Management International Review

, Volume 49, Issue 6, pp 733–758

International Strategy Configurations of the World’s Top Family Firms

Another Factor Affecting Performance


    • University of Edinburgh
  • Suzanne Bateman
    • University of Edinburgh
Research Article

DOI: 10.1007/s11575-009-0018-3

Cite this article as:
Carr, C. & Bateman, S. Manag Int Rev (2009) 49: 733. doi:10.1007/s11575-009-0018-3


  • Many variables have been studied, rather inconclusively, to determine behavioural differences of family firms and any impact on performance. This article focuses on just one under-researched variable: their international strategic choices or rather ‘configurations’. We compare 65 of the world’s top family firms with those of a matched sample of non-family firms.

  • Results suggest that family firms’ ‘international ‘configurations’ were just as worldwide and profitable as those pursued by non-family firms. In the case of such large companies, researchers must consider this international dimension alongside other more acknowledged variables, if they are to advance understanding in this important emerging new field.


International strategyFamily firmsWorldwidePerformanceConfigurations

1 Introduction

The importance of family businesses in developed and emerging markets, and even among top global companies, is far greater than previously acknowledged (Birley 2001, Bird et al. 2002, Burkart et al. 2003, Gomez-Mejia et al. 2003, Anderson et al. 2003, p. 264, Kim et al. 2004, Miller/Le Breton-Miller 2005, Villalonga/Amit 2006, Hiscock 2008). Yet we still know surprisingly little about the international strategies of even the largest worldwide family and family-influenced firms.

Internationally many family firms enjoy unique, recognised advantages such as remarkable long-term commitment and employee loyalty (Miller/Le Breton-Miller 2005). How many companies match Wal-Mart for sheer size or Hoshi Ryokan for longevity? Historically, family capitalism has sustained many such successes (O’Hara 2004, James 2006, Landes 2008). Families frequently wield power far beyond equity stakes, controlling companies such as Porsche (Peng 2009, p. 423), which in turn has now acquired VW. Many are also notably agile, niche-focused, high-quality customer service providers (Poza 2004). They include famous brands like Benetton, Heineken and BMW. Yet ‘despite the enormous growth in family business research’ surprising little has appeared in well recognised strategy or international business journals (Chrisman et al. 2008).

This article investigates the largest world family firms’ international strategies, comparatively as against non-family peers, and critically assesses whether this affects performance. Following a review of what we already know from other studies, we compare and contrast the international strategic configurations of 65 of the world’s largest family firms against a peer selected group of non-family firms. Finally we discuss the significance of this additional perspective against more established explanations of family firm behaviour and performance.

1.1 Family vs. Non-family Firms: Competitive Behaviours and Performance

Until recently, ‘most business scholars ignored family businesses’ or down-played them as ‘inefficient anachronisms’, antithetical to success in international markets (Chrisman et al. 2006, p. 720, see also Chandler 1977 and 1990, Dyer/Handler 1994, Colli 2006). Major advances in family business research (Chrisman et al. 2003, Miller/Le Breton Miller 2005, Miller et al. 2007) are not yet mirrored in international business journals or textbooks (Chrisman et al. 2008, pp. 929–931 et seq.), or in commensurate international research. This is partly because of methodological and definitional difficulties (Brockhaus 1994, Hoy/Verser 1994, Wortman 1994, p. 18, Miller et al. 2007).

From a resource-based theoretical perspective (Barney 1991, Grant 1991, Wernerfelt 1984, Collis 1995, Habbershon et al. 1999 and 2003, Barney 2002, Colbert 2004, Eddleston et al. 2008, Sirmon et al. 2008), sustainable competitive advantages would need to reflect idiosyncratic resource-orientated biases of family vis-à-vis non-family firms, enabling them to offset any overarching threat of imitation1. Patient capital, leading in turn to longer term stewardship and sustained innovation, emerges as one such critical advantage (Miller et al. 2008, Sirmon et al. 2008), though this is not always the case (Carney 2005).

Such barriers to imitation derive not only from technical but also social complexity (Barney 1991, Colbert 2004, Eddleston 2008)2. Families approach business relationships differently and may benefit from unique trust-based connections (Fukiyama 1995, Tai/Ghoshal 1998, Zahra 2003, Sjogren 2006, Zahr/Zahr 2006) and knowledge creation processes (Nahapiet/Ghoshal 1998 and 2005, Lee/Macmillan 2008). This may be more advantageous in ‘low trust’ societies such as France, Italy, and China, than in higher trust societies such as the USA, Germany and Japan less marred by undue government interventions (Fukiyama 1995, Zahr/Zahr 2006)3. Institutional factors matter (Peng et al. 2008). Historically family businesses have particularly flourished in less developed, more regulated economies reflecting less perfect market-mechanisms (Jones/Khanna 2006)4.

