Outward Foreign Direct Investment and Supply Chain Concentration: Evidence from China

Abstract

The potential risks arising from excessive supply chain concentration have received more and more attention. In practice, high supply chain concentration hinders firms’ sustainable development, especially for firms in emerging markets. However, how to reduce supply chain concentration remains an underexplored problem. Outward foreign direct investment (OFDI) has been an important strategy for firms to reallocate and optimize their supply chain relationships. Thus, we tried to explore whether OFDI can serve as an effective method to reduce supply chain concentration by analyzing their relationship. Using the data of Chinese A-share listed firms from 2008 to 2022, this paper empirically examines the impact of OFDI on supply chain concentration through a two-way fixed-effects model. We find that OFDI significantly reduces supply chain concentration, and this conclusion still holds after using alternative measures of the independent variable and dependent variable, endogeneity treatment, the Heckman two-stage model test, controlling for provincial factors, and excluding special samples. Channel analysis reveals that OFDI reduces supply chain concentration mainly through increasing market share, enhancing innovation capacity, and improving reputation. Cross-sectional analysis shows that the negative effect of OFDI on supply chain concentration is more pronounced in state-owned firms, firms under high industry competition, and high-tech firms. Our findings have important implications for firms to build sustainable supply chain relationships and strengthen supply chain resilience, and for policymakers to guide OFDI reasonably. Future research could further explore the effect of firms’ investment decisions on supply chain configuration.

1. Introduction

Supply chain concentration is an important characteristic of a firm’s supply chain relationship [1]. A firm has high supply chain concentration when its sales mainly depend on a small number of customers or its purchases mainly depend on a few suppliers. A highly concentrated supply chain facilitates stable sales and purchase relationships, improves asset utilization, reduces operating costs [2], and promotes R&D cooperation between upstream and downstream firms [3]. However, high supply chain concentration also brings a series of negative effects, including weakened bargaining power [1,4] and increased vulnerability to supply chain shocks [5,6,7]. Moreover, firms have stronger incentives to engage in opportunistic behaviors, such as implementing earnings management [8] and hiding negative information [9], which cause stock price crash risks [10]. Under the current situation of high global economic uncertainty, firms’ supply chains are experiencing frequent shocks. The potential risks caused by excessive supply chain concentration become more prominent and may harm the sustainability of firms’ supply chain relationships and business operations. Therefore, how firms can reduce their supply chain concentration and enhance their supply chain resilience has become an important theoretical and practical issue.

Existing studies have investigated the determinants of supply chain concentration mainly from the perspectives of external environment and firm characteristics. Studies on the external environment show that the opening of intercity high-speed railways [11] and low customer market competition [12] increase firms’ supply chain concentration, while high economic policy uncertainty reduces firms’ supply chain concentration [13]. Studies on firm characteristics suggest that firm innovation [14], high ESG performance [15], digital transformation [16], foreign residency rights of the controlling person(s) [17], positive management tone [18], and digital technology adoption [19] reduce firms’ supply chain concentration. Nevertheless, existing studies have not paid much attention to the impact of firms’ investment decisions, as one of the important aspects of firms’ strategies, on supply chain concentration. Therefore, we aim to fill this gap and investigate the relationship between outward foreign direct investment (hereafter, OFDI) and supply chain concentration. In addition, we try to provide practical references for firms and policymakers on the countermeasures of excessive supply chain concentration.

According to the OECD, OFDI generally refers to a category of outward investment in which an investor resident in one economy establishes a lasting interest in and a significant degree of influence over an enterprise resident in another economy. As an important way to participate in international division of labor and seek high-quality business partners, a firm’s OFDI may have a significant impact on its supply chain configuration. OFDI may affect supply chain concentration through three plausible channels: The first channel is market share. Through developing international markets and expanding their customer base [20,21], OFDI firms can increase their market share and strengthen their bargaining power, which places them in a better position in cooperative relationships and helps them to configure their customers and suppliers more broadly. The second channel is innovation. Firms can obtain reverse technology spillover effects through OFDI [22,23] and enhance their innovation capacity. The improvement of firms’ innovation capacity can help create more new products, open up blue ocean markets, and increase the customer base. It also increases the complexity of firms’ demand for intermediates and pushes them to seek more innovative suppliers. The third channel is reputation. Firms can improve their reputation by acquiring internationally renowned enterprises or learning and practicing advanced management methods [24], which will strengthen the trust and recognition among potential customers and suppliers, thus attracting more upstream and downstream partners. In summary, OFDI might be an effective strategy for firms to alleviate their dependence on large customers and suppliers, and to reduce their supply chain concentration. Relevant studies on OFDI and supply chains show that OFDI promotes global value chain upgrading in the home country (region) [25], boosts the GVC position of the manufacturing industry [26], expands the spatial layout of local suppliers [27], and reduces global supply chain risks [28], but these studies have not provided a comprehensive analysis of the relationship between OFDI and supply chain concentration. Therefore, we try to supplement novel empirical evidence for the impact of OFDI on supply chain concentration from the perspectives of the customer, supplier, and overall supply chain.

