economic and organizational factors, Strategic Management Journal 10: 399-411. *Harveston, P. D., Kedia, B. L. and Francis, J. D. (1999) MNEs dependence on foreign operations and performance: a study of MNEs from the Triad regions, International Business Review 8: 293-307. Hempel, C. (1965) Aspects of scientific explanation, New York: Free Press. *Herrmann, P. and Datta, D. K. (2002) CEO successor characteristics and the choice of foreign market entry mode: an empirical study, Journal of International Business Studies 33: 551-569. Hirsch, S. and Lev, B. (1971) Sales stabilization through export diversification, Review of Economics and Statistics August: 258-266. *Hirschey, M. (1982) Market power and foreign involvement by U.S. multinationals, Review of Economics and Statistics, 64: 343-346. *Hitt, M. A., Hoskisson, R. E. and Kim, H. (1997) International diversification: effects on innovation and firm performance in product-diversified firms, Academy of Management Journal 40: 767-798. Hofstede, G. (1980) Cultures consequences: international differences in work-related values, Beverly Hills, CA: Sage Publications. Hoopes, D. G. (1999) Measuring geographic diversification and product diversification, Management International Review 39: 277-292. Hoskisson, R. E., Hitt, M. A., Wan, W. P. and Yiu, D. (1999) Theory and research in strategic management: Swings of a pendulum, Journal of Management 25: 417-456. Hunter, J. E., Schmidt, F. L. and Jackson, G. B. (1982) Meta-analysis: cumulating research findings across studies, Beverly Hills, CA: Sage Publications. Hunter, J. E. and Schmidt, F. L. (1990) Methods of Meta-Analysis, Newbury Park, CA: Sage Publications. Hymer, S. H. (1976) The international operations of national firms: a study of foreign direct investment, Cambridge, MA: MIT Press. Inkpen, A. C. and Beamish, P. W. (1997) Knowledge, bargaining power, and the instability of international joint ventures, Academy of Management Review 22: 177-202. Johanson, J. and Vahlne, J. (1977) The internationalization process of the firm: a model of knowledge development and increasing foreign market commitments, Journal of International Business Studies 8: 23-32. Johansson, J. K. and Yip, G. S. (1994) Exploiting globalization potential: U.S. and Japanese strategies, Strategic Management Journal 15: 579-601. 41

Jones, G. R. and Hill, C. W. L. (1988) Transaction cost analysis of strategy-structure choice, Strategic Management Journal 9: 159-172. *Jung, Y. (1991) Multinationality and profitability, Journal of Business Research 23: 179187. Kaplan, A. (1964) The conduct of inquiry, New York: Harper and Row. Kilduff, M., Angelmar, R. and Mehra, A. (2000) Top management team diversity and firm performance: examining the role of cognitions, Organization Science 11: 21-34. Kogut, B. (1985) Designing global strategies: Profiting from operational flexibility, Sloan Management Review 27: 27-38. Kogut, B. and Singh, H. (1988) The effect of national culture on the choice of entry mode, Journal of International Business Studies 19: 411-432. Kotabe, M., Srinivasan, S. S. and Aulakh, P. S. (2002) Multinationality and firm performance: the moderating role of RandD and marketing capabilities, Journal of International Business Studies 33: 79-97. *Kumar, M. S. (1984) Comparative analysis of UK domestic and international firms. Journal of Economic Studies 11: 26-42. Lawrence, B. S. (1997) The black box of organizational demography, Organization Science 8: 1-22. Lessard, D. R. (1973) International portfolio diversification: a multivariate analysis for a group of Latin American countries, Journal of Finance 28: 610-633. Lessard, D. R. (1976) World, country, and industry relationships in equity returns: implications for risk reduction through international diversification, Financial Analysts Journal 32: 32-38. Levitt, T. (1983) The globalization of markets, Harvard Business Review 61: 92-102. Levy, H. and Sarnat, M. (1970) International diversification and investment portfolios, American Economic Review September: 668-675. *LHeureux, R. A. (1994) The effect of internationalization on the relationship between TMT characteristics and firm performance Unpublished doctoral dissertation, The Florida State University. *Lu, J. W. and Beamish, B. W. (2001) The internationalization and performance of SMEs, Strategic Management Journal 22: 565-586. Luo, Y. and Peng, M. W. (1999) Learning to compete in a transition economy: experience, environment, and performance, Journal of International Business Studies 30: 269-296. *Majocchi, A. and Zucchella, A. (2001) Internationalization and performance: findings from a set of Italian SMEs Working paper, University of Insubria. Makhija, M., Kwongsoo, K. and Williamson, S. (1997) Measuring globalization of industries using a national industry approach, Journal of International Business Studies 28: 679710. Martinez, I. (2002) The value impact of internationalization: theory and evidence, the French case, Managerial Finance 28: 73-82. Martinussen, M. and Fjukstad, B. (1997) MetaDos: a computer program for meta-analysis calculations, Version 2, Faculty of Social Science, University of Troms, Norway. Mathur, I., Singh, M. and Gleason, K. C. (2001) The evidence from Canadian firms on multinational diversification and performance, Quarterly Review of Economics and Finance 41: 561-578. *Mauri, A. J. and Sambharya, R. B. (2001) The impact of global integration on MNE performance: evidence from global industries, International Business Review 10: 441454. Mayer, M. and Whittington, R. (2003) Diversification in context: a cross-national and crosstemporal extension, Strategic Management Journal 24: 773-781.