From a resource-based view (RBV), familiness may thus motivate idiosyncratic strategic decisions and behaviours, in turn affecting performance (Sirmon/Hitt 2003, Carney 2005, Chrisman et al. 2005, Arregle et al. 2007)5. Empirical findings regarding more specific strategies are, though, more mixed: and most studies also depend heavily on establishing statistical associations, raising endogeneity concerns and questions about the direction of causality.

Arguments regarding patient capital are born out by evidence of family firms’ remarkable long-term commitment (Ward 1988, Gudmundson et al. 1999, James 1999 and 2006, Kiley 2004, Chrisman et al. 2006, Le Breton-Miller/Miller 2006, Sjogren 2006, Landes 2008). Inspired leadership (often over generations) has often led to tighter knit cultures and strong positive relationships (Goldwasser 1986, Allio 2004). On the other hand reviewing strategy literature, Harris et al. (1994) highlighted family firms’ inward efficiency orientation (Cohen/Lindberg 1974) and lower emphasis on strategic planning6; resulting in lower growth and participation in global markets (Gallo 1995, Westhead/Howorth 2006). Family firms sometimes do face trade-offs because of constrained resources (Morck et al. 2000, Fernandez/Nieto 2005, Graves/Thomas 2008) or family interests (Trostel/Nichols 1982, Ward 1988, Gordon/Nicholson 2008), and can thus favour lower capital intensity (Friedman/Friedman 1994).

1.2 Empirical Investigations on Performance

An on-going debate on performance has proven somewhat inconclusive. Brockaus (1994), McConaughy (1994), Surowiecki (2000) and Martinez et al. (2007) found evidence of higher profitability ratios in family vs. non-family firms7; yet Morck et al. (2000) and Faccio et al. (2001) found the opposite. Gallo (1995)’s family firms exhibited better profitability ratios but lower market shares and growth rates; whilst Storey (1998) found no difference either way8. Country and scale may also matter. Among just US Fortune 500 companies, family influenced firms achieved marginally higher returns on assets and substantially higher sales growths between 1994 and 2000 (Villalonga/Amit, 2006). Yet Miller et al. (2007) found better performances here were confined to ‘only businesses with a lone founder’ (ibid p. 856).

Miller et al. (2007, p. 829) ‘found that the out-performance of family business was the result of how these businesses were defined’. However, their recommendation to exclude ‘lone founders’ from the definition of ‘true family businesses’ is contentious, so no-one can be dogmatic as to the right definition. Chua et al. (1999)’s comprehensive study of 250 research articles on family business cited 21 different definitions: and in spite of diligent research (Westhead/Cowling 1998, Birley, 2001, Dyer 2003) none is yet widely accepted (Littunen/Hyrsky 2000, Stewart 2003). This article adopts Ward’s (1986) definition that, to qualify as a family business, the family has to own over 50% of the business in a private firm or more than 10% of a public company. This definition may be old and equally contentious, but it meets Astrachan et al. (2002)’s criteria in that it is functional, unambiguous, transparent, and quantifiable.

Perhaps in view of such problems of definition, even studies controlling for levels of family control have proven rather inconclusive (Daily/Dalton, 1992, Daily/Dollinger, 1992, Daily/Thompson 1994). Without distinguishing ‘lone founders’, Anderson and Reeb (2003) did, however, find that the relationship between family control and performance among large US firms was non-monotonic: performance first increased as family ownership increased (up to about one-third of the firm’s outstanding equity) but then decreased.

Beyond Miller et al. (2007)’s recent distinction in respect to ‘lone founders’, the number of contexts and variables potentially impacting on performance is huge (Steier et al. 2004, Miller/Le Breton Miller 2005, Carney 2005, Kellermanns/Eddleston 2006). Performance studies per se are notoriously inconclusive. Villalonga and Amit (2006)’s study of Top US Fortune 500 companies suggests yet further governance and family succession factors; but it is almost impossible to control for all, whilst simultaneously responding to their call for research across more countries. Our article has been conceived more globally, so we now focus on just one less-researched additional variable: family firms’ actual international strategies.

1.3 International Strategies: Another Factor Influencing Performance?

Sirmon et al. (2008) further develop the RBV theoretical perspective, by proposing internationalisation and R&D as two critical mediating factors, offsetting imitability, thereby impacting on performance. Empirical support is afforded by French Small and Medium Sized Enterprises (SMEs) value-added performance data analysis. The choice of SMEs in a low trust country such as France might be viewed as a special case; but for the first time, this establishes a clear theoretical link between distinctive internationalisation strategies and the relative performance of family firms.

