The Role of the State in Explaining the Internationalization of Emerging Market Enterprises

While the competitive advantages of firms from developed economies are well understood, knowledge of the advantages that enable emerging market enterprises (EMEs) to expand overseas remains limited. Our analysis goes beyond theorizing that focuses on firm resources, enhancing the understanding of how EMEs expand abroad by internalizing home‐country institutional advantages that extend beyond the firm boundaries. More specifically, we examine how the state and institutional idiosyncrasies in the home country help EMEs internationalize. We demonstrate that state ownership has a strong independent effect on the international expansion of EMEs. This effect, however, is contingent upon firms' own resources and other location‐ and industry‐specific forces pertaining to the market orientation of each subnational region and the institutional policies within a given industry.


Introduction
Although globalization was for several decades driven by multinationals from developed countries such as the UK and the USA, emerging market enterprises (EMEs) have recently expanded into new markets through outward foreign direct investment (OFDI). The internationalization of EMEs as a new phenomenon challenged the previous explanations and underlying assumptions of established theorizing on the subject (Wright, Filatotchev and Hoskisson, 2005) and led to an interesting puzzle: how can firms that only rarely possess strong resources and advantages internationalize and compete with multinational enterprises (MNEs) from developed countries? While prior research on developed country MNEs focused on the exploitation of firm-specific assets to explain cross-border expansion (Buckley and Casson, 1976), EMEs differ significantly from their counterparts in the West. Not only do they not possess typical advantages such as technological capabilities and strong brand names Dunning and Lundan, 2008a), but they also operate in under-developed institutional environments in their home country. Because emerging economies are characterized by high levels of state ownership and weak market-based mechanisms, the state often acts as a functional substitute for market failure, controls the opportunities given to EMEs and shapes their strategic choices, including internationalization. As a result, EMEs often need to pursue unconventional non-market strategies to exploit governmentrelated advantages and compensate for their weaknesses (Luo, Xue and Han, 2010;Rugman and Verbeke, 1992;Wang et al., 2012a).
The peculiar characteristics of EMEs prompt the need for a new explanation of the sources of competitive advantages that enable such firms to internationalize. Although the literature has acknowledged that the state may assist EMEs' international expansion (Cui and Jing, 2012;Ramasamy, Yeung and Laforet, 2012;Wang et al., 2012b), knowledge of how and under what institutional circumstances such effects occur is rather limited. Indeed, the role of the state -which according to institutional theory may influence the objectives, priorities, resources and capabilities of the firm -remains under-theorized. To increase understanding of this phenomenon, we develop and test a multilevel framework explaining why the involvement of the state in facilitating EME internationalization is contingent upon location-, industry-and firm-specific idiosyncrasies.
Our study differs from and contributes to previous research in two ways. First, most research on internationalization implicitly assumes that there is institutional homogeneity within a given nation (Aguilera, 2005), i.e. although it acknowledges the existence of institutional variations across countries, it usually assumes that institutional forces remain similar across different locations and industries in a given country. By contrast, we argue that there are significant within-country institutional variations (Gao et al., 2010;Ma, Tong and Fitza, 2013;Meyer and Nguyen, 2005) and advance the premise that this important source of variation shapes EMEs' inclination and ability to internationalize by affecting the environment in which strategic decisions are taken. Therefore, instead of merely asking the generic question of whether the state matters, we demonstrate that EMEs' internationalization is driven by location-and industryspecific institutional contexts that are not equally beneficial for all EMEs. This approach enables us to bring separate views on firm-level, industrylevel and subnational region-level institutional determinants of international business within one unifying framework.
Second, previous studies have provided insightful explanations as to how state ownership results in different advantages and therefore in overseas expansion (Cui and Jiang, 2012;Dunning and Lundan, 2008a;Ramasamy, Yeung and Laforet, 2012). Prior analyses, however, often rely on the assumption that all EMEs are equally able to use institutional factors to implement their international plans. We argue that as EMEs possess different resources and capabilities, they differ in their capacity to respond to institutional pressures and exploit government-related advantages and therefore vary in their ability to internationalize successfully. More specifically, we theorize that the effect of the state on EMEs' internationalization is contingent upon the firms' own technological and marketing resources. We therefore propose that variations in internationalization exist not only as a result of institutional forces and government related advantages but also because of the resources that enable the firm to respond to isomorphic pressures and make use of such advantages. By modelling the internal effects of institutions on internationalization in this manner, we extend international business theory (e.g. Dunning, 1993) that recognizes that firms may derive competitive advantages from home institutions but does not specify what drives the firm's ability to exploit such advantages. To test our conceptual framework, we use firm-and project-level data for a large emerging economy (namely China). Although we acknowledge that emerging economies differ from one another, we anticipate that some of the predictions of our framework could be adapted to other emerging economies.