In this paper, we empirically examine the relationship between OFDI and supply chain concentration using a sample of Chinese A-share listed firms from 2008 to 2022. We find that OFDI significantly reduces customer concentration, supplier concentration, and overall supply chain concentration. This conclusion still holds after using alternative measures of the dependent variable and the independent variable, endogeneity treatment, the Heckman two-stage model test, controlling for provincial factors, and excluding special samples. The potential channels through which OFDI affects supply chain concentration are increasing market share (the size effect), enhancing innovation capacity (the innovation effect), and improving firm reputation (the reputation effect). Cross-sectional analysis reveals that the impact of OFDI on supply chain concentration varies significantly according to firms’ equity nature, degree of industry competition, and technological property. The inhibitory effect of OFDI on supply chain concentration is more prominent in state-owned firms, firms under high industry competition, and high-tech firms.

This study makes the following contributions: First, we enrich the literature on the determinants of supply chain concentration. The existing literature mainly focuses on the impact of the external environment [11,12,13] and firm characteristics [14,15,16,17,18,19]. However, these studies overlook the role of firm investment decisions. This paper investigates the impact of OFDI on supply chain concentration from three perspectives: customer concentration, supplier concentration, and overall supply chain concentration. Our results provide important empirical evidence for the determinants of supply chain concentration.

Second, we extend the research perspectives on the microeconomic consequences of OFDI. Relevant studies have mainly analyzed the impact of firms’ OFDI on exports [20,21], total factor productivity [22,29], innovation [23,30], supplier spatial layout [27], stock price crash risk [31], social responsibility performance [32], and green transition [33]. However, few studies have extended the economic consequences of OFDI to the supply chain relationship aspect. Our findings provide incremental evidence for microeconomic consequences of OFDI, especially from the perspective of supply chain concentration.

Third, this paper has clear policy implications for firms’ investment decisions, sustainable supply chain development, and supply chain resilience improvement, since supply chain concentration has gained widespread attention as a potential risk factor that can hinder the sustainability of firms’ development and operations. Our results can provide important practical references for firms to make reasonable use of OFDI as a method to build sustainable supply chain relationships, enhance their competitiveness, and strengthen their supply chain resilience. Moreover, policymakers can enhance the supply chain information disclosure standards and build risk prediction systems that are conducive to the sustainability of the overall supply chain.

2. Theoretical Analysis and Hypothesis Development

OFDI is an important way for firms to connect domestic and international markets and to seek high-quality resources around the globe. The motivations of OFDI mainly include increasing export trade, absorbing advanced technology, acquiring natural resources, reducing production costs, etc. [24]. From the supply chain perspective, OFDI expands firms’ range of supply chain partner choices and enhances their competitiveness in the supply chain. This is conducive to firms’ selection of more competitive customers and suppliers, and to reducing their supply chain concentration. Specifically, the impact of OFDI on supply chain concentration is mainly reflected in the following three aspects.

2.1. OFDI and Supply Chain Concentration: The Size Effect

First, OFDI reduces supply chain concentration by increasing firms’ market share. OFDI enables firms to bypass tariff and non-tariff barriers, enter the international market with lower trade costs [25], and expand the scale of their exports [20,21]. This part of the incremental international customers and market revenue will dilute the proportion of sales to the original major customers to a certain extent, increasing firms’ total sales revenue and market share [25]. On this basis, OFDI also helps to attract more customers and increase market share through improved productivity and product quality. From the standpoint of the firm itself, the increased business scale and market share from OFDI can strengthen its bargaining power in the selection of customers and suppliers, which helps the firm to establish cooperation with more competitive customers and suppliers [34], thereby alleviating its dependence on the original major customers and suppliers. From the standpoint of customers and suppliers, cooperation with firms with a high market share in the industry is also conducive to their own revenue growth and corporate development. As a result, the increase in the firm’s market share will increase the willingness of upstream and downstream firms to cooperate with it [34]. In brief, OFDI can increase the firm’s market share, strengthen its bargaining power in the supply chain [35], and attract more upstream and downstream enterprises to cooperate with it, thus reducing its supply chain concentration.