Inclusion of R&D as such a prime variable may, though, be harder to justify theoretically. Innovation more generally does play a critical role in resource-based theory; but in many sectors, notably in services, R&D levels are minimal. Comparable entry barriers offsetting imitation are also afforded by branding investments (perhaps better reflected in Sales and General Administration expenditure ratios), and Capital Expenditure levels more generally. Arguably all three variables should be considered, and they may not always prove primary.

Beyond this the number of potential variables is high. Global involvement may also be moderated by strategic factors, family issues and top management attitudes (Gallo/Garcia Pont 1996). The main control variables used by scholars investigating export performance between 1998 and 2005 were: foreign market characteristics, export marketing strategies, firm characteristics, management characteristics (Sousa et al. 2008).

As to whether family firms do or do not internationalise less rapidly, corroborative evidence has been mixed. Simon (1996) identified many thriving family firms, particularly in Germany, with high levels of sales from foreign markets and strong global orientations; but Spanish family firms have been slower to internationalise, focusing more narrowly on customer needs in local markets (Gallo/Garcia Pont 1996, Gallo/Sveen 1991, p. 1839). US family firms also seem to have been slower in responding to foreign competition (Dertouzos 1989) and in undertaking international activities (Zahra 2003). Paradoxically, though, levels of internationalisation related positively to the percentage share of family ownership. A systematic search of literature since Zahra’s (2003) study reveals a dearth of more systematic evidence on family businesses’ distinctive international strategies: even Sousa et al. (2008)’s comprehensive export literature review makes no mention of family businesses.

Meanwhile global strategy per se has been hotly debated. Advocates of more aggressive stances (Hout et al. 1982, Levitt 1983, Ohmae 1985, 1990, 2001, Carr 1993, Bryan et al. 1999, Yip 2003, Friedman 2006, Nolan et al. 2007, Sirkin et al. 2008, Van Agtmael 2008) have been challenged by advocates of regional or semi-global strategies (Baden-Fuller/Stopford 1991, Rugman/Verbeke 2004, Rugman 2005, Mourdoukoutas 2006, Ghemawat 2003, 2005 and 2007, Ghemawat/Ghadir 2000 and 2006, Ghemawat/Hout 2008). The latter arguments challenge more sweeping portrayals of globalisation (Rugman 2001), and highlight a continuing requirement for local adaptation. The right balance depends partly on sector characteristics (Yip 2003, Dicken 2007): much discussion has centred on corresponding internal organisational choices within MNCs (Prahalad/Doz 1987, Bartlett/Ghoshal 1989), and on their likely country and FDI choices. This literature offers no evidence on family firms’ global or international strategies, but accentuates the importance of the issue.

1.4 Requirement for a System Classifying Family Businesses’ Global Orientation

Of the international business contributions discussed, Rugman (2005) provides potentially the most useful, clearly defined classifications: home-region orientated, host-region orientated, bi-regional and global. Some such classification is essential to determine family firms idiosyncratic behaviour vis-à-vis non-family firms. Less helpfully, Rugman’s orientations are essentially arithmetical constructs, unrelated to coherent international strategic or policy choices. His host-regional orientation is also conceptually problematic: such footloose MNCs should arguably be re-classified based on their region of maximum sales. More practically, only about 2% of even the world’s largest companies are classified as global or host-region orientated, 5% are bi-regional, leaving over 90% home-region orientated. Differentiating family and non-family international strategies on this basis, though desirable, is unlikely to lead to statistically reliable results in the absence of extremely large sample sizes.

An alternative approach, more closely related to firms’ strategic choices, is to draw on research on international configurations. Calori et al. (2000) provided the theoretical justification for separating out eight international strategy types, empirically validating such ‘configurations’ through executive interviews in a range of ‘mixed industries’ across Europe10. They identified four types of relatively conservative international challengers: country-centred players; geographic niche players; international opportunistic challengers and continental leaders, and four worldwide players:global luxury niche players; worldwide specialists; quasi-global players and transactional restructurers. Leknes and Carr (2004) integrated an additional global shapers category, evidenced in more internationalised sectors by Bryan et al. (1999). Their empirically supported, nine ‘configurations’ framework is detailed in Appendix A: its pertinence to strategic choices and to performance analysis has also been demonstrated.

This new ‘configurations’ framework allows a more nuanced approach to comparing family firms’ international strategies and suggests two research hypotheses:

Hypothesis 1a: Family firms pursue less worldwide configurations than non-family firms.

Hypothesis 1b: Family firms’ worldwide configurations correlate with the level of the family’s control.

Hypothesis 2: Family firms with worldwide configurations perform less well than comparable non-family firms.

