Strategic Management Journal Strat. Mgmt. J., 30: 207–219 (2009) Published online 10 November 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.732 Received 22 May 2006; Final revision received 6 August 2008

FOREIGN OWNERSHIP AND LONG-TERM SURVIVAL DORTE KRONBORG1 and STEEN THOMSEN2 * 1

Center for Statistics, Copenhagen Business School, Copenhagen, Denmark Center for Corporate Governance, Department of International Economics and Management, Copenhagen Business School, Copenhagen, Denmark 2

We study the relative survival of foreign- and domestically owned companies in Denmark over more than a century (1895–2005). Contrary to previous studies that have emphasized the liability of foreignness, we find evidence of a significant survival premium for foreign-owned companies; however, the premium declines over time and disappears entirely in the last decade leading up to 2005. Further evidence indicates that the foreign survival premium is negatively influenced by new foreign entry, and that the long-run decline is caused by increasing competition between foreign subsidiaries. Copyright  2008 John Wiley & Sons, Ltd.

INTRODUCTION How does foreign ownership affect the survival of companies? Does it improve company survival by resource transfers from the parent company; or does it reduce survival because of the liability of foreignness and internal bureaucracy? While there have been several studies of the survival of foreign subsidiaries (e.g., Mitchell, Shaver, and Yeung, 1994; Shaver, 1995, 1998; Li, 1995; Delios and Beamish, 2001, 2004; Dhanaraj and Beamish, 2004; Chung and Beamish, 2005), only a few have compared the survival of foreign-owned and domestic companies (Li and Guisinger, 1991; Pennings, Barkema, and Douma, 1994; Zaheer and Mosakowski, 1997; Thomsen, 2000; Mata and Portugal, 2002). Moreover—somewhat surprisingly given the apparent competitiveness of multinational companies in the global economy—some Keywords: ownership; survival; foreignness; subsidiaries

∗ Correspondence to: Steen Thomsen, Center for Corporate Governance, Department of International Economics and Management, Copenhagen Business School, Porcelænshaven 23 A Building 65, 2000 Copenhagen F, Denmark. E-mail: [email protected]

Copyright  2008 John Wiley & Sons, Ltd.

of these studies (e.g., Zaheer and Mosakowski, 1997) have emphasized the liability of foreignness (Zaheer, 1995), while others have found no significant survival advantage compared to domestically owned companies (Pennings et al. 1994; Mata and Portugal, 2002). In this study we examine the long-run survival of foreign subsidiaries in Denmark compared to a matched sample of domestically owned firms. Our data is unique in that it extends over a 110-year period (1895–2005), which enables us to examine how foreign versus domestic ownership influences survival over long periods of time. This is important, because there are strong theoretical reasons to assume that relative survival rates will change over time due to a declining liability of foreignness, new entry by foreign or domestic companies, and new policy regimes. We find a statistically and economically significant survival premium for foreign-owned companies: domestically owned companies have a two times higher exit risk. But the foreign survival premium has declined over time and disappears completely by the end of the 110-year study

208

D. Kronborg and S. Thomsen

period. Additional evidence indicates that competition between foreign entrants has eroded the comparative advantages of foreign-owned companies. The time span covered is unique and allows an examination of the long-run dynamics of foreign subsidiaries. Nevertheless, there are both costs and benefits to a very long-term historical study. In particular, because the number of foreign subsidiaries increased rapidly over the past century, our sample is skewed so that we have many more observations at the end of the observation period than we have at the beginning. We correct for this by grouping observations in time periods, but there remains a danger of near-term bias (i.e., overemphasis on recent observations), which needs to be taken into consideration when interpreting the results.

THEORY, EVIDENCE, AND HYPOTHESES The standard theory of the multinational enterprise (Caves, 1996; Buckley and Casson, 1976; Dunning, 1981; Hennart, 1991) proposes that foreign direct investment will occur and be successful when ownership, location, and internalization advantages exceed the costs of establishing and running foreign subsidiaries (Dunning, 1981). Similar arguments figure prominently in the international business literature as a rationale for the establishment of sales subsidiaries rather than independent distributors (Anderson and Gattignon, 1986). The theory predicts that foreign subsidiaries have performance advantages when company-specific resources matter while transaction costs are high, and several empirical studies have supported this hypothesis (Caves, 1996). The implication is that foreign subsidiaries will have higher survival rates if their competitive advantage relative to domestic companies exceeds their added costs of operating in a foreign country. To the ownership and location advantages we must add the possible options-like character of foreign direct investment (Kogut and Kulatilaka, 1994; Mata and Portugal, 2002). Multinational companies may decide to maintain a presence in a country despite adverse conditions, because foreign subsidiaries can provide them with valuable options for future expansion. This may improve their survival power relative to domestic companies. Copyright  2008 John Wiley & Sons, Ltd.