2.2. OFDI and Supply Chain Concentration: The Innovation Effect

Second, OFDI reduces supply chain concentration by enhancing firms’ innovation capacity. OFDI firms can integrate into local innovation networks by setting up branches or R&D centers in host countries (regions) with advanced technologies and obtain reverse technology spillover in many ways [26]. Specifically, firms can enhance their innovation capacity by learning cutting-edge technologies [23,26,36], acquiring high-level talents [36], and accessing demand information of the new products [22]. Furthermore, fierce international competition also puts pressure on OFDI firms to improve their innovation capacity [23]. The improved innovation capacity will cause changes in firms’ customer and supplier relationships. From the customer’s perspective, firms with strong innovation capacity can not only use cutting-edge technology to create new products that are ahead of the market, open up blue ocean markets, and create the image of “pioneers” in the customer base, but also utilize advanced technology to improve the productivity of existing products and reduce production costs and selling prices [14,37]. Both factors will attract more customers to collaborate with firms that are more technologically innovative [38], reducing their reliance on major customers. From the supplier’s perspective, firms’ technological innovation implies a higher demand for the quality and technological sophistication of intermediate goods [39], which raises the possibility that the original major suppliers are unable to meet the firms’ requirements [35]. Under this circumstance, firms will seek more innovative suppliers with stronger responsiveness to new requirements to collaborate with [40], thus reducing their supplier concentration. In sum, the enhanced innovation capacity caused by OFDI improves firms’ attractiveness to customers and presents more sophisticated requirements to suppliers, which, in turn, helps the firms to reduce their supply chain concentration.

2.3. OFDI and Supply Chain Concentration: The Reputation Effect

Third, OFDI reduces supply chain concentration by improving firms’ reputation. According to Herbig and Milewicz [41], reputation is people’s estimation of firms’ consistency in price, quality, marketing skills, etc. Firms’ reputation, which includes ethics and social responsibility, brand image, public praise, and product and service quality, is a crucial resource for the firms to gain and maintain a competitive edge [42]. OFDI improves firms’ reputation mainly through the following two ways: Firstly, a firm can acquire internationally renowned enterprises [24] and leverage their brand and goodwill to improve its reputation [43]. Secondly, a firm can learn and apply advanced management concepts and business models through OFDI [29,32], thus strengthening its ability to fulfil its commitments to stakeholders and improving its reputation. A good reputation is a positive signal sent from the firm to its potential upstream and downstream partners, which can help the firm build trust [42] and increase the willingness of customers and suppliers to cooperate [15]. Therefore, OFDI can improve firms’ reputation, which expands their customer and supplier resources, thus reducing their supply chain concentration.

Based on the above analysis, OFDI can increase market share, enhance innovation capacity, and improve reputation, thereby reducing supply chain concentration (see Figure 1). Therefore, we propose the following hypotheses:

H1:

OFDI reduces supply chain concentration.

H2a:

OFDI reduces supply chain concentration by increasing market share.

H2b:

OFDI reduces supply chain concentration by enhancing innovation capacity.

H2c:

OFDI reduces supply chain concentration by improving firms’ reputation.

3. Research Design

3.1. Sample and Data

We chose Chinese A-share listed firms from 2008 to 2022 as the initial sample and applied the following treatment according to existing research conventions: (1) Exclude financial firms because the capital structure of financial firms is different from that of other firms, and the interpretation of financial firms’ indicators is also different; including them in regression analysis can cause bias. (2) Exclude ST (special treatment) firms and firms with abnormal trading conditions (termination of listing, suspension of listing, and trading halt), because those firms are not under normal operations and their financial data can be problematic or include extreme outliers. (3) Exclude observations missing main variables to ensure the completeness of empirical analysis. The OFDI information of the listed firms was derived from the China Stock Market and Accounting Research (CSMAR) database. According to Liu et al. (2021) [31], listed firms’ subsidiaries, joint ventures, and joint operations can be considered as OFDI affiliates if they are registered outside the Chinese mainland and the shareholding ratio of the Chinese listed firms exceeds 10%. Furthermore, we excluded the OFDI affiliates that are registered in tax havens, such as the Cayman Islands, the British Virgin Islands, and Bermuda, because such investments are primarily motivated by tax avoidance. The financial data and supply chain data of the listed firms were also sourced from the CSMAR Database, and the patent data were sourced from the Chinese Research Data Services (CNRDS) platform. Finally, we obtained a sample of 31,718 firm-year observations. To reduce the influence of outliers, we winsorized continuous variables at the 1st and 99th percentiles.