Our underlying theoretical model, adapted from Sirmon et al. (2008), is summarised in Fig. 1. We substitute the broader concept of innovation in place of R&D. We further accentuate patient capital and also social capital effects, reflecting idiosyncratic approaches to relationships as discussed.
Fig. 1

Overview of Conceptual Model (Adapted from Sirmon et al. 2008)

Given the number of other potential explanatory variables affecting performance and the somewhat inconclusive performance debate to date, our primary aim is to address the first hypothesis focusing only on top world companies. We hope though to shed some light on the second hypothesis, by taking some account of scale, country, sector and levels of family control.

2 Methodology

Our sample of the world’s largest family firms was primarily selected from the Fortune Global Top 500 (2003) and Forbes Top 2000 (2003), according to turnover. They were defined and identified in line with Ward (1986), whereby a family has to own over 50% of the business in a private firm or more than 10% of a public company. We adopted Ward’s four group classification: family ownership less than 30%, family ownership between 30% and 50%, family ownership greater than 50%, and lastly where the family had raised capital from the stock market. The Forbes Rich List11 and the Family Business Magazine provided details of families’ involvement and levels of ownership, which were further investigated through company websites and annual reports. After excluding firms such as Mars, too secretive on financial data, we classified a sample of 65 top family firms as shown in Appendix B.

We then paired these 65 family companies with peer non-family firms. Thomson One Banker-Analytics was chosen as our primary data source, since it had the most comprehensive data on global companies in all countries and for its analytical tools, in turn enhanced by specialist Excel-based macros we had recently developed. For each selected family firm, Thomson’s ‘peer’ function identified those most closely matched on market segment/activity match (primarily but not exclusively on Standard Industrial Classification, SIC codes). Secondly it then allowed identification of the closest match, based on turnover to provide for a size match. We also checked that selected ‘peers’ were under no form of family ownership.

We then analysed the international strategic configurations of all 130 sample companies according to Calori et al. (2000) and Leknes and Carr (2004) criteria shown earlier in Appendix A. Data was obtained using secondary information from an extensive range of industry reports, academic articles and websites, in addition to Thomson-One-Banker our prime source of comparable financial data. Such secondary-based analysis proved arduous and set the upper limit to our overall sample number. Statistically though, it was desirable to achieve roughly 30 for each of Calori’s two main groups central to our argument (worldwide players vs. international challengers), and for both family and non-family groups. Classifications of international configurations are shown in Appendix C. We further analysed Rugman (2005)’s four international classifications for 17 family and 16 non-family firms, where our samples overlapped.

Company performances were assessed using comparable ratios from Thomson-One-Banker for all key performance ratios. Our primary dependent performance variable was Return on Capital Employed (RoCE) between 1998 and 2003, and particularly five year averages (as advocated by Porter12). We assessed longer term performances for a slightly smaller sample of 56 from each group, examining 5-year averages for the last 20 years, for RoCE and also sales growth (our secondary dependent performance variable).

Both Villalonga and Amit (2006) and Miller et al. (2007), by contrast, generate vastly more data sets for a similar 6 year period by using more frequent stock-market-based performance data for the largest US 500 companies. International strategic success sometimes, though, can only be judged over decades. Relatively short-term stock market fluctuations overstate the number of genuine data points from the viewpoint of measuring strategic success. Even Miller et al. (2007)’s use of ‘Tobin’s q’ over 3 years is depend upon subjective stock market prices and base year effects. Whilst popular with statisticians, such performance relationships prove highly unstable, when investigated over longer time periods. Stock market increases are just as likely to reflect bad performance at the start of any given period, and a multitude of stock market related variables notoriously difficult to control for. Over our 20 year period investigated, we could have generated many, many combinations and permutations of such short term, base-year dependent rankings – each quite different13.

Addressing this performance time horizon issue, however, radically reduces the number of data sets and explanatory variables that can be reliably controlled for. We have therefore focused sharply on key variables most pertinent to our research hypotheses. Control variables included sector, country of origin, level of ownership (in the case of the family sample), and revenue. We also examined pre-tax profit margins and further benchmarks, including R&D/Sales, Capital Expenditure/Sales and Sales and General Administration Costs/Sales, taking up points raised by Sirmon et al. (2008) and RBV theory. We were unable to go beyond this to investigate the highly extensive set of further potential control variables identified earlier, in the absence of internal access to all our worldwide companies.

It is important to recognise that our study compares the complete population of the 65 largest accessible family-influenced companies worldwide, alongside non-family peers. They are not statistical samples, purporting to represent smaller companies whose behaviours may well prove different. Yet there is a sense in which we too are modelling processes which influence behaviour and performance: from this perspective we can still treat our two groups of matched companies as ‘samples’ of a notionally larger population, provided we are very cautious. Going beyond descriptive statistics to naïve testing, as for example t-tests, neglects synchronous effects even from those multiple factors, identified as critical.