Theoretical background
As institutions -the political, legal, economic and social rules within countries (Scott, 1995) -shape an organization's decisions, the determinants of internationalization extend beyond economic optimization and strategic justification (Oliver, 1997;Peng, Wang and Jiang, 2008). An institutional environment can be conceptualized as a set of three pillars (Scott, 1995): regulatory (the legal system and its enforcement), normative (the established norms and professionalization) and cultural-cognitive (shared social beliefs and values). These three pillars generate isomorphic pressures that may encourage or constrain the behaviour and actions of organizations. Hence, according to institutional theory (Scott, 1995), firm strategy (including internationalization) is

46
J. Hong, C. Wang and M. Kafouros driven not only by firm resources and capabilities, but also by various isomorphic pressures and the need to meet environmental demands. Recent advances in institutional theory challenge the deterministic view of organizations, emphasizing the importance of active agency over embedded institutional factors. Because firms differ in the models they adopt and in their degree of independence from institutions, they may choose to respond to institutional forces in different ways (Kostova, Roth and Dacin, 2008). Firms also co-evolve with institutions as they can exert significant influence on the institutional environment by managing non-market relationships (Boddewyn, 2003;Cantwell, Dunning and Lundan, 2010;Wang et al., 2012b). Such interfirm variations in the effects of institutions are particularly important in emerging economies that experience political and institutional reforms and institutional voids (Cantwell, Dunning and Lundan, 2010). These variations make an institutional approach particularly relevant to the analysis of EME internationalization (Gaur and Kumar, 2009;Khanna and Rivkin, 2001).
On the other hand, studies that use the resource-based view (e.g. Peteraf, 1993) suggest that differences in international expansion are the result of differences in firms' resources and capabilities. According to this reasoning, foreign expansion represents a firm's attempt either to internalize the use of existing resources − such as technology and brand names − in international markets (Buckley and Casson, 1976;Hsu and Pereira, 2008) or to acquire new assets (Gaur and Kumar, 2009;Kafouros, Buckley and Clegg, 2012). The resource-based view is a useful theoretical avenue for explaining how the availability and combination of resources helps firms internationalize. The resource-based view and the institutional-based view can complement one another in explaining EMEs' internationalization as differences in firm resources may also influence a firm's inclination and ability to respond to regulatory, normative and cultural-cognitive pressures. For instance, firms with strong resources are usually more able to respond to such isomorphic forces (Scott, 1995), exploit government linkages (Sun, Mellahi and Thun, 2010), take advantage of intermediary state services and influence regulations to create favourable conditions (Danneels, 2002;Luo, Xue and Han, 2010;Wang et al., 2012b). By contrast, firms with limited resources may respond to isomorphic pressures in an ad hoc manner (Winter, 2003) and may struggle to pursue institutional conformity.
Hence, although state ownership can be an important institutional advantage (Wang et al., 2012b), the ability to manage regulatory, normative and cultural-cognitive pressures and benefit from non-market relationships depends on a firm's resources (Sun, Mellahi and Thun, 2010). Thus, variations in internationalization are a result of three things: firm resources, relationships with government and the ability to leverage such resources to benefit from non-market relationships. This prediction is consistent with established ownership-based theories of MNEs. For instance, while Dunning's (1993) OLI paradigm focuses more on the firm as an economic agent and less on how it is embedded in its environment, it posits that ownership advantages can also be built through interacting with the firm's home institutional environment. Dunning andLundan (2008a, 2008b) re-conceptualized the term 'ownership advantages' explicitly to incorporate not only typical advantages such as technology, but also institutional advantages including firmspecific norms and values and the institutional environment of the firm. In a similar vein, Rugman and Verbeke (1992) suggest that the international configuration of MNEs hinges upon home-country-specific advantages. This work implies that EMEs with weak firm-specific advantages may still be able to internationalize by leveraging home-country advantages.

Theoretical framework and hypotheses
The role of the state in emerging and developed markets Emerging economies are countries that exhibit a high rate of economic growth and have started favouring economic liberalization and international openness (Arnold and Quelch, 1998;Hoskisson et al., 2000). Although the role of the state is important in both emerging and developed countries, for various reasons the role of government tends to be much more influential in emerging markets, particularly with respect to EMEs' internationalization. First, whilst governments in developed countries keep their direct involvement in businesses at a lower level, the degree of state ownership in most emerging countries remains

Internationalization of Emerging Market Enterprises 47
high. Second, because markets in emerging economies are more recently established than those in developed countries, government involvement is usually stronger and more influential in emerging economies, thus increasing regulatory and coercive pressures (Child and Rodrigues, 2005;Dunning and Narula, 1996;Hoskisson et al., 2000). Third, due to the limited experience, resources and capabilities of EMEs, government is often a key force driving EMEs' internationalization (Meyer and Peng, 2005;Peng, Wang and Jiang, 2008;Ramasamy, Yeung and Laforet, 2012;Wang et al., 2012aWang et al., , 2012b. By contrast, the internationalization of firms in developed economies is largely a market-driven strategic choice that is not so much influenced by institutions and government. However, although the role of the state tends to be more important in emerging countries, we should also recognize that emerging countries differ from one another in terms of institutional development, market reforms and government structure. Due to these variations, the role of government might be more prominent in some emerging economies than in others. For instance, prior research suggests that government involvement is particularly prominent in large emerging economies such as China, India and Indonesia (Dunning and Narula, 1996).