However, ownership, internalization, and option advantages must be balanced against the disadvantages of foreign ownership. Hymer (1976) suggested that multinational companies that establish subsidiaries in a foreign country incur costs because they lack knowledge of business conditions in the host country and because of policy discrimination. These costs can be thought of as the ‘liability of foreignness’ (Zaheer, 1995). The liability of foreignness should be distinguished from the liability of newness (Stinchcombe, 1965), which newly established foreign and domestic companies both face. If foreign parent companies act under a veil of ignorance compared to their host country competitors, they are likely to make more mistakes that could result in a higher exit rate. However, it is not evident that foreignness is necessarily a liability in all cases. Mata and Portugal (2002) argue that new foreign subsidiaries have better survival chances than new domestic ventures, because they can draw on the expertise, experience, and legitimacy of the parent company. Brannen (2004) argues that foreignness was an advantage for Disneyland Tokyo and suggests that foreign firms may to some extent recontextualize their operations to achieve a better semantic fit with the host country environment. While the theoretical relationship between foreign ownership and survival is ambiguous, research on multinational companies tends to focus on survival advantages, given the massive increase in the volume and stock of foreign direct investment over the past decades (Caves, 1996). Despite an improved understanding of the costs of foreignness, we therefore propose what we believe to be the classical a priori hypothesis for empirical testing. Hypothesis 1: The foreign survival premium: foreign subsidiaries are likely to have higher survival rates (lower exit risk) than domestically owned companies. Dynamics and sustainability of the foreign survival premium While the standard theory is expressed in a comparative static framework, we propose that the costs and benefits of foreign ownership will change over time. In particular, long-run effects may differ from short-run effects. Based on previous research, Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Foreign Ownership and Long-term Survival we distinguish between changes at the firm, industry, and policy level. At the firm level the relative survival of foreign subsidiaries is likely to increase over time (with firm age) because the initial liability of foreignness is gradually overcome as a foreign subsidiary learns more about the host country environment and develops better connections to local business networks (Johanson and Vahlne, 1977; Barkema, Bell, and Pennings, 1996; Zaheer and Mosakowski 1997; Lu and Beamish, 2004). However, in the longer term, the positive age effect on foreign survival is likely to level out (Zaheer and Mosakowski, 1997). At the industry level, successful entry is likely to be imitated, which will increase competitive pressures (Lieberman and Montgomery, 1988, 1998). Over longer calendar time periods, competition from late entrants, technological shocks, the industry life cycle, and other factors are likely to erode the advantages of early entrants (Schumpeter, 1950). For example, first mover advantages may be partially eroded by imitators (Lieberman and Asaba, 2006: 373). Host country employees may quit and start competing businesses. Uncertainty and ambiguity may lead other foreign companies to enter (c.f. the bandwagon effect observed by Knickerbocker, 1973). Competitors may feel forced to enter in order to counter competition in the local market (Lieberman and Asaba, 2006; Leahy and Pavelin, 2003). Miller and Eden (2006) argue that the negative effect of local density on foreign subsidiary performance has been overlooked in previous studies. Admittedly it is possible that there are positive synergies (agglomeration effects) between foreign subsidiaries up to a point, particularly if they are not strong competitors (Mitchell et al., 1994; Delios and Makino 2003). But beyond this point competitive pressures will almost surely reduce survival chances as the number of foreign competitors increases (Hannan and Freeman, 1977; Mitchell et al., 1994; Chang and Park, 2005). We propose that an increasing number of foreign subsidiaries will lower the relative survival rates of foreign more than domestic companies, because multinational subsidiaries operate in niches that differ from those of domestically owned companies (Caves, 1996; Dunning, 1981). Apart from industry and size effects, this is reflected in physical location (Chang and Park, 2005), strategy group association (Porter, 1979), advertising and Copyright  2008 John Wiley & Sons, Ltd.

209

research and development (R&D) intensity (Caves, 1996), or workforce education (Mata and Portugal, 2002). Moreover, the competitive pressure exercised by domestic firms is likely to intensify in the long run as surviving domestic firms increasingly become multinationals themselves and benefit from the classic ownership, location, and option advantages of multinational enterprise. Finally, policy regimes influence the cost and benefits of foreignness (Hymer, 1976; Jones, 2005). In the short run, currency restrictions, import tariffs, or outright expropriation may force foreign companies to withdraw (and so reduce their relative survival rates) while deregulation would have the opposite effect. However, in the long run, deregulation may also lower entry barriers and attract more competition from other foreign firms. Deregulation can also make it easier to service a market by exporting, which can lead to plant closures in the host country. Altogether, controlling for firm age and policy regime, we propose that increasing density (Miller and Eden, 2006) will drive down the long-run survival rates of foreign subsidiaries. This leads to: Hypothesis 2: Given company age, we expect the foreign survival premium (exit hazard) to decline over time as a consequence of entry by other foreign subsidiaries. Previous empirical studies In the first seminal study on a small sample (81 foreign business failures) without control variables, Li and Guisinger (1991) found that domestic manufacturing firms failed almost four times as frequently as the foreign companies. Pennings et al. (1994) reported that 462 ‘expansions’ of large Dutch multinationals failed more often if they were international than if they were domestic. However, the effect of foreignness became insignificant when controlling for firm and calender time effects. Zaheer and Mosakowski (1997) examined the survival rates of 2,667 foreign exchange trading rooms over the period 1974–1993. Foreign trading rooms were found to have higher exit rates, particularly in the first 10–15 years after market entry. Mata and Portugal (2002) compared the survival of newly established firms in Portugal over the period 1983–1989. They found no significant survival differential when controlling for other factors. Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

210

D. Kronborg and S. Thomsen

Related literature has indicated that the liability of foreignness may be attributable to both institutional and cultural distance between host and home country (Kostova, 1998; Kostova and Srilata, 1999; Kogut and Singh, 1988). Miller and Parkhe (2002) measured it by x-effciency (failure to reach the production possibility frontier) and found higher x-ineffciency in foreign-owned banks. Mezias (2002) found that foreign subsidiaries in the United States face more labor lawsuits than domestic firms. In contrast, Nachum (2003) found higher rates of return among foreign financial firms in the city of London. Brannen (2004) found foreignness to be a liability for Disneyland Paris, but an advantage for Disneyland Tokyo. In a study of first mover advantages controlling for entry size, Mascarenhas (1997) found that foreign-owned companies in off-shore oil drilling had higher exit rates.