3.2. Variable Definitions

3.2.1. The Dependent Variable: Supply Chain Concentration

A firm’s supply chain concentration includes three aspects: customer-base concentration, supplier-base concentration, and overall supply chain concentration. Based on the research by Zhou et al. (2024) [19], we constructed three indicators, including customer concentration CusCon, supplier concentration SupCon, and overall supply chain concentration SCCon, to portray the firm’s supply chain concentration. The larger the value of CusCon/SupCon/SCCon, the higher the customer concentration/supplier concentration/overall supply chain concentration the firm has.

3.2.2. The Independent Variable: OFDI

This paper considers the OFDI at the firm level. Referring to the work of Zhang and Sun (2023) [33], we use the firm’s number of OFDI affiliates to measure its OFDI level. Considering that the firm’s number of OFDI affiliates is right-skewed, we treat it by adding 1 and taking the natural logarithm.

3.2.3. Control Variables

Referring to the studies of Qian et al. (2023) [17] and Zhou et al. (2024) [19], we adopted the following firm-level control variables: firm size, age, leverage, profitability, growth, relative value, operating cash flow, net working capital, and equity nature. We controlled for these variables because firms’ operating and financial conditions are likely to influence their supply chain management decisions and negotiation power in supply chain relationships. For example, larger firms usually need more diverse sales channels and intermediate goods and have stronger industry positions, thus having lower supply chain concentrations. The definitions of the main variables are shown in

Table 1. Variable definitions.

Symbol	Name	Definition
CusCon	Customer concentration	The firm’s sales to its top five customers/total annual sales
SupCon	Supplier concentration	The firm’s purchases from its top five suppliers/total annual purchases
SCCon	Overall supply chain concentration	(Supplier concentration + customer concentration)/2
OFDI	Outward foreign direct investment (OFDI)	Natural logarithm of 1 plus the number of the firm’s OFDI affiliates
Size	Firm size	Natural logarithm of the firm’s total assets
Age	Age	Natural logarithm of 1 plus the number of years that the firm has been established
Lev	Leverage	Total liability/total assets
Roa	Profitability	Net income/total assets
Revgrow	Growth	(Current year revenue − previous year revenue)/previous year revenue
Tbq	Relative value	Market value/total assets
Opcash	Operating cash flow	Net cash flow from operating activities/total assets
Nwc	Net working capital	(Liquid assets − liquid liability − cash)/total assets
Soe	Equity nature	A dummy variable that equals 1 if the firm is state-owned and 0 otherwise

.

3.3. Model Setting

To test the impact of OFDI on supply chain concentration, we built the following two-way fixed-effects model:

$$Concentratio{n}_{it}={\alpha }_0+{\alpha }_{1}OFD{I}_{it}+{\alpha }_2{X}_{it}+{\delta }_j+{\mu }_t+{\epsilon }_{it}$$

(1)

where the dependent variable Concentration_it is the supply chain concentration, including customer concentration CusCon_it, supplier concentration SupCon_it, and overall supply chain concentration SCCon_it. The independent variable OFDI_it is the firm’s outward foreign direct investment. X_it represents a series of firm-level control variables. δ_j and μ_t represent industry-level and year-level fixed effects, respectively, which are used to control for industry-level factors that do not change over time and year-level factors that do not change over individuals. ε_it is the random error term.

4. Empirical Analysis

4.1. Summary Statistics

Table 2. Summary statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
CusCon	31,718	0.3279	0.2279	0.0131	0.9702
SupCon	31,718	0.3508	0.1987	0.0532	0.9320
SCCon	31,718	0.3394	0.1702	0.0331	0.9511
OFDI	31,718	0.6482	0.8243	0.0000	5.3753
Size	31,718	22.1483	1.2800	19.8429	26.1201
Age	31,718	2.9205	0.3431	0.6931	4.1744
Lev	31,718	0.4076	0.2042	0.0528	0.9038
Roa	31,718	0.0385	0.0650	−0.2662	0.2041
Revgrow	31,718	0.1665	0.3804	−0.5303	2.2872
Tbq	31,718	2.0291	1.2979	0.8476	8.5342
Opcash	31,718	0.0492	0.0686	−0.1563	0.2479
Nwc	31,718	0.0659	0.2051	−0.4788	0.5752
Soe	31,718	0.3449	0.4753	0.0000	1.0000