42

McEvoy, G. M. and Cascio, W. F. (1987) Do good or poor performers leave? A metaanalysis of the relationship between performance and turnover, Academy of Management Journal 30: 744-762. Miller, D. and Friesen, P. (1980) Archetypes of organizational transitions, Administrative Science Quarterly 25: 268-299. Miller, K. I. and Monge, P. R. (1986) Participation, satisfaction, and productivity: a metaanalytic review, Academy of Management Journal 29: 727-753. *Mishra, C. S. and Gobeli, D. H. (1998) Managerial incentives, internalization, and market valuation of multinational firms, Journal of International Business Studies 29: 583598. Mitchell, W., Shaver, M. J. and Yeung, B. (1992) Getting there in a global industry: impacts on performance of changing international presence, Strategic Management Journal 13: 419-432. *Morck, R. and Yeung, B. (1991) Why investors value multinationality?, Journal of Business 64: 165-187. Morck, R. and Yeung, B. (1992) Internalization: an event study test, Journal of International Economics 33: 41-56. *Nazar, V. M. (1999) The degree of internationalization and multinational corporations profitability Unpublished doctoral dissertation, Nova Southeastern University. Nehrt, C. (1999) Degree of foreign sales and foreign assets as joint indicators of firm performance Paper presented at the ABAS international conference 1999, Barcelona. *Newbert, S. L. and Contractor, F. J. (2001) The relationship between multinationality, size, and performance: a research note and empirical examination Paper presented at the annual meeting of the Academy of Management 2001, Washington. Nonaka, I. and Okumura, A. (1984) A comparison of management in American, Japanese and European firms, Management Japan 17: 23-29. Palich, L. E., Carini, G.R. and Seaman, S. L. (2000) The impact of internationalization on the diversification-performance relationship: a replication and extension of prior research, Journal of Business Research 48: 43-54. Palich, L. E., Cardinal, L. B. and Miller, C. C. (2000) Curvilinearity in the diversificationperformance linkage: an examination of over three decades of research, Strategic Management Journal 21: 155-174. Phene, A. and Nehrt, C. (2000) Are all your eggs in one basket? Diversified knowledge networks and innovatory performance Paper presented at the annual meeting of the Academy of Management 2000, Toronto. Porter, M. E. (1985) Competitive advantage: creating and sustaining superior performance, NY: Free Press. Porter, M. E. (1990) The competitive advantage of nations, NY: Free Press. Powell, T. C. (1996) How much does industry matter? An alternative empirical test Strategic Management Journal 17: 323-334. Prahalad, C. K. and Hamel, G. (1990) The core competence of the corporation, Harvard Business Review 68: 79-91. Premack, S. L. and Wanous, J. P. (1985) A meta-analysis of realistic job preview experiments, Journal of Applied Psychology 70: 706-719. Qian, G. (1997) Assessing product-market diversification of US firms, Management International Review 37: 127-149. Qian, G. (1998) Determinants of profit performance for the largest U.S. firms 1981-92, Multinational Business Review 6: 44-51. Qian, G. and Li, J. (1998) Multinationality, global market diversification, and risk performance for the largest U.S. firms, Journal of International Management 4: 149170. 43