Multilinear regression analysis was therefore used additionally, simply to handle the issue of synchronicity, and to try to distinguish key drivers for high performance within the family and non-family data samples. Family influences or chosen international strategy configurations must here be considered against other key alternative factors responsible for good performance. The purpose of our statistical analysis was to identify correlations between performance; ownership (in the case of the family sample); international strategy, and country of origin and industry segment.

Although statistical modelling is often viewed as the most reliable way to handle multiple variables, it only delivers explanations up to a point (roughly 14% of the story as suggested by our R-squared figures). Further, given several variables ‘sample’ numbers are inevitably low in particular categories (from the viewpoint of statistical rigour), these results must be interpreted with considerable caution. It would therefore be wrong to over-play the significance of results from multilinear regression, or to under-play the relative importance of simple averaged statistics: the latter do, after all, represent the whole population of extremely large family-influenced companies and their peers.

3 Results and Analysis: Family vs. Non-family Internationalisation Configurations

As the prime alternative classification approach, we analysed Rugman (2005, pp. 242–254 et seq.)’s four international classifications for 17 family and 16 non-family firms, where our companies coincided with Rugman’s Global Top 500 companies. Only one non-family firm (BAe) was classified by Rugman as other than home-region orientated (being bi-regional); this compares with four of our family firms – LVMH (global), Sodexho Alliance (host-region), and Michelin and Motorola (both bi-regional). On Rugman’s criteria, our family firms thus emerge as more globally orientated. However, as expected, sample sizes of companies classified as non-home-region orientated are statistically very low: and these classifications do not justice to the spread of firms’ subtler international strategic choices.

We therefore moved on to a more comprehensive investigation of all nine international configurations, suggested by the alternative approach, based on Leknes and Carr (2004)’s Calori/McKinsey based classifications, to address our research hypotheses directly. Figure 2 analyses configurations for both (full) samples of family and non-family firms, for the period 1999–2003.
Fig. 2

Family and Non-family Internationalisation

Non-family firms were more highly represented in country centred and opportunistic international challenger and continental leader categories than family firms. Family firms displayed higher engagement in worldwide categories, particularly as global luxury and worldwide niche players, but also (contrary to our expectations) in global shaper and even transnational restructurer categories, both usually associated with heavy investments. Focusing on the five most globally-orientated categories (i.e. excluding Calori’s four less worldwide international challenger categories), 68% of family firms pursued worldwide configurations vs. 54% of non-family companies.

Focusing on essentially the same ratio – the number of firms pursuing these five worldwide configurations divided by the number pursuing Calori’s four less worldwide international challenger configurations – Fig. 3 next examines these ratios for non-family, and for family firms categorised according to ownership level.14 This depicts a positive trend with regard to level of family ownership at higher levels and international activity; but there is a just slight dip in levels of global participation when family ownership was between 30% and 50%.
Fig. 3

Ratio of Global Players to International Challengers for Different Levels of Ownership

3.1 Impact on Performance

It is surprising that family firms have exhibited such distinctive and worldwide strategies, but has this factor resulted in better performance? Table 1 first provides some longer term performance assessment (over a full 20 years) based on our slightly more restricted samples.
Table 1

5 Year Averaged RoCEs: Family vs. Non-family Over Last 20 Years

5 Yr Periods















5-year RoCE averages for family were higher, as compared with non-family firms, for 1998–2003, 1988–1993 and 1983–1988, but virtually identical between 1993 and 1998. For the most recent five years, we split our samples into three turnover-size categories (< $3.5bn, $3.5–$10bn, > $10bn) and found that family again outperformed non-family companies in all three size categories.

Figure 4 then compares RoCEs, averaged over just the most recent five years, according to all nine international configuration types analysed earlier in Fig. 2.
Fig. 4

Average RoCE ‘03–’99 by International Strategy Configuration

Contrary to our expectations, family businesses exhibited higher RoCEs for all international strategy types, with the exception of global shapers, where non-family firms performed a little better. Family global luxury players and worldwide specialists did very much better, perhaps reflecting their dedication to core activities. Global shapers generally enjoyed relatively high RoCEs, suggesting that dynamic innovative international approaches pay off15.

3.2 Multilinear Regression Analysis of all Variables

For the more recent five years, multilinear regression enabled more systematic performance analysis, taking into account a wider range of key drivers which may have been acting simultaneously. Performance data was investigated on a year-by-year basis for each sample set of 65 companies, using RoCE as the prime dependent variable. The explanatory variables processed by SPSS were: Sales ($m) in Years 2000, 2001, 2002, 2003 (1999 taken as the dummy variable for SPSS); five types of company ownership/control (including one type for non-family, and using totally private firms as the dummy variable); our international strategy types as above (taking country centred players as the dummy variable); area of origin – America, European, Latin America (taking Rest of the World as the dummy variable); Industry – Retailing, Media, Food/Drink and Tobacco, Household and Personal, Auto Manufacturing, Pharmaceutical and Building products (taking Miscellaneous sectors as the final dummy variable). Data was analysed on SPSS first using the Enter option to assess the significance of all variables, and then using Forward Multilinear Regression analysis to establish systematically key drivers only. Results are shown in Table 2 firstly for non-family firms.
Table 2

All Variables Against RoCE 1999–2003 for 65 Non-family Firms

Variables Correlated with RoCE



Standardised Coefficients

t Score






























Adjusted R Squared = 0.150.