State ownership and EMEs' internationalization
The state can influence firm internationalization through macro-level policies and firm-level ownership arrangements. Governments, as the regulators of the business environment (Luo, Xue and Han, 2010), create policies that facilitate or constrain internationalization. Governments may also support EMEs' internationalization through international treaties that protect OFDI (Luo, Xue and Han, 2010) and incentives that encourage certain industries to expand overseas. For example, although the Chinese government promotes OFDI in certain sectors through tax rebates, foreign exchange assistance and financial support, it also creates regulatory and coercive pressures using legislation to limit firm internationalization in other industries. Furthermore, isomorphic pressures that influence internationalization differ depending on the level of the government involved (e.g. state, provincial and citylevel) (Wang et al., 2012b). While central government formulates regulatory frameworks to guide internationalization, ease capital controls and provide information and guidance on investment opportunities, governments at lower levels are responsible for implementing central government's policies (Kumar and Worm, 2004) by, for example, using incentives to encourage and direct EMEs to expand abroad.
Governments also influence international expansion through direct ownership in firms. Industry-level forces affect most organizations in a given environment, whereas institutional pressures through state ownership are specific to the firm. As a main shareholder in a firm, government impacts EMEs' internationalization more directly to achieve its political, economic and social objectives. Since state ownership represents an important institutional attribute (Zhang, Zhou and Ebbers, 2011), the associated institutional pressures increase the scope for deviations from profit-maximizing objectives, push EMEs to make decisions that are not always economically optimal and influence both the inclination and ability of EMEs to internationalize.
A government may affect EMEs' inclination to internationalize by creating normative pressures for state-owned enterprises (SOEs) to comply with the objectives of the state and gain legitimacy. As a result, managers in SOEs try to implement state policy and often internationalize in order to accommodate political objectives, improve geopolitical standing and 'show the national flag' (Deng, 2009;Wang et al., 2012b). Such institutional conformity is driven by normative expectations in the sense that internationalization may benefit SOE managers' career and rewards. In addition, because emerging country governments increasingly want to integrate their nations in the global economy, the coercive and normative pressures that SOEs face increase (Peng, Wang and Jiang, 2008). Since SOEs are particularly sensitive to such influences, state ownership plays a crucial role in affecting EMEs' inclination to expand abroad.
The state also influences the ability of EMEs to internationalize. The firm's ability to implement new strategic initiatives, such as internationalization, depends on its access to new and complementary assets and resources (Pfeffer and Salancik, 1978). Because the market for critical inputs in emerging economies is under-developed, state ownership comes with advantages (e.g.

48
J. Hong, C. Wang and M. Kafouros stronger protection of property rights, better public provision, insider information and control over important assets) that help EMEs offset ownership disadvantages in foreign markets (Aggarwal and Agmon, 1990;Dunning and Lundan, 2008a). In contrast, non-SOEs are less likely to access these advantages through nonmarket channels and therefore have to rely on internal resources to internationalize (Nee, 1992). Therefore, as state ownership positively influences both the inclination and ability of EMEs to expand abroad, firms with a higher degree of state ownership are more likely to expand overseas. Nevertheless, because the involvement of the state is not equally beneficial to all firms (Peng and Luo, 2000), a contingency analysis is required to capture such variations. Although we expect state ownership, firm-specific resources and other institutional factors to influence OFDI independently, we further hypothesize that the effects of state ownership on international expansion are not uniform across EMEs but rather depend upon the EME's own resources and other location-and industry-specific institutional idiosyncrasies ( Figure 1). We accordingly develop a number of hypotheses.

Location-specific institutional variations: the role of marketization
Marketization refers to the extent to which a business environment is driven by market forces and firms, rather than the state and government. Hence, a higher degree of marketization implies that the role of government involvement is less influential. In contrast to the typical view that institutions within a given nation are similar, emerging countries exhibit significant crossregional variations in the extent to which the state is involved in the coordination of economic actors. In regions with well-established institutions, markets function well and interruptions from governments are minimal. Hence, uncertainty and transaction costs are lower than they are in regions with a lower level of marketization. These variations are particularly significant in countries such as China which are changing their structure from a centrally planned economy to a

State ownership Firm's overseas investment
Control variables -Firm size -Firm age -Group affiliation -Foreign ownership -Export intensity -Human resources -Industry effects -Time effects

Institutional environment
Firm-specific resources market economy (Luk et al., 2008). As a result, the transition process in these countries is spatially uneven (He, Wei and Xie, 2008). While some regions exhibit the characteristics of a command economy (i.e. the state has control over the market and owns many organizations), nonmarket influences and state interventions are less common in market-oriented regions. Indeed, extant research suggests that regions within China differ in the pace and extent of market reforms (Wei and Hao, 2010). For instance, local governments in the coastal provinces of China (which are more market oriented than inland provinces) intervene less frequently in firms' decisions (NERI, 2001). Accordingly, the relationship between state ownership and foreign investment varies significantly across regions. In regions where the degree of marketization is higher, firms resemble their counterparts from developed economies. The coercive pressure that they receive from government is lower. They also rely on the principles of a market economy and take advantage of markets for coordinating factors for internationalization. As the institutional context evolves and becomes market oriented, firms move from institutionbased strategies to resource-based strategies (Hoskisson et al., 2000). We therefore expect internationalization to be largely driven by the firm itself rather than by government involvement. Because of market reforms and improved governance, SOEs in these regions are likely to become 'market' oriented and organize themselves around incentive-based structures that promote performance. By contrast, regions with a lower level of marketization are more likely to have under-developed institutions, which reduce competitive and imitative pressures (Chacar and Vissa, 2005). Firms in such regions are significantly influenced by regulatory forces and are therefore keen to build connections with government. Government officials may be particularly influential and may even coerce firms to align their goals with government interests. Government may use its power to implement objectives, including globalization, through means such as direct provision of critical resources and subsidies to support internationalization. In these cases, internationalization is likely to be institutionally embedded and supported by the regulatory and normative pillars of the institutional environment. Overall, we expect the degree of marketiza-tion in a region to moderate negatively the effects of state ownership on a firm's overseas investment: H1: The effects of state ownership on the level of overseas investment will be weaker in subnational regions with a higher degree of marketization than in subnational regions with a lower degree of marketization.