reports the summary statistics of the main variables. The mean value of customer concentration is 32.79%, with a standard deviation of 22.79% and a maximum value of 97.02%. The mean value of supplier concentration is 35.08%, with a standard deviation of 19.87% and a maximum value of 93.2%. The mean value of overall supply chain concentration is 33.94%, with a standard deviation of 17.02% and a maximum value of 95.11%. According to the study of Leung and Sun (2021) [13], the average proportion of firms’ total sales to major customers (sales ratio ≥ 10%) among listed US firms is 6.9%. These results suggest that the supply chain concentration of Chinese listed firms is relatively high. Some of the sample firms have over 90% customer concentration or supplier concentration, which means that they are strongly dependent on their major customers and suppliers. Although a concentrated supply chain can facilitate stable sales relationships and reduce operating costs, excessively high supply chain concentration can trigger a series of significant negative effects, such as lower bargaining power, increased vulnerability to supply chain disruptions, and more frequent opportunistic behaviors, which may eventually lead to stock price crashes. Therefore, it is of great practical significance to explore feasible paths to reduce supply chain concentration for Chinese firms. The mean value of OFDI is 0.6482 and the standard deviation is 0.8243, which indicates that there is a prominent difference in the OFDI levels between Chinese listed firms. The descriptive statistics of other variables are similar to those in previous studies, indicating that the measurement of the main variables is reasonable.

4.2. Baseline Results

Table 3. Baseline results.

Variables	(1)	(2)	(3)
	CusCon	SupCon	SCCon
OFDI	−0.0073 **	−0.0183 ***	−0.0128 ***
	(0.0033)	(0.0028)	(0.0024)
Size	−0.0381 ***	−0.0339 ***	−0.0360 ***
	(0.0031)	(0.0025)	(0.0022)
Age	−0.0403 ***	−0.0100	−0.0252 ***
	(0.0092)	(0.0079)	(0.0067)
Lev	−0.0099	−0.1023 ***	−0.0561 ***
	(0.0199)	(0.0183)	(0.0149)
Roa	−0.0882 ***	−0.0735 **	−0.0809 ***
	(0.0332)	(0.0287)	(0.0250)
Revgrow	0.0462 ***	0.0277 ***	0.0370 ***
	(0.0039)	(0.0034)	(0.0029)
Tbq	0.0065 ***	0.0111 ***	0.0088 ***
	(0.0021)	(0.0019)	(0.0016)
Opcash	−0.1599 ***	−0.0961 ***	−0.1280 ***
	(0.0269)	(0.0232)	(0.0200)
Nwc	0.0313 *	−0.0273 *	0.0020
	(0.0173)	(0.0160)	(0.0130)
Soe	0.0171 **	−0.0031	0.0070
	(0.0069)	(0.0062)	(0.0052)
Controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes
Observations	31,718	31,718	31,718
Adjusted R-squared	0.2623	0.2084	0.2692

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

reports the baseline regression results. Columns (1)–(3) test the impact of OFDI on customer concentration, supplier concentration, and overall supply chain concentration, respectively. In column (1), the coefficient estimate of OFDI is −0.0073 and significant at the 5% level, which implies that OFDI significantly reduces customer concentration. In column (2), the coefficient estimate of OFDI is −0.0183 and significant at the 1% level, which implies that OFDI significantly reduces supplier concentration. In column (3), the coefficient estimate of OFDI is −0.0128 and significant at the 1% level, which implies that OFDI significantly reduces overall supply chain concentration.

In terms of economic significance, an increase of one standard deviation in OFDI causes CusCon to decrease by 0.60% (0.8243 × 0.0073), SupCon to decrease by 1.51% (0.8243 × 0.0183), and SCCon to decrease by 1.06% (0.8243 × 0.0128). Considering that the average total annual sales and average total annual purchases in our sample were approximately 9413.9 million RMB and 11190.8 million RMB, respectively, the 0.60% decrease in CusCon is equivalent to a 56.5 million (9413.9 × 0.60%) RMB reduction in sales to the top five customers, and the 1.51% decrease in SupCon is equivalent to a 169 million (11190.8 × 1.51%) RMB reduction in purchases from the top five suppliers. Overall, the baseline results suggest that OFDI reduces supply chain concentration, which validates H1.