This suggests two main drivers of profitability – sector and broad region – pharmaceuticals and retailing being associated with respectively 16.5% and 7% higher RoCEs. The only other statistically significant variable was country of origin: U.S. non-family firms being associated with almost 6% higher RoCEs16. However, once these key drivers are controlled for, there is no residual convincing evidence of any link between international strategy configuration and performance: not even for global shapers which, in isolation, exhibited markedly higher RoCEs. Clearly such sector and country effects should be controlled for.

Regression results for family firms, which effectively control for these and other variables, are shown in Table 3.
Table 3

Multilinear Regression Analysis. All Variables Against RoCE 1999–2003 for 65 Family Firms

Variables Correlated with RoCE



Standardised Coefficients

t Score












Opportunistic International Challenger






Global Shaper


















Worldwide Specialist






Global Luxury Niche






Adjusted R Squared = 0.133.

*denotes significance level at 99.9%

**denotes significance level above 99%

***denotes significance level above 98%

****denotes significance level above 96%

Interestingly, for family firms, the most statistically significant effect was actually the international configuration type pursued. Firms identified as global shapers, worldwide specialists or global luxury players were all associated with significantly higher RoCEs: 9%, 5% and 4.5% respectively. Conversely, international opportunistic challengers were associated with 3.6% lower RoCEs. Just one sector, again retailing, emerged was associated with significantly higher RoCEs at just over 5% higher. Here size as determined by revenue, our other key control variable, was significantly, negatively associated with RoCE. Thus revenue, sector and regions should all be viewed as key control variables, when comparing family and non-family firms. Once we do this, it is family firms’ unique international configuration choices per se, which are most strongly associated with our observed performance differences.

4 Discussion, Conclusion and Suggestions for Future Endeavours

Our ‘sample’, more correctly speaking, constitutes the ‘population’ of 65 of world’s largest family companies. This comparative study against 65 non-family peers was conceived globally, whereas other studies have generally been restricted regionally, often to single countries. Other recent studies, such as Villalonga and Amit (2006) and Miller et al. (2007) provide for larger samples and more numerous control variables (particularly for governance) and so potentially greater statistical robustness. On the other hand, these other studies have not been validated outside the USA, not do they completely control for sector, which emerged alongside regions as critical control variables. Our ‘matching’ principles, based first on industry group and second on size, may though still allow just some regional bias. Whilst US representation is almost the same for both samples, our family sample contained proportionately more European as opposed to Asian companies. This could possibly enhance performance figures for family firms.

Our international strategic classifications are perhaps subtler than other similar studies, but direct field research (though difficult worldwide) would have increased our confidence in judgements based on databases and company websites. Adopting relatively old, though arguably sound, definitions of family ownership levels, could have affected our findings; on the other hand, we could find no more convincing alternative, commanding universal support.

With these reservations, our findings suggest family firms are slightly more internationally orientated than non-family firms17. This finding is in complete contrast to Cohen and Lindberg (1974), Gallo (1995) and Gallo and Sveen (1991) whose samples all extended to smaller family firms18. Zahra (2003)’s US findings likewise indicated that family firms overall lagged behind on international activities, yet paradoxically found a positive relationship with the level of family ownership. Our results suggest that, after just a slight dip in international involvement when ownership is between 30% and 50%, this increased markedly as compared to non-family firms. Our results would lead us to reject our first research Hypothesis 1. For our population of top world family firms, we do not find they are less worldwide in terms of international configurations adopted: and any relationship with increasing levels of control appears broadly positive.

We found no evidence to support research Hypothesis 2. Family firms’ international configurations do emerge as different to those of non-family firms, but from a long-term 20 year perspective, we find this has generally been associated with better RoCE performances. In the absence of all desirable control variables this is not fully conclusive; but in assessing the effectiveness of international strategies this longer term, more global research perspective is of some pertinence and value.

Controlling for just key variables identified, using multiple regression analysis over just the most recent 5 years, we still cannot be completely conclusive on performance relationships. For non-family firms, regional and sector effects emerged as critical performance drivers, which future studies may need to control for. Once we controlled for these counter-factual variables, we found no significant relationship between choice of international configurations and performance.