METHODOLOGY AND SAMPLING Our study compares the survival of foreign subsidiaries in Denmark to a control group of domestically owned companies (for convenience we occasionally refer to these companies as ‘foreign’ and ‘domestic,’ respectively). We were able to track companies over the period 1895–2005 using a unique data source, Green’s Handbook of Danish Funds and Stocks (Greens-Børsen, various years), which has published—since 1887—a fairly comprehensive register of companies operating in Denmark. We use a matched sample methodology. The objective is to approximate a controlled experiment by matching a treatment group (foreignowned companies) to a control group that is as similar as possible to the study group in relevant dimensions (Stouffer, 1950; Breslow, 1996; Mezias 2002). In this study we match the foreignowned companies with domestic twins in the same industry and size class. We use the dataset on foreign subsidiaries gathered by Thomsen (2000), updated with a follow-up study in 2005. We were also able to add information on the causes of exit in the last part of the period. To illustrate the methodology, one of our observations is Ford Motor Company, which established a European subsidiary in Denmark in 1919 (Ford’s second in Europe). This subsidiary, which supplied Copyright  2008 John Wiley & Sons, Ltd.

the Nordic and Baltic markets, Poland, and Germany, was from the very beginning one of the largest manufacturing companies in Denmark. It produced Ford T, Ford A and later V8 engines (i.e., Anglia) and tractors. Production was closed down in 1965, but the company continues as a sales and service operation. We match Ford with The United Danish Automobile Manufacturers (De Forenede Danske Automobilfabrikker), which resulted from a 1918 merger of three Danish producers of cars, trucks, busses, and, later, motor trains, until production was closed in 1957. Thereafter the company began to import foreign brands (e.g., British Triumph). The company was closed in 1993. We register this case as evidence that the foreign company (Ford) survived and survived longer—so far (as of 2008) for 89 years compared to 75 years for the domestic control company. However, at the time of entry Ford (share capital 30 million DKK) was much larger than The United Danish Automobile Manufacturers (share capital 2 million DKK), which indicates the need for controlled statistical studies. The sampling design is as follows: all foreign subsidiaries of manufacturing companies registered in the years 1895, 1907, 1915, 1925, 1934, 1946, 1956, 1964, 1975, 1985, and 1995 were sampled with a follow-up in 2005. Our data collection was made as bias free as possible by 1) identifying all foreign-owned companies in the company register, 2) choosing as control the company in the same (four-digit) industry category that came closest to the foreign company by assets (primary criteria) and share capital (secondary criteria). Thus we have tried to avoid subconsciously collecting data to confirm prior hypotheses. We identified 533 foreign-owned and 528 matching domestic-owned companies. Our database therefore consists of 528 pairs of companies. Data limitations We sample all foreign subsidiaries of manufacturing companies once every decade, which means that some companies may enter and exit after a few years without being included, but we see no bias arising from this source. Banks, shipping and service subsidiaries are omitted. The register includes companies over a certain minimum size. The minimum size varies over time, for example at the end of the period only companies with either more than 45 employees, assets >45 million DKK, or gross Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Foreign Ownership and Long-term Survival profits >35 million DKK are included. This means that new entrants to the dataset may occasionally be more than 10 years old because they were added after they had grown large enough. We sample only subsidiaries in which a single foreign corporation holds more than 50 percent of the stock. Minority owned foreign subsidiaries (Delios and Beamish, 2004) and companies owned by foreign individuals are not included. Finally, it is important to note that information early on is scarce. Thus, we may be able to observe that a certain company is active at some point in time, but has disappeared in the next period without knowing precisely what happened to it. Naturally, we cross-checked our data source for possible name changes, omissions, information about mergers, etc., but in some cases we were unable to verify what happened. Our ‘exit’ event therefore is not synonymous with ‘liquidation’ or death, but includes exit by merger into a larger entity (if the company disappears as an independent unit). Moreover, although we did try to account for this, a company may in a few cases survive as a corporate entity (owning property or a bank account) while actual business operations have ceased. After entering the database, the companies were tracked in each of the above sampling years with a follow-up in 2005. We recorded the evolution of the company in terms of share capital, assets, industry, and when relevant, the year of exit. The company register also provided information on establishment year and, thereby, company age. We do a one-country study of Denmark, a small European country (EU member) of five million people (comparable to Minnesota in the United States) and a per capita gross national income of 40,650 U.S. dollars compared to U.S. average 41,400 U.S. dollars (World Bank, 2005). Despite a large welfare state, government intervention in the private sector is very limited and the economy is open to international trade and investment (exports constitute some 45% of gross domestic product). Historically, following the golden age of free trade up to 1930, currency, investment, and trade restrictions became severe during the Depression and the war years 1930–1945 (Boje, 2000). After the Second World War some subsidiaries (i.e., German and Japanese) were nationalized or dissolved (Boje, 2000). But restrictions on trade, investment, and capital transfers were gradually removed after 1950 and completely abolished in Copyright  2008 John Wiley & Sons, Ltd.

211

1982. Altogether, this corresponds well to the historic periods identified by Jones (2005). Table 1 lists control variables. Apart from ownership a, number of other variables are expected to influence company survival. We focus on size, industry, and time period effects. Variables like knowledge stock, R&D intensity, brand capital, and so on could also influence survival rates, but cannot be measured empirically over a long period of time. Instead we rely on the twin study (matched sample) methodology to control for unobservable heterogeneity.

DESCRIPTIVE STATISTICS A first look at the data, without taking the matching or the external variables into account, indicates a marginally higher exit risk for domestic companies. At the end of the follow-up period 301 (57%) domestic companies and 339 (64%) foreign subsidiaries were still in operation while the remaining 227 (43%) domestic and 189 (36%) foreign subsidiaries had ceased their activities before 2005. In other words, at first glance, the exit rates seem to be lower for the foreign subsidiaries, and foreign firms seem to have higher survival power. Table 2 shows that the differences in exit causes between the two groups, foreign and domestic, seem small. For example, the fraction of foreign and domestic firms exiting by mergers and acquisitions is almost identical. However, the bankruptcy frequency—though as low as a few percent for both kinds of firms—is 2.7 times higher for domestically owned firms (5.1%) than for foreign subsidiaries (1.9%). Apparently, foreign parent companies prefer to avoid bankruptcies and have the financial strength to carry on. The matching strategy succeeded with respect to industries. Food, chemicals, and metal products are the main industries in which foreign subsidiaries are active. Together they make up 82 percent of the included firms. This agrees with previous studies (Caves, 1996; Chandler, 1990), which find that multinationals are particularly active in industries characterized by (multi-firm) economies of scale and scope. Our preferred size measure is (inflation adjusted) share capital. This is both because it is always available and because of endogeneity problems with measures that reflect the relative success of the company. Companies that are not competitive Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Miller and Eden (2006)

Jones (2005: .20 ff)

Agarwal and Gort (1999), Jovanovic and MacDonald (1994), Klepper (1996), Dunning (1981), Caves (1998), Thomsen and Pedersen (1998), Chandler (1990)

Copyright  2008 John Wiley & Sons, Ltd.