*Qian, G. (2002) Multinationality, product diversification, and profitability of emerging US small- and medium-sized enterprises, Journal of Business Venturing 17: 611-633. *Qian, G. and Li, L. (2003) Profitability of small- and medium-sized enterprises in high-tech industries: the case of the biotechnology industry, Strategic Management Journal 24: 881-887. Ramanujam, V. and Varadarajan, P. (1989) Research on corporate diversification: A synthesis, Strategic Management Journal 10: 523-551. Ramaswamy, K. (1992) Multinationality and performance: a synthesis and redirection, Advances in International Comparative Management 7: 241-267. *Ramaswamy, K. (1995) Multinationality, configuration, and performance: a study of MNEs in the US drug and pharmaceutical industry, Journal of International Management 1: 231-253. Ramaswamy, K., Kroeck, K. G. and Renforth, W. (1996) Measuring the degree of internationalization of a firm: a comment, Journal of International Business Studies 27: 167-178. *Ramirez-Aleson, M. and Espitia-Escuer, M. A. (2001) The effect of international diversification strategy on the performance of Spanish-based firms during the period 1991-1995, Management International Review 41: 291-315. Reeb, D. M., Kwok, C. C. Y. and Baek, Y. H. (1998) Systematic risk of the multinational corporation, Journal of International Business Studies 29: 263-279. Riahi-Belkaoui, A. (1996) Internationalization, diversification strategy and ownership structure: implications for French MNE performance, International Business Review 5: 367-376. Riahi-Belkaoui, A. (1999) The degree of internationalization and the value of the firm: theory and evidence, Journal of International Accounting, Auditing and Taxation 8: 189-196. Riahi-Belkaoui, A. and Picur, R.D. (1998) Multinationality and profitability: the contingency of the investment opportunity set, Managerial Finance 24: 3-14. Riahi-Belkaoui, A. (2002a) The effects of multinationality on earnings response coefficients, Managerial Finance 28: 97-106. *Riahi-Belkaoui, A. (2002b) Multinationality as a determinant of earnings persistence, Managerial Finance 28: 83-96. Ronen, S. and Shenkar, O. (1985) Clustering countries on attitudinal dimensions: a review and synthesis, Academy of Management Review 10: 435-454. Rosenthal, R. (1979) The file drawer problem and tolerance for null results, Psychological Bulletin 86: 638-641. Rosenthal, R. (1984) Meta-analytic procedures for social research, Beverly Hills, CA: Sage Publications. *Roth, K. (1995) Managing international interdependence: CEO characteristics in a resource-based framework, Academy of Management Journal 38: 200-231. Roth, K. and ODonnell, S. (1996) Foreign subsidiary compensation strategy: an agency theory perspective, Academy of Management Journal 39: 678-703. *Ruigrok, W. and Wagner, H. (2003) Internationalization and performance: an organizational learning perspective, Management International Review 43: 63-83. Sagie, A. and Koslowsky, M. (1993) Detecting moderators with meta-analysis: an evaluation and comparison of techniques, Personnel Psychology 46: 629-640. *Sambharya, R. B. (1995) The combined effect of international diversification and product diversification strategies on the performance of U.S.-based multinational corporations, Management International Review 35: 197-218. *Severn, A. K. and Laurence, M. M. (1974) Direct investment, research intensity, and profitability, Journal of Financial and Quantitative Analysis 29: 181-190. 44