Institutional variations across industries
Because each industry is coordinated by a unique configuration of institutional arrangements (Hollingsworth, 2000), there are significant institutional variations across industries. Industries in emerging economies vary considerably in their nature and the amount of support that they receive from the state. OFDI is encouraged and subsidized in some industries, but it is not a priority for government in other sectors. We hypothesize that such variations are an important source of different competitive advantages that may result in contrasting internationalization outcomes. We thus expect the relationship between state ownership and overseas investment to vary across sectors. In industries where governments do not place emphasis on internationalization, political interests, normative forces and institutional policies on OFDI are not always consistent with the decision of the SOE to invest overseas. These coercive pressures may lead to conflicts between managers and government officials and may change the firm's internationalization plans − particularly in SOEs where direct government involvement is high.
By contrast, state ownership will have a strong positive effect on overseas investment for firms that operate in sectors where global integration is a priority. Firms and SOEs in these industries respond to coercive and normative pressures and gain legitimacy by implementing their international strategies because such objectives are compatible with those of government (Oliver, 1997). This may lead to synergistic effects between institutional policies and firms' strategic planning. The process of getting government approval for OFDI projects is likely to be easier and less time consuming in those industries. Hence, higher levels of state ownership will enable such firms to benefit disproportionately from institutional policy and from industry-specific initiatives that support overseas expansion (Luo, Xue and Han, 2010). Overall, although government will only rarely block firms' foreign investment entirely, we hypothesize that the positive relationship between state ownership and internationalization will be stronger in sectors where OFDI is a priority: H2: The effects of state ownership on the level of overseas investment will be stronger in firms that operate in industries where OFDI is encouraged by government than in firms that operate in other industries.

The moderating role of technological and marketing resources
Furthermore, the effects of state ownership on OFDI vary across firms, depending on the firms' technological and marketing resources. Technological resources refer to the assets used to develop new products and innovative processes (Moorman and Slotegraaf, 1999), while marketing resources are those which help firms differentiate their products from competitors, build brand names (Erramilli, Agarwal and Kim, 1997;Kotabe, Srinivasan and Aulakh, 2002) and better position themselves in the product value chain. Both types of resources facilitate EMEs' internationalization because they involve high levels of specificity (Dierickx and Cool, 1989). R&D, for instance, enables firms to create ambiguity and barriers to imitation (Reed and DeFillippi, 1990), to develop innovative technologies to differentiate themselves from international rivals (Kafouros and Buckley, 2008) and to introduce a steady stream of new products (Dutta, Narasimhan and Rajiv, 1999). Similarly, marketing resources encourage internationalization by enabling firms to market their products abroad, enhance their bargaining power with suppliers (Kotabe, Srinivasan and Aulakh, 2002) and influence foreign customers' choice behaviour (Dutta, Narasimhan and Rajiv, 1999).
The state may complement the firm's innovation efforts by improving framework conditions and by providing critical resources (Mahmood and Rufin, 2005). As SOEs are closely associated with government, they have access to public R&D, patents and other government-controlled intangible assets not available to other companies (Wang et al., 2012b). These can in turn help firms develop distinctive structures, respond to isomor-phic pressures more effectively (Scott, 1995) and accelerate the organizational learning needed to compete in international markets. Furthermore, the level of intellectual property protection depends on the status of the firm, which means that regulatory pressures also differ across firms (Li, Park and Li, 2004). State-owned firms are well protected by government, capturing the value of innovation. The state also helps EMEs acquire scientific talent and technology from foreign MNEs (Hitt et al., 2004). Yet, not all firms can exploit these institutional advantages equally. We argue that this ability is a function of the firm's own technological resources. Although SOEs can access intermediary state services (Khanna, Palepu and Sinha, 2005), the exploitation of these benefits depends upon firms' own R&D because this enhances the ability to identify external technology providers and successfully internalize the knowledge that government agencies and public research institutes can provide.
Marketing activities such as international market research and contract protection involve high risk and uncertainty (Khanna and Palepu, 2000). The state can reduce such risks by bringing EMEs in contact with foreign institutions and investors and by reducing the cost of market research (Malik and Kotabe, 2009). It can also provide information about consumer preferences (Moorman and Slotegraaf, 1999) and facilitate channel relationships that may create barriers to entry. SOEs can further reduce the cost of marketing by accessing intelligence that the state collects regarding foreign opportunities and product attributes (Khanna, Palepu and Sinha, 2005). EMEs can therefore increase their understanding of foreign consumers' preferences and reduce the uncertainty associated with foreign markets. However, to benefit from such governmentendowed marketing resources and intelligence, firms must possess complementary marketing resources (Morgan, Vorhies and Mason, 2009). Such resources enable firms to react to the coercive and normative pressures of internationalization and help them integrate such strategic ingredients with other internal resources or institutional advantages in ways that enhance their international competitiveness (Amit and Shoemaker, 1993). By contrast, these benefits will be less useful for firms that possess weak marketing resources. These firms are less able to change and take full advantage of intermediary services,