However this was not the case for top family firms. Not only did they exhibit different international configurations, but these do appear to have influenced performance (even after allowing for these same control variables). Regressing against RoCE for the family firms confirms that 3 of five worldwide international strategies19 significantly enhanced performance, whilst the less worldwide opportunistic international challenger strategy diminished it (as compared with all other configurations). We have expressed some reservations about using multiple regression techniques, particularly given so many potential variables, but there is no better way for controlling for several variables acting simultaneously. Thus it would seem reasonable to conclude that international configurations as a variable appear to be an important factor bearing on family firms’ relatively higher performances. Certainly it would seem worthwhile bringing in as an additional variable to be considered by future researchers.

Possible explanations for our more positive findings for our family influenced firms may lie within some parameters highlighted by studies cited earlier. Long-term commitment, emphasised by Ward (1986), Gudmundson et al. (1999), Miller and Le Breton Miller (2005) and Landes (2008), and was borne out by their surprisingly higher levels of capital expenditure, as compared with sales (in contrast to Friedman and Friedman (1994)’s findings). Appendix F shows capital expenditure, sales and administration and also R&D/sales ratios, which were actually lower for family firms. This suggests Sirmon et al. (2008)’s findings may be more confirmation of ‘patient capital’, than of R&D/sales policies per se (as proposed in their theoretical framework), though our findings concur on the issue of internationalisation. Future research on the impact of R&D or innovation policies more generally would be valuable. Our 20 year longitudinal results concur with recent historical studies, again highlighting patient capital and long term commitment. This seems to have been a factor in the cases examined of Walmart and Wrigley20.

Other explanations lie in the field of governance, though indeed we could not do due justice to all such variables, particularly those identified after our study was initiated. Villalonga and Amit (2006)’s findings, based on the Fortune US Top 500 companies, also recorded just slightly better performance for family-influenced large companies. However their findings suggest that family ownership may create more value where the founder still serves as CEO or as Chairman with a hired CEO. Miller et al. (2007)’s findings, again based on Fortune Top companies, similarly suggests differentiating ‘lone founders’. It may be that other family firms lack the same degree of CEO commitment, allowing other less positive family influences to prevail. Miller and Le Breton-Miller’s (2005) five key winner types21 also look extremely promising in terms of future research themes and new classifications for investigation.

Finally, in spite of such logical explanations and theoretically constructed arguments earlier, we cannot entirely dispense with concerns as to whether the direction of causality runs the other way: i.e. that it is the better longer term performance of family firms which has facilitated their more worldwide international strategies. New statistical approaches could assist future researchers in further addressing such endogeneity concerns (Bascle 2008), though well-founded theoretical arguments (Hult et al. 2008) or longitudinal/historical evidence remain the best safeguards here (Morck/Yeung 2007).

Clearly, there is a need for more research on the distinctive international strategy choices of family businesses, whilst still controlling for degrees of family control, business sectors, at least broad global regions with proportionate coverage in North America, Asia and Europe, and also size. Researchers also need to sharply distinguish this group of Large Family (Influenced) Enterprises (or LFEs); SME family businesses may be less internationally progressive.

Aside from such promising arenas for future research, we end by summarising our key conclusions drawn from this study. We found no evidence of family firms adopting more ‘inward’ orientated strategic choices. In contradiction to Hypothesis 1, our findings indicate family firms adopt a higher proportion of the most worldwide configurations, as compared to non-family firms. This propensity actually rises with the degree of family ownership and control. Finally, in contradiction to Hypothesis 2 and even allowing for control factors, family firms pursing these more worldwide configurations appeared to achieve higher, not lower, levels of profitability.

Performance studies, per se, are rarely totally conclusive but we suggest that family firm’s internationally configurations represent an important factor for future research, and an important additional control variable that should not be overlooked.


Strictly speaking the RBV focuses on four aspects of resources and capabilities: their value (V), rarity (R), imitability (I), and organizational (O), known as the VRIO framework (Barney 2002 pp. 157–174); but literature applying RBV to family businesses has tended to see facing the competitive threat from imitability as a more all-embracing challenge. Clearly the other three elements also potentially contribute to addressing this challenge.


Indeed social capital theory (Coleman 1988, Baker 1990, Burt 1992 and 1997, Uzzi 1996, Portes 1998, Steier 2001, Adler/Kwon 2002, Sharma 2004, Agndal et al. 2008) is itself almost a testament to advantages from being part of some notional ‘extended family’ (Sanders/Nee 1996, Steier 2001, Sharma 2004).


Fukiyama’s position here is ultimately not dissimilar to Chandler (1977, 1990). Interestingly the recent Credit Crunch may now profoundly undermine trust relationships in even in countries previously viewed as ‘high trust’. Fukiyama would predict this as likely to lead to far greater government intervention, reducing these countries’ competitiveness. This in turn would reduce some of the advantages he sees for corporate vs. family firms.