Shuman, J. C. and Seeger, J. A. (1986) The theory and practice of strategic management in smaller rapid growth companies, American Journal of Small Business 11: 7-18. *Siddharthan, N. S. and Lall, S. (1982) Recent growth of the largest U.S. multinationals, Oxford Bulletin of Economics and Statistics 44: 1-13. Siegel, P. H., Omer, K. and Leavins, J. (1995) Corporate international diversification and performance: a comprehensive analysis of risk and return, Advances in Financial Planning and Forecasting 6: 77-94. Siegel, P. H., Omer, K., Rigsby, J. T. and Theerathorn, P. (1995) International diversification: a review and analysis of the evidence, Managerial Finance 21: 50-77. *Soenen, L. A. (1990) Stock market recognition of multinationality: an empirical study of U.S. markets, Akron Business and Economic Review 21: 64-73. Solnik, B. (1974) An equilibrium model of the international capital market, Journal of Economic Theory 8: 500-525. Sullivan, D. (1994a) Measuring the degree of internationalization of a firm, Journal of International Business Studies 25: 325-342. Sullivan, D. (1994b) The threshold of internationalization: replication, extension, and reinterpretation, Management International Review 34: 165-186. Sullivan, D. (1996) Measuring the degree of internationalization of a firm: A reply, Journal of International Business Studies 27: 179-192. Taveggia, T. (1974) Resolving research controversy through empirical cumulation, Sociological Methods and Research 2: 395-407. Tallman, S. and Li, J. (1996) Effects of international diversity and product diversity on the performance of multinational firms, Academy of Management Journal 39: 179-196. *Tihanyi, L., Ellstrand, A. E., Daily, C. M. and Dalton, D. R. (2000) Composition of the top management team and firm international diversification, Journal of Management 26: 1157-1177. *Tihanyi, L., Johnson, R. A., Hoskisson, R. E. and Hitt, M. A. (2003) Institutional ownership differences and international diversification: the effects of boards of directors and technological opportunity, Academy of Management Journal 46: 195211. Tukey, J. W. (1977) Exploratory data analysis, Reading, MA: Addison-Wesley. Tung, R. L. and Miller, E. L. (1990) Managing in the twenty-first century: the need for global orientation, Management International Review 30: 5-18. UNCTAD (1999) United Nations Conference on Trade and Development, World investment report 1998: Trends and determinants, United Nations: New York and Geneva. UNCTAD (2000) United Nations Conference on Trade and Development, World investment report 1999: Trends and determinants, United Nations: New York and Geneva. Vachani, S. (1991) Distinguishing between related and unrelated international geographic diversification: a comprehensive measure of global diversification, Journal of International Business Studies, second quarter: 307-322. Venkatraman, N. and Ramanujam, V. (1986) Measurement of business performance in strategy research: a comparison of approaches, Academy of Management Review 11: 801-814. *Vermeulen, F. and Barkema, H. G. (2002) Pace, rhythm, and scope: process dependence in building a profitable multinational corporation, Strategic Management Journal 23: 637-653. Wacker, J. (1998) A definition of theory: research guidelines for different theory-building research methods in operations research, Journal of Operations Management 16: 361385. Weber, Y., Shenkar, O. and Raveh, A. (1996) National and corporate cultural fit in mergers and acquisitions: an exploratory study, Management Science 42: 1215-1227. 45

Weick, K. (1995) What theory is not, theorizing is, Administrative Science Quarterly 40: 385-390. Werner, S. (2002) Recent developments in international management research: a review of 20 top management journals, Journal of Management 28: 277-305. Wernerfelt, B. (1984) A resource-based view of the firm, Strategic Management Journal 5: 171-180. Whetten, D. A. (1989) What constitutes a theoretical contribution?, Academy of Management Review 14: 490-495. Whitener, E. M. (1990) Confusion of confidence intervals and credibility intervals in metaanalysis, Journal of Applied Psychology 75: 315-321. Wolf, F. M. (1986) Meta-analysis: quantitative methods for research synthesis, Sage University Papers, Series: Quantitative Applications in the Social Sciences. Yip, G. S. (1989) Global strategy in a world of nations?, Sloan Management Review 31: 2941. Zaheer, S. (1995) Overcoming the liability of foreignness, Academy of Management Journal 38: 341-363. Zaheer, S. and Mosakowski, E. (1997) The dynamics of the liability of foreignness: a global study of survival in financial services, Strategic Management Journal 18: 439-464. *Zahra, S. A., Ireland, D. R. and Hitt, M. A. (2000) International expansion by new venture firms: international diversity, mode of market entry, technological learning, and performance, Academy of Management Journal 43: 925-950. *Zahra, S. A., Sambharya, R. B. and Lee, J. (2000) The effect of international and product diversification on MNCs innovation Paper presented at the annual meeting of the Academy of Management 2000, Toronto. Note: Studies marked with an asterisk have been included in the meta-analysis.