Internationalization of Emerging Market Enterprises 51
information and other resources provided by governments. Hence: H3: The effects of state ownership on the level of overseas investment will be stronger for firms with greater (a) technological and (b) marketing resources than for firms with weaker resources.

Empirical setting and data
We test our conceptual framework using a sample of Chinese manufacturing and mining firms. Starting from virtually no OFDI in 1979, China is now an important contributor to the world's OFDI. Chinese OFDI achieved an average growth rate of 48.94% during our sample period (2005−2007). This growth in OFDI concurred with a transition from a centrally planned economy to a market-based system. Although institutional changes have dismantled many barriers to business operations during this transition, the state is still crucial in shaping business in China and the behaviour of firms. This is evidenced by the fact that SOEs accounted for 82% of Chinese OFDI in 2006 (Yeung and Liu, 2008). Furthermore, China features considerable crossregion and cross-industry institutional differences, allowing us to examine the ways in which within-country variations impact the effects of state ownership on internationalization. China therefore offers an appropriate setting for examining whether and under what conditions the state and institutional forces influence OFDI. Our sample contains 626 Chinese firms including 615 manufacturing firms and 11 mining firms. The data are obtained from two sources. Information on firms' outward investment was obtained from the Ministry of Commerce of China (MCC). Due to a legacy of institutional dependence, OFDI projects by Chinese firms are still subject to governmental approval (Child and Rodrigues, 2005). The Chinese government wants to ensure that such investments are in line with its resource allocation plans and priorities (Child and Lu, 1996) and to reinforce China's global influence (Luo, Xue and Han, 2010). The need for approval is also important because capital accounts are not fully open for free trade in China, meaning that the Chinese government monitors all foreign exchange outflows. All firms that have invested overseas are also required to report data to their corresponding provincial Bureau of Commerce. 1 These data are then reported to the MCC, which has collected such data since 2006. The database provides information on all Chinese firms' OFDI in 2006 and 2007 -a period in which China's OFDI achieved significant growth. The MCC data set provides a wide variety of information, including the name of the parent firm and its foreign subsidiaries, the host country, the total capital of the project and the capital invested by Chinese firms. 2 All other data are obtained from the Annual Report of Industrial Enterprise Statistics (ARIES) compiled by the National Bureau of Statistics. Chinese law requires all firms to submit financial information to the National Bureau of Statistics. This data set provides a wide variety of information on firm ownership structure, group affiliation, industry affiliation, geographical location, establishment year, employment, assets, R&D, advertising, value-added, sales, new product sales and exports for all manufacturing industries. The use of this multi-industry sample increases both variance and the number of observations. The ARIES is a comprehensive firm-level data set, accounting for about 90% of total output in most industries.
Matching the MCC and ARIES databases produces a sample of 676 projects. We added the value of all projects together for firms that had invested in both years (43 firms in total). This produces a sample of 633 firms. After dropping four firms for missing observations and three outlier firms, we have ended up with a final sample of 626 firms for which there is a complete set of information about the empirical variables. To construct our dependent variables, we use data for 2006−2007. Our independent variables are lagged by one year and therefore are based on data for 2005−2006. Our sample excluded small investors with annual turnover of less than 5 million renminbi (approximate US$815,661). For this reason, our data set is not representative of very small Chinese firms. 1 Large overseas investment projects need to be reported to the Ministry of Commerce directly. 2 The data set does not allow us to discriminate between different modes of entry by Chinese firms (greenfield versus acquisition).