Number of foreign subsidiaries

Time periods (phases)

Survival increases with company size because of selection effects, larger investment commitment, higher sunk costs, lower sensitivity to economic shocks Survival varies by industry because of differences in life cycle, intensity of competition, predictability of demand, and scale economics and technology change. Multinationals cluster in scale and scope industries Differences in policy regime 1. The first global economy 1880–1929 2. Disintegration, protectionism and war 1930–1950 3. Liberalization 1950–1969 4. The new global economy 1969– Survival of foreign subsidiaries decreases with density Company size

Industry effects

Rationale Control variable

Table 1. Control variables

Jovanovic (1982), Caves (1998), Agarwal and Gort (1999), Mata, Portugal, and Guimaraes (1995) Caves (1996)

D. Kronborg and S. Thomsen

References

212

are more likely to experience cutbacks and size reductions and then to exit, but it would be wrong to interpret this as a negative effect of company size on exit rates. In contrast, the share capital reflects the (in cash or in kind) capital contribution that is ex ante invested in the company before its relative fitness is tested in practice. It was not possible to find exact matches according to company size, but the overall picture is that company size (share capital) is comparable for matched companies. Due to the sampling strategy, we could not expect that firm ages at entry time were similar within pairs. However, the age difference was relatively small with a slight tendency for domestic companies to be older than their foreign matches. More importantly, we find no systematic trend in age differences during the sampling period. We observe a rapid increase in the number of foreign companies operating in Denmark during the sampling period, particularly after the late 1970s.

STATISTICAL METHODS The statistical analyses are based on the hazards model for analysis of censored survival data as proposed by Cox (1972, 1975). But unlike the ‘standard’ hazards model, companies do not enter the study at the same age, and to avoid problems with length biased sampling, each company only contributes to the partial likelihood from the age of entering into the study—that is, the observations are considered as left truncated (Kalbfleisch and Prentice, 2002). To be precise, let Tij , i = 1, 2 (domestic, foreign), j = 1, . . . , N (indexing pairs) be independent survival times with hazard rate functions λij , λij (t) = limt→0+ P (t ≤ Tij < t + t|Tij ≥ t)/ t, right-censoring times Uij and left-truncation times Vij , such that P (Vij < Uij ) = 1. As in Andersen et al. (1993) we use the notation T˘ij = Tij ∧ Uij > Vij and Dij = I(T˘ ij = Tij ). Our lefttruncated and right-censored survival sample is then given as the i.i.d. triplets (Vij , T˘ij , Dij ), i = 1, 2, j = 1, . . . N. We want to test hypotheses about the relation between λ1j (t) and λ2j (t), j = 1.., N. Under the Cox model the hazard function for the j’th pair may be written 

λij (t) = λ0j (t)eβ Zij , i = 1, 2, j = 1, . . . , N, (1) Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Foreign Ownership and Long-term Survival Table 2.

213

Exit causes for domestically owned companies and foreign subsidiaries

Cause Merger, etc Liquidated Bankrupted Empty Unknown Censored Total

Domestic 89 19 27 11 81 301 528

Foreign

(16.9%) (3.6%) (5.1%) (2.1%) (15.3%) (57.0%) (100.0%)

84 24 10 5 66 339 528

where λ0j (t) s are the unknown baseline hazard for the j’th pair of companies at time t. Zij is the vector of covariates and β the vector of unknown regression parameters. Alternatively, the model for the relation between the two intensities can be formulated as 

λ1j (t) = λ2j (t)eβ (Z1j −Z2j ) , j = 1, . . . , N. Estimation is carried out by letting each matched pair of triplets define a stratum, and only pairs where T˘1j ∧ T˘2j do not correspond to a censoring contribute information to the test statistic. This means that the analysis relies only on intrapair comparisons, and the interpretation of regression parameters is within pairs. Results will be given in terms of hazard ratios (relative exit risks) for domestic compared to foreign companies, λ1j (t)/λ2j (t). A relative risk larger than one means that the risk for exit is larger for domestic companies than for foreign subsidiaries. The 95 percent confidence limits are Wald based. In this model it is not possible to estimate main effects of covariates with constant values within matched pairs of companies (nested effects). However, it is possible to calculate test statistics for comparison of relative risks for groups of pairs by including interaction terms and thereby to evaluate the influence of, for example, industry. For this purpose we have used likelihood-ratio test statistics, −2 log LR, which are evaluated in the approximating χ 2 distributions. When accounting for competitive advantages the model can be expressed as λ1j (t) = λ2j (t)eβ0 +β1 Zj j = 1, . . . , N,

(2)

where Zj describes the number of foreign companies operating at time of entry for the j’th foreign company. The model is extended to account for Copyright  2008 John Wiley & Sons, Ltd.