46

Appendix Meta-analytic procedure and statistical formulas

The meta-analysis applied in this study can be divided into three main steps: calculation of confidence interval, calculation of credibility interval, and subgroup test. Formulas used for these statistical analyses are provided by Hunter and Schmidt (1990). First, a confidence interval is calculated for the overall sample population to determine the direction and strength of the internationalization-performance relationship at the aggregate level. This 95% confidence interval around the sample-size weighted mean effect size r corrected for sampling error is estimated with the standard error (SE) and the following formulas:

r -1.96*SE < r < r +1.96*SE ,

r =

(N i * ri ) , Ni

where ri = observed effect size for each sample unweighted by sample size and uncorrected for sampling error. r = sample-size weighted mean effect size uncorrected for sampling error. r = sample-size weighted mean effect size corrected for sampling error. K = number of samples. T = number of studies. Ni = number of observations per sample. It should be noted that the variance of observed effect sizes (r2) is composed of the true variance (p2) and variance stemming from sampling error (e2); thus, r2 = p2 + e2.
2 Accordingly, SDres is the residual variance after the variance stemming from sampling error 2 has been removed from the observed variance (i.e., SDres = p2 = r2-e2). The variance in

observed effect sizes and the variance stemming from sampling error are calculated by the following formulas:

47

2 N i (ri r ) , r2 = Ni

(1 r ) T . =
2 2

Ni

In the second step, the existence or non-existence of sub-populations - i.e., moderators of the internationalization-performance relationship - is determined by calculating the credibility interval around the sample-size weighted mean effect size corrected for sampling error. The range of the 95% credibility interval is based on the corrected standard deviation (SDc) in effect sizes. It is estimated by the following formulas:

r - 1.96 * SDc < r < r + 1.96 * SDc ,

SDc =

1 K (ri r )2 . K 1 i =1

We applied three customary criteria (cf. Hunter et al., 1982) to determine whether our data set is homogeneous (no subgroups exist) or heterogeneous (subgroups or moderators do exist). According to these criteria, data populations are homogeneous when 1) the residual standard deviation is smaller than 25 percent of the populations sample-size weighted mean effect size, 2) at least 75 percent of the observed effect size variance is explained by sampling error, and 3) the null-hypothesis of the chi-square test of homogeneity is significant. The formula for the chi-square test of homogeneity is as follows:

2 =

K 2 * SDc . 2 2 (1 r )

If the three criteria are fulfilled, the third and final meta-analytic step involves subgroup analyses to quantitatively identify individual moderator variables. The Z-test used produces 48

critical values that indicate whether effect sizes between subgroups are significantly different. Z-test significance is determined by a two-tailed test, i.e., when the 5 percent critical value is larger than +1.96 or smaller than -1.96, moderator variables are identified at a significant level. The critical value Z is calculated by the following formula:

Z=

r1 r2

r21 r22 + K1 K 2

where r1,2 = mean effect size for compared subgroups weighted by sample size and uncorrected for sampling error. = variance of the effect sizes for compared subgroups weighted by sample size and uncorrected for sampling error. = number of samples for compared subgroups.

r1, 2
K1, 2

49

Table 1 Studies and coding used in the meta-analysis* Publica- Sample Internationa- Perfortiona sourceb lizationc manced Y Y Y Y Y Y Y Y Y Y Y Y N N Y Y Y Y Y N Y Y N Y Y N Y Y N Y Y Y N N N Y Y N N N N Y Y Y 50 US E US E US US US US US US E US US E US US J E US US E J US J US US US US J,E,US J US J,E,US US US US E US US US US US W US US I F F F I I F F F F S I I F I I S S F S S S S, F, I F S S I S F S S, F, I F I I F, I S S, F F S, F S, I F S S, F S, F M A A A, O A A A A A A A A A A A A, M A A A A A, M A A, M A A O A A A, M, O A A, O A, O A A, M, O A A M A A A, M A, O A, M A, O A