For further supporting evidence see also Khanna, Palepu and Sinha (2005) and Khanna (2007) highlighting the effects of ‘institutional voids’. Lin and Zhang (2005) and Zain and Ng (2006) highlight SME network effects exploited by family firms respectively in Taiwan and Malaysia. Zhao and Hsu (2007) show the effect on their foreign market entry strategies.


We should note, as highlighted by Miller et al. 2007, that ‘familiness’ is not necessarily the same thing as the level of family ownership and control, though in practice Sirmon et al. (2008) utilise the latter for their empirical measures.


Arthur Andersen/MassMutual (1997)’s survey found that 69.4% of American family businesses reported not having written a strategic plan. Ward (1997) argued that the rejection of planning in family firms was due to a culture restricting information flow which, in turn, was likely to restrict growth. Harris et al. (1994) found little difference in family firms’ strategic management processes with regard to internationalising, as compared with non-family firms. However, family firms which do use formal strategic planning appear to benefit developmentally (Mazzola/Marchisio/Astrachan 2008) and to internationalise more actively (Rienda et al. 2005).


Ding et al. (2008) also found that Chinese family firms performed significantly better than state-owned companies on five financial ratios between 1999 and 2004.


Reported in The Economist (U.S), Dec 2, 2000, p. 7.


Though export/sales ratios proved similar comparing just medium and large sized companies.


‘Mixed industries’ are defined as industries which, whilst not exceptionally globalised, nevertheless display considerable internationalisation in terms of firm strategies.


The Rich List is a good place to start the process of identifying family companies. Many of the world’s richest people are founders or descendants of family businesses. For example, at the top of the Forbes list are all the members of the Walton family who are heirs to the retail giant, Wal-Mart.


Presentation given by Michael E Porter, ‘Competitive Strategy for Profitable growth.’ Given Thursday 10th Feb 2005, Edinburgh International Conference Centre.


Miller et al. (2007) report that there findings were ‘indeed highly sensitive both to the way that family businesses were defined and to the nature of the sample’, thus advancing the case for their advocated new variables. However, such sensitivity might equally merely reflect the instability inherent in their statistical model and chosen performance parameters. Robustness tests might be required over successive time periods to establish this.


Appendix D contains pie charts for each category of ownership and shows the percentage of companies found in each Calori strategy type. The main difference is that fully private family companies seemed to have moved beyond opportunistic international challenger and continental leader configurations and to have been most worldwide, as quasi-global transnational restructurers and global shaper players.


Appendix E also compares averaged 5 year sales growth figures for all types over the same period. Family firm types displaying relatively faster sales growth, as compared with non-family firms, were continental leaders and worldwide specialists and, most especially, opportunistic international challengers. There was no consistent trend of internationalisation in general benefiting sales growth. Non-family firms were growing relatively faster in country centred and transnational restructurer types.


This is re-assuring. Our two samples contained the same proportions of US firms. Had this not been the case there would otherwise have been a danger of our results having been skewed by this country performance effect. No statistically significant effect was noted for family firms. There is some residual danger that some differences in the proportions of European vs. Asian companies could slightly skew our results, but these statistical results suggest there should be no systematic skewing.


Using Rugman (2005)’s four alternative categorisations of global, bi-regional, host-regional and home regional, family firms emerged as even more global, but as expected this alternative classification approach yielded less statistically reliable results. The majority of both family and non-family firms would all be classified by Rugman as simply home-regional, affording little discrimination in terms of different types of international strategies.


Cohen and Lindberg (1974)’s sample comprised in fact entirely smaller firms.


These being global shapers, worldwide specialists and global luxury niche players. Other configurations all being on par with each other: country centred, geographical niche, continental leaders, and (in terms of more worldwide players) quasi-global and transnational restructurers.


Wrigley ranked top on RoCE averages over the last 10 years (at 31%) in our sample of 56 family firms, analysed by CIMacro, and Walmart 9th. Anderson and Reeb (2003) likewise noted Wrigley’s superior long-term RoCE figures, as compared with its key rival Hershey Foods and US averages.


These winning family influenced firm types are: Brand Builders (e.g. Estée Lauder, Levi Strauss, Hallmark); Craftsmen (e.g. Coors, Timken, Nordstrom); Operators (e.g., Cargill, Ikea, Walmart); Innovators (e.g. Corning, Michelin, Motorola); Deal Makers (e.g., Bechtel, Bombardier, J.P. Morgan).



The authors would like to acknowledge very helpful feedback on the first draft of this article from Peter Rosa, Professor of Family Business at the University of Edinburgh. We would also like to acknowledge the contribution of an anonymous MIR reviewer, which has had a very considerable impact in furthering the intellectual development of this article since its original inception.

Copyright information

© Gabler-Verlag 2009