(15.9%) (4.5%) (1.9%) (0.9%) (12.5%) (64.2%) (100.0%)

Total 173 43 37 16 147 640 1056

(16.4%) (4.1%) (3.5%) (1.5%) (13.9%) (60.6%) (100.0%)

other factors like industry by allowing β0 to vary over industries. Hereby, eβ0 for a given industry relative to another is the multiplicative factor for increase/decrease in the relative risk for otherwise comparable companies, that is, companies with the same competitive conditions. As part of the analysis we analyze the evolution of relative exit risk over calendar time. In a given calendar period [S1 ,S2 ], the sample thus consists of triplets (Vij , T˘ij , Dij ), where the left-truncation times and the right-censoring times are such that pairs at risk in [S1 ,S2 ] are considered left truncated at their age at calendar time S1 if the pair is included before S1 , and right censored if the exit occurs after calendar time S2 . Model (1) is applied to each of these strata, and comparison of excess risk for domestic versus foreign companies is performed as a likelihood ratio test. This, in fact, corresponds to analysis of an ‘interaction’ between excess risk and calendar time. As mentioned above, only pairs of companies where one of the companies exits before either exit or censoring of the other contributes to the test statistics, and, consequently, the effective sample size can occasionally be small. Accordingly, the precision of the estimated coefficients as well as the power of hypothesis tests may be weakened. Therefore, analysis of models with simultaneous inclusion of several factors is limited to a few relevant cases.

RESULTS For illustrative purposes, we plot the standard marginal Kaplan-Meier survival curves for both categories in Figure 1, which portrays the estimated survival probability as a function of firm age. In addition, we show the spline smoothed hazard rates for the two ownership categories. The two Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

214

D. Kronborg and S. Thomsen

Figure 1.

Marginal Kaplan-Meier survival curves and spline smoothed estimated hazard rates for domestically owned companies and foreign subsidiaries

ownership categories seem to have similar hazard rates for the first 10 years of age, while the foreign subsidiaries have lower exit risk for older companies. We can confirm this visual impression statistically. Overall, the domestic companies have higher exit risk. The estimated relative risk is 1.45 (95% confidence limits: (1.19, 1.76)), and the usual score test for comparison of two survival curves results in a test statistic equal to 14.1, which with one degree of freedom corresponds to p = 0.0002. However, neither the above estimate of the relative risk nor the test statistic utilizes the matched design. Taking the matching into account by applying the stratified Model (1) the overall relative risk of exit is increased to 1.96 (95% confidence limits: (1.40, 2.75)) and the test statistic becomes highly significant, p < 0.0001 (see Table 3). In other words, Hypothesis 1 is strongly supported. The relatively wide confidence limits for the relative risk are attributable to the fact that the ‘effective’ sample size is reduced to 252 pairs when analyzing the stratified model. Since the two hazards are very similar for the first age decade we cannot support the hypothesis that domestic companies have an advantage in the time period just after establishment because of greater familiarity with the institutional environment. But higher relative risk for older firms indicates that the costs of foreignness decrease with the Copyright  2008 John Wiley & Sons, Ltd.

age and experience of the company. We can confirm the visual impression analytically. Utilizing the stratified matched sample design, we find that the exit risk of domestic companies was 2.0 times higher than that of foreign companies (p < 0.0001) for companies older than 10 years, while the hazard risk for domestic and foreign companies did not differ (p = 0.282) for younger companies until 10 years of age. We report our statistical tests in Table 3. First, we report the relative risk as a function of year of entry. Our sampling period is divided into five periods (according to time of entry) where foreign investment intensity was almost constant. Ignoring the first period, we find a decreasing trend over the last century. From a value of 2.0 on average for the first three decades, domestic relative exit risk increased to around 10 in the 1930s–1950s, but has since dropped steadily to a point where domestic firms now have the same exit risk as foreign companies. We interpret this as an effect of increasing competition among foreign subsidiaries (cf. below). The relative risk of domestic companies is highest during the protectionist 1930s, the war economy (1940s) and the subsequent rationing period (1950s), which demonstrates the expected importance of policy regimes. The relative risks were found to differ significantly by entry years (p = 0.017). Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Foreign Ownership and Long-term Survival

215

Table 3. Estimates of relative exit risks (domestic/foreign) and likelihood ratio test statistics for comparison of exit risks for domestically owned companies and foreign subsidiaries Analysis of matched model Overall comparison of risk: domestic vs foreign Relative risk (95% confidence limits) Overall Year of entry 1895–1925 1934–1956 1964–1975 1985 1995 Calendar-period 1895–1929 1930–1949 1950–1969 1970–2005 1990–2005

Significance test

No. of pairs (effective sample size)

−2 log LR

p-value

16.2

<0.000

528 (252)

1.96∗∗∗ (1.40, 2.75) 2.00∗ 10.50∗∗∗ 2.43∗ 1.54 1.06

(1.00, (2.45, (1.01, (0.77, (0.54,

4.00) 44.7) 5.85) 3.09) 2.05)

4.08 18.3 4.40 1.50 0.03

0.044 0.000 0.038 0.221 0.866

42 (38) 33 (26) 58 (38) 107 (46) 288 (104)

6.00∗ 4.48∗∗ 3.00 1.57∗ 1.40

(0.72, (1.52, (0.61, (1.07, (0.84,

49.7) 13.3) 14.9) 2.31) 2.31)

3.96 9.64 2.09 5.38 1.67

0.046 0.002 0.148 0.020 0.196

42 (7) 60 (27) 53 (9) 484 (209) 434 (163)

Determinants of relative risk Significance tests controlled for the effect of number of foreign companies at entry time Share capital Main industry

−2 log LR = 0.91, χ 2 (1), p = 0.320 −2 log LR = 6.20, χ 2 (3), p = 0.102

Table 3 also shows that foreign companies that entered in recent years (1985 and 1995) do not have significantly higher survival rates than domestic companies, whereas foreign companies entering earlier had significantly higher survival rates than domestic companies. To examine the exit risk within different calendar periods, the relative risk was estimated in four time periods (stratification). In each period, the risk set consists of pairs of companies where both are active in the beginning of the calendar period and those surviving the calendar period are considered to be censored at the end of the period. In this way some companies may be in the risk set for several periods. However, the estimates of the relative risks are not affected by pairs of companies where both survive a given calendar time period and, consequently, it is possible to compare the calendarspecific relative risks. The calendar-specific relative risks are found not to differ significantly (p = 0.149). Due to the low effective number of pairs of companies, the estimated precision of the relative risks is low—especially in the periods 1895–1929 and 1950–1969 and have to be interpreted cautiously. However, there is a tendency to decreasing differences in exit risk of foreign versus domestic companies. In 1970–2005, the relative risk has Copyright  2008 John Wiley & Sons, Ltd.