Study 1. Carpenter and Sanders, 2004 2. Capar and Kotabe, 2003 3. Qian and Li, 2003 4. Ruigrok and Wagner, 2003 5. Tihanyi, Johnson, Hoskisson and Hitt, 2003 6. Carpenter, 2002 7. Carpenter and Wade, 2002 8. Ellstrand, Tihanyi and Johnson, 2002 9. Herrmann and Datta, 2002 10. Qian, 2002 11. Vermeulen and Barkema, 2002 12. AlNajjar and Riahi-Belkaoui, 2001 13. Broaden and Samii, 2001 14. Capar, 2001 15. Carpenter and Fredrickson, 2001 16. Carpenter, Sanders and Gregersen, 2001 17. Lu and Beamish, 2001 18. Majocchi and Zucchella, 2001 19. Mauri and Sambharya, 2001 20. Newbert and Contractor, 2001 21. Ramirez-Aleson and Espitia-Escuer, 2001 22. Delios and Beamish, 2000 23. Eden and Thomas, 2000 24. Geringer, Tallman and Olsen, 2000 25. Palich, Carini and Seaman, 2000 26. Phene and Nehrt, 2000 27. Tihanyi, Ellstrand, Daily and Dalton, 2000 28. Zahra, Ireland and Hitt, 2000 29. Zahra, Sambharya and Lee, 2000 30. Delios and Beamish, 1999 31. Gomes and Ramaswamy, 1999 32. Harveston, Kedia and Francis, 1999 33. Nazar, 1999 34. Annavarjula, 1998 35. Athanassiou and Nigh, 1998 36. Barkema and Vermeulen, 1998 37. Mishra and Gobeli, 1998 38. Best, 1997 39. Black, 1997 40. Carpenter, 1997 41. Gomes, 1997 42. Gomez-Mejia and Palich, 1997 43. Hitt, Hoskisson and Kim, 1997 44. Tallman and Li, 1996

45. Ramaswamy, 1995 46. Sambharya, 1995 47. LHeureux, 1994 48. Jung, 1991 49. Morck and Yeung, 1991 50. Soenen, 1990 51. Chang and Thomas, 1989 52. Haar, 1989 53. Grant, Jammine and Thomas, 1988 54. Benvignati, 1987 55. Buehner, 1987 56. Grant, 1987 57. Errunza and Senbet, 1984 58. Kumar, 1984 59. Hirschey, 1982 60. Siddharthan and Lall, 1982 61. Aggarwal, 1979 62. Severn and Laurence, 1974
*

Y Y N Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y

US US W US W US US J,E,US, E US E E US E US US W W

S S, F I F S S, F F F F I F F S, F F F F S, F S

A A A A M M A A A A A, M A M A M A M A

Twenty-seven studies identified in the scanning process were not included in our meta-analysis database because, although authors reported empirical findings, they did not provide statistical summaries including correlation coefficients or other statistics that could have been transformed. Studies that have investigated the internationalization-performance relationship (and therefore are referred to in this paper) but could not be included in our database are as follows: AlNajjar and Riahi-Belkaoui (1999, 2001); Bae and Jain (2003); Berry (2001); Bodnar, Tang and Weintrop (2000); Christophe (1997); Click and Harrison (2000); Contractor et al. (2003); Denis, Denis and Jost (2002); Doukas and Lang (2003); Dragun (2002); Fatemi (1984); Garrod and Rees (1998); Han, Lee and Suk (1998); Kotabe et al. (2002); Martinez (2002); Mathur et al. (2001); Nehrt (1999); Qian (1998); Qian and Li (1998); Ramaswamy, Kroeck and Renforth (1996); Ramirez-Aleson and Espitia-Escuer (2001); Riahi-Belkaoui (1996, 1999, 2002a); Riahi-Belkaoui and Picur (1998); Siegel et al. (1995). Published (Y = Yes; N = No). Home country / region of companies investigated (E = Europe; J = Japan; US = United States). Study samples that were composed of firms from multiple clusters are coded as W = World. Internationalization conceptualization mode applied (coding based on Figure 2). Performance conceptualization mode applied (coding based on Figure 2).

a b

c d

51

Table 2 Study characteristics over time* Conceptualization mode Internationalization Sb Overall 40% 40% 40% F 41% 49% 34% I 18% 10% 26% A 74% 75% 74% Performance M 17% 22% 13% O 8% 3% 13% US 66% 66% 67% Sample source Europe 15% 19% 11% Japan 10% 5% 14% World 9% 10% 8% Degree of internationalization US 35% 32% 38% Europe 50% 37%c 63% Japan 29% 21% 36% World 33% 33% ---d

1988a
> 1988
*

Numbers in the Internationalization, Performance, and Sample source sections represent proportions of variable conceptualization modes applied and sample sources investigated in previous inquiry. Due to number rounding, the percentages do not always add up to 100. Numbers in the Degree of internationalization section represent the average scale of internationalization of investigated firms headquartered in different countries / regions. 1988 = median year investigated. Abbreviations for conceptualization modes are adopted from Figure 2. See Endnote #6 for an explanation of why this number is conservative compared to those for U.S. and Japanese firms. No studies available.