Methods
Dependent variable. Following many previous studies (e.g. Wang et al., 2012b) and for ease of comparison, we used the actual amount of annual overseas investment undertaken by each firm to measure foreign investment.
Independent variable and moderators. The key predictor variable, state ownership, is measured as the share of state-owned paid-in capital over the total paid-in capital of the firm. We include four variables that we hypothesize may moderate the effects of state ownership. First, a measure of region-specific marketization takes into account the fact that the effects of state ownership may vary across regions depending on the development of market-based mechanisms. This measure developed by Fan, Wang and Zhu (2006) is a comprehensive composite index that evaluates a province's policies and institutions regarding economic freedom in five key areas, including the role of market relative to government, the development of the private sector, the development of commodity and factor markets and the development of free market institutions. 3 Twenty-four indicators are employed to assess these five dimensions and estimate a marketization index. The values of the marketization index, according to Fan, Wang and Zhu (2006), range from 2.89 (Tibet) to 11.80 (Shanghai). The higher the value of marketization, the more developed the marketbased system in a region.
Second, the Chinese government strongly encourages OFDI in some industries. Firms in the 'encouraged' group of industries receive more institutional support and incentives than their counterparts in sectors where OFDI is not a priority. To capture such inter-industry differences in our industry-specific policy, we constructed a dummy variable that distinguishes between the two groups of industries (it equals 1 if the firm operates in an 'encouraged' industry). These industry data were collected from the official document jointly issued by several Chinese government agencies under the State Council. 4 Finally, building on prior studies (e.g. Tseng et al., 2007), technological and marketing resources are captured by R&D expenditures per employee and marketing expenditures per employee, respectively (Dutta, Narasimhan and Rajiv, 1999;Erramilli, Agarwal and Kim, 1997;Gatignon and Anderson, 1988;Tseng et al., 2007;Wang et al., 2012b). As these operationalizations reflect the actual investment made in R&D and marketing, they directly capture the R&D and marketing resources of each firm. By dividing R&D and marketing expenditure by the firm's number of employees, we normalize the measures for firm size.
Control variables. Since differences in size may influence the hypothesized effects (Yiu, Lau and Bruton, 2007), we use the logarithm of the number of employees to control for firm size and skew (Tabachnick and Fidell, 2001). Second, we control for the effects of firm experience by including a measure of firm age, using the number of years since the establishment of the organization. Third, group affiliation is captured by a dummy variable (equal to 1 if the firm is affiliated to a group) (Khanna and Palepu, 2000). 5 Fourth, foreign ownership may stimulate a firm's internationalization due to knowledge spillovers. We control for foreign ownership using the ratio of 3 This index has been adopted by many studies (e.g. Du, Lu and Tao, 2008;Firth et al., 2009;Li, Yue and Zhao, 2009

Internationalization of Emerging Market Enterprises 53
paid-in capital owned by foreign investors to total paid-in capital. Fifth, exporting often represents the initial stage of internationalization and provides valuable information about new markets. We use the ratio of export sales to total sales to control for export intensity. Our model also controls for human resources by measuring each firm's training expenditure per employee. Finally, we introduce dummy variables to account for unobserved industry-and time-specific effects.

Results
Table 1 provides descriptive statistics. All correlations are fairly low, and the average value of the variance inflation factor is well below the acceptable level of 10 (Neter, Wasserman and Kutner, 1985), suggesting that there is no serious multicollinearity problem. Nevertheless, we mean-centred the interaction terms to avoid any potential problems of multicollinearity and increase the interpretability of interactions (Aiken and West, 1991). Since a firm's actions may take some time to influence OFDI, we lagged all independent variables by one year. To deal with possible heteroscedasticity, we estimated ordinary least squares (OLS) regressions using Huber−White's robust standard error (White, 1980). Table 2 reports the results of our hierarchical OLS regression. Model 1 serves as a baseline model that includes only control variables. Firm resources are introduced in model 2. The effects of both resource variables on OFDI are statistically insignificant. Whilst these results contradict conventional wisdom that predicts that firms need to possess intangible resources to conduct FDI, they support the view that firm-specific resources are not always important for the internationalization of EMEs (Wang et al., 2012a;Yiu, Lau and Bruton, 2007). Model 3 incorporates the three institutional variables. Two of them (state ownership and industry-specific policies) are statistically significant, supporting the view that homecountry institutional forces, in particular government involvement, lay the foundations for the international expansion of Chinese firms (Deng, 2009;Luo, Xue and Han, 2010;Wang et al., 2012b).
Models 4−7 present the results for Hypotheses 1−3. Following the usual practice in moderated regression analysis, we enter two-way interactions in these models successively. Model 4 shows that the coefficient of the interaction term between state ownership and location-specific marketization is negative and statistically significant, providing support for Hypothesis 1. Similarly, as the interaction term between state ownership and industry-specific policy (model 5) is positive and statistically significant, Hypothesis 2 is corroborated. The interaction term between state ownership and technological resources in model 6 is insignificant, while that between state ownership and marketing resources is significant (model 7). Therefore, Hypothesis 2a is not supported, whilst Hypothesis 2b is corroborated. To explain better the moderating effects of government involvement, these relationships are presented in Figure 2.

Robustness checks
The possibility that OFDI affects some firm characteristics is less likely as most Chinese OFDI projects have started only recently. Nevertheless, some explanatory variables (e.g. state ownership, technological and marketing resources) may be influenced by OFDI, causing potential problems of reverse causality. We reduce such concerns by incorporating several variables that account for firm characteristics as thoroughly as possible. We also implement a lag structure for independent variables. However, despite these measures, the obtained estimates may still lack some of the required properties. Therefore, we employ the Hausman test (Hausman, 1978) to examine the possibility of simultaneity between explanatory variables and OFDI. We conduct the Hausman test using instrumental variables (IVs). Following standard practice (Gujarati, 2009;Wooldridge, 2009), we use lagged variables for technological resources, marketing resources and state ownership as IVs. We are able to trace back firms from our sample that were operating in 2004 from the China Economic Census. 6 The combination of our data with information from the China Economic Census in 2004 creates a sample of 516 firms. We use these data to test the exclusion restriction that each IV does not affect the dependent variable through channels other than the suspected endogenous variables. This is done by regressing the residuals from the second-stage estimations on the IVs. If the IVs affect the key variables through other mechanisms, the residuals from the second-stage estimations will be correlated with the IVs (Lu and Tao, 2009). Table 3 displays the results. As all the estimated effects on IVs are statistically insignificant, the 6 As China did not conduct industrial census before 2004, data for previous years are not available. *, **, ***Significance level at 10%, 5% and 1% (two-tailed tests), respectively.