dropped to 1.57 (1.07, 2.31). In this period, neither differences in inflation-adjusted share capital nor in main industry have a significant influence on survival after adjusting for ownership (share capital: −2 log LR = 0.36, df = 1, p = 0.61; main industry: −2 log LR = 0.587, df = 3, p = 0.89). For the last part of the period 1990–2005, relative risk has dropped to 1.4, which is not significantly greater than one. In other words the foreign survival premium has disappeared. Excluding bankruptcies from the analysis does not qualitatively alter the results. The overall relative exit risk decreases to 1.87 (1.31, 2.67) and is still significantly different from one (p = 0.0004). Similar to the analyses based on the full dataset, neither share capital nor main industry have a significant influence on the relative exit risk. Moreover, year of entry affects the relative exit risk significantly, whereas the calendar-specific relative risks do not differ significantly. The general picture of the estimated relative risks within groups is almost unaltered, but the estimated risk ratios are slightly lowered. To examine the long-run decline of the foreign survival premium we included both the number of foreign companies overall and the number of foreign companies within the individual Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

216

D. Kronborg and S. Thomsen

Figure 2.

Estimated relative risk (domestic relative to foreign) as a function of number of foreign companies operating at time of entry. Note: The dashed curves are estimated 95% confidence limits

industry sectors active in the calendar year of entry as an explanatory variable in Model (1) as described in (2). Both measures influence the relative risk significantly with almost identical estimated parameters. However, the industry density measure becomes insignificant when controlling for the overall density (p = 0.48). The overall measure of competition intensity influences the relative risk significantly (p = 0.014). The estimated parameters are βˆ0 = 1.163 (SE = 0.272) and βˆ1 = −0.0025 (SE = 0.001). The estimated relative risk as a function of the number of foreign companies is seen in Figure 2. We find that the relative exit risk (domestic/foreign companies) decreases with the number of foreign subsidiaries, which provides additional support for Hypothesis 2. The hazard rate for foreign subsidiaries tends to approach the hazard rate for domestic companies as more foreign subsidiaries enter the market. Regarding the matching variables, we did not find any significant influence on relative risk when controlling for the increased competition. As seen from the bottom of Table 3, analysis of possible interaction between firm size and ownership showed no significant effect of firm size measured by the logs of the inflation-adjusted share capital (p = 0.320). We also analyzed relative risk as a function of industry. Given the matched design and the relatively small effective number of companies, the analysis of differences between industry groups was performed on aggregated industry level within four groups, namely food, chemicals, metal products, and other industries. It appears that the relative exit risk for domestic companies is highest in the chemical industries (estimated relative Copyright  2008 John Wiley & Sons, Ltd.

risk of 2.4 times estimated relative risk for ‘other industries’) and metal industries (estimated relative risk 3.9 times estimated relative risk for ‘other industries’), whereas companies in food industries have almost the same exit risk as ‘other industries’ (estimated relative risk of 1.2 times estimated relative risk for ‘other industries’). This is consistent with the idea that the relative performance of multinational companies is better in industries characterized by proprietary technological advantages. However, the overall industry effect is not statistically significant (p = 0.102). Adding size (inflation-adjusted share capital) to the model does not change the insignificance (p = 0.110).

DISCUSSION To summarize, we find that domestic companies have a significantly higher exit risk—roughly two times higher—than foreign subsidiaries. Thus Hypothesis 1 is supported. Apparently, the benefits of foreign ownership, including access to capital, brands, knowledge and other resources from the parent company, outweigh the costs of foreignness. We cannot confirm the existence of a liability of foreignness, but rather find a foreign survival premium. This is important since previous studies have failed to confirm a survival advantage for multinational subsidiaries although international business research has emphasized the competitiveness of multinational companies (Caves, 1996). We also find that the survival advantages of foreign ownership have been eroded over time and provide evidence that the survival premium Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Foreign Ownership and Long-term Survival is negatively correlated with the number of active foreign subsidiaries (a measure of local density and competition). In fact, in the last decade up to 2005 the survival premium appears to have disappeared altogether. The decline in the relative hazard of domestic companies compared to foreign subsidiaries is not monotonous as a function of entry year. There is actually an increase in the relative hazard rate for new entrants during the protectionist period 1934–1956, when there were very few new entrants because of the policy regime. Although we cannot confirm the existence of a liability of foreignness (Zaheer and Mosakowski 1997), we do find that the foreign survival premium is higher for older companies. This is consistent with learning and better access to host country networks over time (Johanson and Valne, 1977; Barkema et al., 1996). One limitation of our research is that we study a single country, which may not be representative of large strategically important host country markets. In principle, however, this makes our conclusions stand out more robustly since we would expect foreign subsidiaries to be more tenacious in larger markets. Thus it seems quite possible that multinational companies have retained more of their survival advantage in larger national markets, whereas we would expect the foreign survival premium to have declined in smaller markets. This is consistent with Mata and Portugal (2002), who examined a small country setting (i.e., Portugal) during the 1990s and found no survival differences between foreign and domestic companies. Another limitation is that we have many more observations at the end of the period than in the beginning. We therefore need to guard against near-term bias. As for the foreign survival premium, we believe that our results are robust since low relative risk during the past decade would tend to understate rather the overall foreign survival premium, which nevertheless remains economically as well as statistically significant. As for the decline in relative risk (Hypothesis 2), we note that the decline in relative risk in the end period 1990–2005 is consistent with the long-term trend and tracks the entry of foreign subsidiaries. A third limitation is that we compare foreign subsidiaries to domestic companies that tend to be independent entities. Our study is therefore a test of the combined effect of foreignness and corporate ‘subsidiarity’ compared to, for example, Copyright  2008 John Wiley & Sons, Ltd.