a b c d

52

Table 3 Internationalization and firm performance: meta-analytic results* # of samples 174 115 59 Sample size 35,631 26,625 9,006 r unweighted .07 .06 .09 r weighted .04 .04 .04 95% Confidence interval .02 : .06 .01 : .06 .01 : .07 95% Credibility interval -.18 : .26 -.20 : .27 -.15 : .23 Critical value

Variable Overall Publication as moderator Published studies Unpublished studies Investigative time frame as moderator 1988 > 1988 Firm size as moderator Small firms Large firms Country / region as moderator** US Europe Japan Internationalization conceptualization as moderator Structural Financial Index Performance conceptualization as moderator Accounting Market Operational

SDres .11 .12 .10

SE .01 .01 .02

0.04a

73 101 12 162 115 26 18

12,843 22,788 8,946 26,685 19,327 4,556 6,518

.05 .08 .03 .07 .07 .15 -.01

.05 .03 -.03 .06 .05 .12 -.02

.11 .12 .14 .10 .13 .10 .03

.02 .01 .04 .01 .01 .02 .01

.02 : .08 .00 : .06 -.11 : .06 .04 : .08 .02 : .07 .07 : .17 -.04 : .01

-.16 : .27 -.20 : .26 -.30 : .24 -.13 : .25 -.22 : .31 -.08 : .31 -.07 : .04

1.16b

-1.98c

-2.62d 4.75e -2.94f

70 71 33

21,071 9,757 4,803

.03 .09 .10

.00 .09 .07

.10 .11 .12

.01 .02 .03

-.02 : .03 .06 : .13 .03 : .13

-.20 : .20 -.13 : .32 -.16 : .30

-4.01g 0.60h 2.44i

129 30 15

26,642 7,176 1,813

.05 .07 .21

.03 .03 .22

.11 .09 .14

.01 .02 .04

.01 : .05 -.01 : .07 .14 : .30

-.19 : .24 -.15 : .20 -.05 : .49

-0.03j -4.16k 4.46l

53

Sample size = number of companies; r unweighted = observed mean correlation unweighted by sample size; r weighted = sample-size weighted mean correlation; SDres = corrected standard deviation in effect sizes; SE = standard error in corrected mean correlation for heterogeneous studies; 95% confidence interval = interval around samplesize weighted corrected mean correlation based on standard error; 95% credibility interval = interval around sample-size weighted corrected mean correlation based on corrected standard deviation; Critical value = Z-statistic based on two-tailed test that indicates whether moderator subgroups exhibit significantly different effect sizes (values of -1.96 or lower and +1.96 or greater indicate statistical significance at the 5% level). The sample number (K = 159) and sample size (N = 30,401) for this subgroup analysis is slightly lower than the overall sample population (K = 174; N = 35,631) because the region code W (World) is not included. Comparison between published and unpublished studies. Comparison between investigated year 1988 and > 1988. Comparison between small firms and large firms. Comparison between U.S. and Europe. Comparison between Europe and Japan. Comparison between U.S. and Japan.
g h i j k l

**

a b c d e f

Comparison between structural and financial. Comparison between financial and index. Comparison between index and structural. Comparison between accounting and market. Comparison between market and operational. Comparison between operational and accounting.

54

Table 4 Matrix of population correlations based on paired variable conceptualization modes*

Variable Accounting (A) Market (M) Operational (O) Comparison (IA, IM, IO) vs. others (OS, OF, OI) vs. others (IA, IM, IO, SO, FO) vs. others
*

Structural (S) .00 (53) [15,670] -.01 (13) [4,876] .12 (4) [525] Mean square .013 .153 .082

Financial (F) .05 (52) [7,541] .04 (11) [1,317] .12 (8) [899] F-value .263 3.480 1.863

Index (I) .02 (24) [3,431] .10 (6) [983] .14 (3) [389] P-value .601 .062 .170

This table is split into two sections. The top section displays the mean effect sizes produced by paired variable conceptualization modes (i.e., SA, SM, SO, FA, FM, FO, IA, IM, IO). The number of company samples on which the respective meta-analyses are based on is in parentheses; the total sample size in terms of companies follows in brackets. The bottom section provides a comparison analysis of conceptualization groups. It was conducted by an ANOVA method drawing mean correlation r differences over sample groups that applied differing variable conceptualizations. The first analysis compares a group of samples that applied the index conceptualization of internationalization with samples using all other internationalization conceptualization modes. The second compares a group of samples that applied the operational performance conceptualization with samples using all other performance conceptualization modes. The third compares a group composed of samples that either applied the index internationalization conceptualization or the operational performance conceptualization with all other conceptualization pairs.