Internationalization of Emerging Market Enterprises 55
IVs are indeed orthogonal to the error term. To examine whether the suspected variables are endogenous, we conduct a Hausman test and compare the OLS and two-stage least squares (2SLS) estimates. Results show that the value of chi-squared is 0.06 (degrees of freedom 10), much lower than the critical value of 15.99 at 10% significance level. Therefore, the null hypothesis that the differences in OLS and 2SLS estimated coefficients are not systematic is not rejected, suggesting that key variables should jointly be treated as exogenous and that using OLS is well justified. Further, we include a full model (model 8 in Table 2) that includes all the interaction terms simultaneously, indicating that the results for these interactions remain qualitatively the same.

Discussion
Although it is widely accepted that EMEs differ significantly from their counterparts from developed economies, knowledge of whether and how   (Child and Rodrigues, 2005), scholars have questioned the ability of previous theories to explain the international expansion of firms from emerging economies. Extant research recognizes the role of the state in EMEs' internationalization (e.g.  but implicitly assumes that the impact of the state is similar across the different locations and industries of a given country. We challenge this assumption by demonstrating that institutional forces and the internationalization effects of state ownership are contingent upon location-and industry-specific idiosyncrasies. Our empirical analysis points to the role of the state in unlocking EMEs' potential for international expansion. More importantly, it further shows that state ownership is not equally beneficial for all EMEs, but rather dependent on the firm's own technological and marketing resources. Our findings have several theoretical and managerial implications. First, unlike studies that presume that there is institutional homogeneity within a given country (e.g. Aguilera, 2005), we advance the premise that institutions vary across different locations in the same country. A theoretical implication is that location-specific institutional variations change the role that emerging country governments play in stimulating foreign investment. It also shows that examining institutions at the country-level without considering within-country crossregional differences may underestimate the power of these differences in explaining variations in cross-border expansion. In contrast to the notion that marketization, due to lower agency costs, may strengthen the relationship between government involvement and firm internationalization, our findings indicate that this relationship is stronger in regions with a lower degree of marketization than in regions with a higher degree of marketization. This counter-intuitive finding indicates that as regions evolve and become market oriented, OFDI is largely driven by the firm itself and its resources, rather than by the state. Although international business scholars recognize the importance of within-country differences in regional endowments, an understanding of how subnational institutional effects influence variations in foreign investment remains limited. Our findings regarding the subnational region-specific institutional effects on foreign investment complement and advance the strategic management and international business literature by showing the subnational region to be an important unit of analysis. Hence, while these effects are typically described in international business theory as 'host-country' institutions, our findings indicate that they are in fact subnational region-specific effects that manifest themselves in different ways within the same country.
Second, we find that the effect of state ownership in enhancing foreign expansion is stronger in industries where internationalization is a strategic priority for the state. This finding suggests that state ownership may not have an independent and homogeneous influence on internationalization across subnational regions. This finding is intriguing, indicating that while firms in a given country face the same macro-level institutions (Scott, 1995), these macro-level forces are not capable of differentiating firms' competitive advantages. By contrast, our findings suggest that EMEs derive contrasting international competitive advantages from different industry-specific institutional contexts. Indeed, although emerging market governments have recently adopted a favourable attitude towards OFDI, policy is often implemented at the industry level. This leads to significant variations in regulatory, normative and cognitive institutional settings that either facilitate or impede firms' international plans and strategic choices. Together with the findings of the region-specific effects, our research has theoretical implications with regard to how an organizational field should be defined and how EMEs succeed in internationalizing by exploiting region-and industry-specific institutional advantages.
Third, our study contributes to institutional explanations of internationalization by showing that only firms with strong marketing resources can exploit certain institutional advantages associated with state ownership. Although marketing resources have an insignificant independent effect on OFDI, they play an important role under the presence of state ownership. Thus, the implication here is that the ability of government to stimulate international expansion is dependent upon firms' own resources. Our findings concerning the significant interaction between marketing resources and state ownership suggest that taking advantage of institutional factors depends on the brand Internationalization of Emerging Market Enterprises 57 name and image that the firm develops. In contrast, it appears that technological resources do not help EMEs build competitive advantages and internationalize. Firms often deploy resources in ways consistent with market forces (Moorman and Slotegraaf, 1999). The value of these resources depends on how difficult it is for rivals to copy and obtain them from the market (Krasnikov and Jayachandran, 2008). Due to its socially complex nature, marketing knowledge is more difficult to codify than technological capabilities (Simonin, 1999). Because capabilities based on easily codified processes may not afford an organization the ability to achieve competitive advantage (Krasnikov and Jayachandran, 2008), technological resources have a weaker effect on EMEs' competitive advantages than marketing resources. Another explanation for this result is that EMEs have a good understanding of their home market and customers, but have not yet developed strong technological resources as they largely focus on standardized products (Buckley, Clegg and Wang, 2007). Taken together, our findings suggest that traditional theoretical frameworks that rest upon the advantages that MNEs develop within their organizational boundaries should be augmented to include advantages derived from home institutional environments. In this respect, we extend prior theorizing that views government involvement as a functional substitute for market failure by showing that the state can indeed be an important source of competitive advantages that help EMEs compete in international markets. By conceptualizing the state as an endogenous part of the firm, the development of relationships with government emerges as an important component of firms' strategy. Recognizing that EMEs internationalize in a dynamic manner (Yaprak and Karademir, 2010) and are influenced by multiple institutional forces, our analysis suggests that the value-creating potential of state ownership largely depends on the moderating role of various institutional parameters. This view differs from prior studies that either focus on organizations or institutions, rather than the interaction between the two (Coriat and Weinstein, 2002), or conceptualize internal and institutionally derived capabilities as two separate and independent types of advantages. Our study therefore shifts the debate from whether state ownership and institutions matter for internationalization to the question of how and under what conditions they can be used to develop competitive advantages.