217

Pennings et al. (1994) who study the effects of foreignness within a sample of subsidiaries. However, we maintain that our choice of control group is relevant for the many strategy and policy decisions that involve a choice between foreign ownership by a multinational parent company and continuation of the domestic company as an independent entity. In this case, our results are inconsistent with commonly held beliefs that foreign ownership is more likely to lead to closures. Subject to confirmation by further research on other countries, our findings have potentially important implications for both theory and practice. Theoretically, we find support for the the long-term survival advantage of multinational subsidiaries, which previous studies have been unable to uncover. By covering decades of evidence and by careful matching, we have been able to disentangle age and time effects. As predicted by theory (Zaheer and Mosakowski, 1997) the foreign survival premium is much larger for older companies. However, we have also found that the advantages of foreign ownership have been eroded over time so that the much-cited advantages of international ownership may be more descriptive of the past than of the future. These findings clearly point to the importance of dynamic effects for international business research. Process theories that emphasize age effects and internal learning in the multinational enterprise need to be supplemented by attention to external learning among competitors and new entry as a consequence of Schumpeterian dynamics. For strategic management, the most important implication—again subject to confirmation by further research—may be that the survival advantages of foreign subsidiaries appear to have been eroded over time and are now no longer significant. Internationalization strategies aimed at establishing and maintaining foreign subsidiaries therefore need to be justified based on the specific business case rather than as a generic strategy component. Obviously, this does not mean that multinationals are no longer competitive. In fact, some of the reduction is attributable to increasing ability to service the domestic market by subsidiaries in other countries (Richbell and Watts, 2000). But the declining foreign survival premium does probably imply that the historical advantages to foreign direct investment have been eroded over time by increasing internationalization. It is clear that past successes Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

218

D. Kronborg and S. Thomsen

do not necessarily continue into the future, and this also applies to foreign direct investment.

ACKNOWLEDGEMENTS We thank the two anonymous referees and Editor Will Mitchell for valuable comments that have improved the article significantly, both in terms of theory and empirical analysis. We are also grateful to Lars Hammer Jespersen and Kristian Jacobsev for research assistance.

REFERENCES Agarwal R, Gort M. 1999. The determinants of firm survival. Available at Social Science Research Network: http://ssrn.com/abstract=167331.DOI:10.2139/ssrn. 10.2139/ssrn.167331 Andersen PK, Borgan Ø, Gill RD, Keiding N. 1993. Statistical Models Based on Counting Processes (Springer Series in Statistics). Springer-Verlag: New York. Anderson E, Gattignon H. 1986. Modes of foreign entry: a transaction cost analysis and propositions. Journal of International Business Studies 27(3): 1–26. Barkema HG, Bell JHJ, Pennings JM. 1996. Foreign entry, cultural barriers, and learning. Strategic Management Journal 17(2): 151–166. Boje P. 2000. Denmark and Multinational Companies before 1950 . Odense University: Odense, Denmark. Brannen MY. 2004. When Mickey loses face: recontextualization, semantic fit, and the semiotics of foreignness. Academy of Management Review 29(4): 593–616. Breslow NE. 1996. Statistics in epidemiology: the casecontrol study. Journal of the American Statistical Association 91(433): 14–28. Buckley PJ, Casson MC. 1976. The Future of the Multinational Enterprise. Palgrave Macmillan: Hampshire, UK. Caves RE. 1996. Multinational Enterprise and Economic Analysis (Cambridge Surveys of Economic Literature (2nd edn). Cambridge University Press: New York. Caves RE. 1998. Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature 36(4): 1947–1982. Chandler A. 1990. Scale and Scope. The Dynamics of Industrial Capitalism. Harvard University Press: Cambridge, MA. Chang S, Park S. 2005. Types of firms generating network externalities and MNC’s co-location decisions. Strategic Management Journal 26(7): 595–615. Chung CC, Beamish PW. 2005. The impact of institutional reforms on characteristics and survival of foreign subsidiaries in emerging economies. Journal of Management Studies 42(1): 35–62. Copyright  2008 John Wiley & Sons, Ltd.

Cox DR. 1972. Regression models and life-tables. Journal of the Royal Statistical Society, Series B (Methodological) 34(2): 187–220. Cox DR. 1975. Partial likelihood. Biometrika 62: 269–276. Delios A, Beamish PW. 2001. Survival profitability: the roles of experience and intangible assets in foreign subsidiary performance. Academy of Management Journal 44(5): 1028–1038. Delios A, Beamish PW. 2004. Joint venture performance revisited: Japanese foreign subsidiaries worldwide. Management International Review 44(1): 69–92. Delios A, Makino S. 2003. Timing of entry and the foreign subsidiary performance of Japanese firms. Journal of International Marketing 11(3): 83–105. Dhanaraj C, Beamish PW. 2004. Effect of equity ownership on the survival of international joint ventures. Strategic Management Journal 25(3): 295–305. Dunning J. 1981. International Production and the Multinational Enterprise. George Allen & Unwin: London, UK. Greens- Børsens. Various years. Greens- Børsens H˚andbogen om Dansk erhvervsliv (formerly Greens Danske Fonds og Aktier[Green’s Handbook of Danish Funds and Stocks]). Børsens Forlag: Copenhagen, Denmark. Hannan MT, Freeman JH. 1977. The population ecology of organizations. American Journal of Sociology 83: 929–984. Hennart J-F. 1991. Transaction costs theory and the multinational enterprise. In The Nature of the Transnational Firm (2nd edn) Pitelis C, Sudgen R (eds). Routledge: London, UK; 72–118. Hymer SH. 1976. The International Operations of National Firms: A Study of Direct Foreign Investment (MIT Monographs in Economics). MIT Press: Cambridge, MA. Johanson J, Vahlne JE. 1977. The internationalization process of the firm: a model of knowledge development and increasing foreign market commitments. Journal of International Business Studies 8(1): 23–32. Jones G. 2005. Multinationals and Global Capitalism: From the Nineteenth to the Twenty-first Century. Oxford University Press: Oxford, UK. Jovanovic B. 1982. Selection and the evolution of industry. Econometrica 50(3): 649–670. Jovanovic B, MacDonald GM. 1994. The life cycle of a competitive industry. Journal of Political Economy 102(2): 322–347. Kalbfleisch JD, Prentice RL. 2002. The Statistical Analysis of Failure Time Data (Wiley Series in Probability and Statistics) (2nd edn). Wiley: New York. Klepper S. 1996. Entry, exit, growth, and innovation over the product life cycle. American Economic Review 86(3): 562–583. Knickerbocker FT. 1973. Oligopolistic Reaction and Multinational Enterprise. Harvard Business School Press: Boston, MA. Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj

Foreign Ownership and Long-term Survival Kogut B, Kulatilaka N. 1994. Operating flexibility, global manufacturing and the option value of a multinational network. Management Science 40: 123–139. Kogut B, Singh H. 1988. The effect of national culture on the choice of entry mode. Journal of International Business Studies 19(3): 411–432. Kostova T. 1998. Success of the transnational transfer of organizational practices within multinational Companies. Working Paper 98-4, Carnegie Bosch Institute for Applied Studies in International Management: Pittsburgh, PA. Available at: http://cbi.tepper.cmu.edu/papers/cbi workingpaper1998 04.html Kostova T, Srilata Z. 1999. Organizational legitimacy under conditions of complexity: the case of the multinational enterprise. Academy of Management Review 24(1): 64–81. Leahy D, Pavelin S. 2003. Follow-my-leader FDI and tacit collusion. International Journal of Industrial Organization 21(3): 439–453. Li J. 1995. Foreign entry and survival effects of strategic choices on performance in international markets. Strategic Management Journal 16(5): 333–351. Li J, Guisinger S. 1991. Comparative business failures of foreign-controlled firms in the United States. Journal of International Business Studies 22(2): 209–225. Lieberman MB, Asaba S. 2006. Why do firms imitate each other? Academy of Management Review 31(2): 366–385. Lieberman MB, Montgomery DB. 1988. First-mover advantages. Strategic Management Journal , Summer Special Issue 9: 41–58. Lieberman MB, Montgomery DB. 1998. First-mover (dis)advantages: retrospective and link with the resource-based view. Strategic Management Journal 19(12): 1111–1125. Lu J, Beamish P. 2004. International diversification and firm performance: the s-curve hypothesis. Academy of Management Journal 47(4): 598–609. Mascarenhas B. 1997. The order and size of entry into international markets. Journal of Business Venturing 12(4): 287–299. Mata J, Portugal P. 2002. The survival of new domestic and foreign-owned firms. Strategic Management Journal 23(4): 323–343. Mata J, Portugal P, Guimaraes P. 1995. The survival of new plants: start-up conditions and postentry evolution. International Journal of Industrial Organization 13(4): 459–481. Mezias JM. 2002. Identifying liabilities of foreignness and strategies to minimize their effects: the case of labor lawsuit judgments in the United States. Strategic Management Journal 23(3): 229–244. Miller S, Eden L. 2006. Local density and foreign subsidiary performance. Academy of Management Journal 49(2): 341–355.

Copyright  2008 John Wiley & Sons, Ltd.

219

Miller SR, Parkhe A. 2002. Is there a liability of foreignness in global banking? An empirical test of banks’ x-efficiency. Strategic Management Journal 23(1): 55–75. Mitchell W, Shaver JM, Yeung B. 1994. Foreign entrant survival and foreign market share: Canadian companies’ experience in United States medical sector markets. Strategic Management Journal 15(7): 555–567. Nachum L. 2003. Liability of foreignness in global competition? Financial service affiliates in the city of London. Strategic Management Journal 24(12): 1187–1208. Pennings JM, Barkema H, Douma S. 1994. Organizational learning and diversification. Academy of Management Journal 37(3): 608–640. Porter ME. 1979. The structure within industries and companies’ performance. Review of Economics and Statistics 61(2): 214–227. Richbell SM, Watts HD. 2000. Plant closures in multiplant manufacturing firms: adding an international perspective. Management Decision 38(2): 80–89. Schumpeter JA. 1950. The process of creative destruction. In Capitalism, Socialism and Democracy (3rd edn). Harper & Brothers: New York; 81–86. Shaver JM. 1995. The influence of industry growth and foreign entry rate on foreign direct investment survival. Academy of Management Journal, Best Papers Proceedings: 201–205. Shaver JM. 1998. Accounting for endogeneity when assessing strategy performance: does entry mode choice affect FDI survival? Management Science 44(4): 571–585. Stinchcombe A. 1965. Social structures and organizations. In Handbook of Organizations, March JG (ed). Rand McNally: Chicago, IL; 142–193. Stouffer SA. 1950. Some observations on study design. American Journal of Sociology 55(4): 355–361. Thomsen S. 2000. Survival of foreign subsidiaries and domestic companies in Danish manufacturing 1895–1995. The Scandinavian Economic History Review 48(2): 72–86. Thomsen S, Pedersen T. 1998. Industry and ownership structure. International Review of Law and Economics 18(4): 385–402. World Bank. 2005. Statistical Appendix. World Development Report 2006: Equity and Development. World Bank: Washington, D.C. Zaheer S. 1995. Overcoming the liability of foreignness. Academy of Management Journal 38(2): 341–364. Zaheer S, Mosakowski E. 1997. The dynamics of the liability of foreignness: a global study of survival in financial services. Strategic Management Journal 18(6): 439–464.

Strat. Mgmt. J., 30: 207–219 (2009) DOI: 10.1002/smj