55

Table 5 Form of internationalization-performance relationship

Method Linear Quadratic Cubic

R-square .016 .030 .038

df 80 79 78

F-value 1.32 1.21 1.03

P-value .25 .30 .38

b 0* .034 -.085 .106

b1** .164 .801 -1.115

b2

b3

-.783 4.788 -4.775

b0 = intercept. b1, b2, b3 = standardized regression coefficients.

56

Table 6 Utility analysis: estimation of performance impact of internationalization* Performance measure Parameter Mean DOIc (FSTSd) Standard deviation (SD1) of DOI Mean firm performance Standard deviation (SD2) of firm performance Effect size r Amount of absolute increase in firm performance due to absolute increase in DOI of one SD1 (effect size r times SD2) Expected firm performance due to increase DOI of one SD1 Percentage increase in firm performance Estimated impact on firm performance in $ million (U.S.)
*

ROSa .40 .20 .05 .05 .04 .002 .052 4.0% 40

Cost efficiencyb .40 .20 .60 .20 .22 .044 .556 7.3% 876

It should be noted that our meta-analyses revealed that, for example, ROS (i.e., accounting performance) and cost efficiency (i.e., operational performance) would increase by r = .04 and .22 standard deviations, respectively, if firm DOI increased by one standard deviation. Based on previous research findings (e.g., Gomes and Ramaswamy, 1999; Haar, 1989; Herrmann and Datta, 2002; Mauri and Sambharya, 2001; Mishra and Gobeli, 1998; Ruigrok and Wagner, 2003; Tallman and Li, 1996), we assume that the standard deviation of a sample-DOI with a mean 40% FSTS is located close to 0.20 and the standard deviations of ROS (mean = .05) and cost efficiency (mean = .60) can be found close to .05 and .20, respectively. Further, the estimation of the performance impact due to an increase in DOI (FSTS) by one standard deviation (i.e., 20%) is based on the following assumptions concerning firm characteristics: firm is currently located at a DOI (FSTS) of 40%, exhibits $20 billion (U.S.) total net sales, an operating income of $1 billion (U.S.), and operating costs (i.e., sum of material and employee costs) of $12 billion (U.S.). ROS = return on sales. Cost efficiency = sum of material and employee costs divided by total sales. DOI = degree of internationalization. FSTS = foreign sales-to-total sales.

a b c d

57

S Structural

F Financial

I Index*

Internationalization

Firm Performance

Accounting A

Market M

Operational O

Researchers calculated index composites by combining measures from the structural and financial conceptualization modes.

Figure 1 Variable conceptualization modes

58

Correlation (rs) Stem .3 .2 .1 .0 .0 .1 .2 .3 .4 .5 .6 Leaf 1 0012667 0001122345568 011111222222233333334444445555777789 00000111222223333333444445555556666777777788888899999 00000001111334445555677788889999 00222233444568999 023345679 1334 1 0 Statistical summary Maximum: Quartile 3 (Q3): Median (Q2): Quartile 1 (Q1): Minimum: Mean: S.d.*: Proportion positive sign: 0.60 0.16 0.05 -0.03 -0.31 0.07 0.15 63%

S.d. = standard deviation.

Figure 2 Stem-and-leaf plot of correlations (r) between internationalization and firm performance (N=174)

59

0.6 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 Structural Financial Index

0.6 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 Accounting Market Operational

Internationalization

Performance

Figure 3 Boxplots for effect size r by variable conceptualization

60

SCALE a

SCOPE b MODE c

INTERNATIONALIZATION

OPERATIONAL PERFORMANCE g

FINANCIAL PERFORMANCE h

OBJECT d

GOAL e PACE & RHYTHM f

Moderator variables can be exemplified as follows: a low vs. high; b culturally related vs. culturally unrelated; c wholly-owned vs. partially-owned; d sales vs. manufacturing; e profitability vs. market share; f evolutionary vs. revolutionary and stable vs. erratic. g Operational performance can be measured, for example, by market share, cost efficiency, labour productivity, or innovation capability. h Financial performance stands for accounting performance (e.g., profitability ratios) as well as market-based performance (e.g., Tobins q).

Figure 4 Research framework for future inquiry: interrelated moderator variables

61