Implications for managers and policy makers
Our findings have implications for EME managers seeking to expand their firms overseas. First, rather than merely focusing on the accumulation and development of traditional firm resources, EMEs should carefully develop a strategic plan and capabilities that will assist firms in managing institutional idiosyncrasies and exploiting external factors, institutional contexts and inputs controlled by the government. Such capabilities may enable EMEs to internationalize even if they do not yet possess the type of resources (e.g. technological assets) that MNEs from developed countries possess. This recommendation differs from the conventional theory of MNEs that suggests that the development of firm-specific competitive advantages is a necessary and sufficient condition for firms to embark on internationalization. Furthermore, the institutional environment is not always exogenously determined and fixed, but can be shaped by firms (Luo, Xue and Han, 2010). Hence, rather than merely reacting to institutional pressures, EMEs should be proactive, internalize government-related advantages and influence institutions in a way that will help their internationalization.
Second, the location choices of the EME and the industry in which it operates may impact its ability to benefit from government involvement. EMEs that establish their operations in regions where the degree of marketization is lower and compete in industries that are encouraged to internationalize are better able to exploit staterelated advantages. Nevertheless, although our findings emphasize the value of state ownership, they do not necessarily suggest that managers should place less emphasis on the development of the firm's own resources. Government involvement in the form of state ownership may lock firms in a particular institutional context that will not assist the development of new capabilities. To overcome this challenge, EMEs must balance the two strategies, be ambidextrous and develop both conventional (internal) resources and capabilities that will enable them to exploit their relationships with the state (Wang et al., 2012b). Indeed, a business model that rests upon the combination of the two will be likely to be the most effective strategy for increasing EMEs' international expansion.
Our findings also have implications for policy. Policy makers in emerging economies expect internationalization to strengthen the global competitiveness of their firms and country. Our findings show that although government can achieve this objective by intervening through state ownership, this practice should not be used as a uniform policy to encourage internationalization. This policy should be adopted only in industries in which OFDI is encouraged by government and in subnational regions where market mechanisms are less developed. State ownership is unlikely to have a similar effect on internationalization in other industries and regions. Furthermore, as firm-specific resources (i.e. marketing resources) enable EMEs to exploit government related advantages, policies that encourage firms to develop such resources will facilitate the role of state ownership.
However, policy makers also have to be aware that what is good for internationalization is not necessarily beneficial for firm growth and the national economy. For example, although welldeveloped market mechanisms may reduce the level of state ownership, they are important conditions for increasing firm capabilities, competitiveness and long-term economic growth. Therefore, although state ownership may facilitate EMEs' internationalization, it should be considered as a temporary policy choice that works in the particular situations of emerging economies.

Limitations and future research
When interpreting the above results, a number of limitations should be considered. First, while prior research has shown that government involvement is particularly important in emerging economies such as India, China, Indonesia and Brazil (Dunning and Narula, 1996;World Investment Report, 2008), our empirical data rely on one country. Although this approach made it possible to control for region-and industryspecific institutional characteristics, our findings may not be equally generalizable to other emerging countries because of various China-specific idiosyncratic characteristics, including the strong role of the state. As market forces and government interventions vary from country to country (Mahmood and Rufin, 2005), the ability to gener-alize our framework and predictions depends on the level of market and institutional development of the country in question as well as on the extent to which institutional differences persist across regions and industries. Hence, a useful avenue for future research will be to explore whether our hypothesized relationships hold true in other emerging economies. Second, the degree of globalization varies between industries, implying that the forces driving internationalization may differ in manufacturing and service sectors. However, data limitations did not allow us to test this proposition and compare how the findings differ between these two sectors. A final concern is the potential endogeneity that can typically be found in this type of research. For instance, while the deployment of resources may lead to OFDI, international expansion may in turn facilitate the accumulation of new resources. Although our robustness analysis suggests that this concern is less important for our data set, future research should employ longitudinal data and consider the dynamics between government involvement, OFDI and the acquisition of resources from overseas markets.