4.3. Robustness Test

4.3.1. Alternative Measures of the Dependent Variable

Considering that the measurement error may lead to bias in the regression results, we used alternative measures of the dependent variable to check the robustness. Specifically, we used CusHHI and TopCus to replace the customer concentration in the benchmark regression, where CusHHI is the Herfindahl index of customer concentration (the sum of the squared ratios of the firm’s sales to the top five customers to the total sales) and TopCus is the ratio of the sales to the top one customer to total sales. In addition, we used SupHHI and TopSup to replace the supplier concentration in the benchmark regression, where SupHHI is the Herfindahl index of supplier concentration (the sum of the squared ratios of the firm’s purchases from the top five suppliers to the total purchases) and TopSup is the ratio of the purchases from the top one supplier to total purchases. TopCus and TopSup have similar meanings to CusCon and SupCon, as they consider the top one customer or the top one supplier instead of the top five. CusHHI and SupHHI further consider the number of customers or suppliers and their sale or purchase distributions. The larger their values, the higher the customer concentration or supplier concentration the firm has. Therefore, we expect negative coefficients of OFDI when using these alternative measures.

Table 4. Alternative measures of the dependent variable.

Variables	(1)	(2)	(3)	(4)
	CusHHI	TopCus	SupHHI	TopSup
OFDI	−0.0068 ***	−0.0074 ***	−0.0055 ***	−0.0085 ***
	(0.0017)	(0.0026)	(0.0014)	(0.0023)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Observations	22,682	22,687	22,334	22,335
Adjusted R-squared	0.2131	0.2097	0.1320	0.1364

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** denote significance at the 1% level.

reports the results using alternative measures of the dependent variable. In columns (1)–(4), the coefficient estimates of OFDI are all significantly negative, consistent with the baseline results.

4.3.2. Alternative Measures of the Independent Variable

To further confirm the robustness of the baseline results, we also used alternative measures of the independent variable. Specifically, we used OFDIDUM, the dummy variable of whether the firm conducts OFDI (equal to 1 if the firm conducts OFDI in the given year, and 0 otherwise), and OFDIW, the breadth of the firm’s OFDI (the natural logarithm of 1 plus the number of countries (regions) in which the firm conducts OFDI), to replace OFDI in the baseline regression. Since we believe that firms can reduce their supply chain concentration by conducting OFDI, we expect OFDI firms to have lower supply chain concentration than non-OFDI firms, and we expect firms that conduct OFDI in a wider range of countries (regions) to have lower supply chain concentration. In other words, we expect the coefficients of OFDIDUM and OFDIW to be negative.

Table 5. Alternative measures of the independent variable.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	CusCon	SupCon	SCCon	CusCon	SupCon	SCCon
OFDIDUM	−0.0097 *	−0.0240 ***	−0.0169 ***
	(0.0050)	(0.0045)	(0.0038)
OFDIW				−0.0127 ***	−0.0245 ***	−0.0186 ***
				(0.0041)	(0.0034)	(0.0030)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	31,718	31,718	31,718	31,718	31,718	31,718
Adjusted R-squared	0.2621	0.2072	0.2685	0.2627	0.2091	0.2702

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and * denote significance at the 1% and 10% levels, respectively.

reports the results using alternative measures of the independent variable. In columns (1)–(3), the coefficient estimates of OFDIDUM are significantly negative. In columns (4)–(6), the coefficient estimates of OFDIW are also significantly negative. The above results show that our main conclusions still hold after using alternative measures of the independent variable.

4.3.3. Endogeneity Treatment

The reverse causality between OFDI and supply chain concentration could be a potential source of endogeneity in our study. Specifically, firms with lower supply chain concentration might be more willing to conduct OFDI. In order to mitigate this concern, we used a one-period lagged independent variable and instrumental variable approach. First, since firms’ supply chain concentration in the current year does not affect OFDI by firms in the previous year, using one-period lagged OFDI (L.OFDI) can mitigate the reverse causality issue to a certain extent. The results in columns (1)–(3) of

Table 6. Endogeneity treatment.

	(1)	(2)	(3)	(4)	(5)	(6)
	One-Period Lagged OFDI			Instrumental Variable Approach
Variables	CusCon	SupCon	SCCon	CusCon	SupCon	SCCon
L.OFDI	−0.0065 *	−0.0168 ***	−0.0116 ***
	(0.0034)	(0.0030)	(0.0025)
OFDI				−0.0088 **	−0.0197 ***	−0.0142 ***
				(0.0042)	(0.0035)	(0.0028)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Kleibergen–Paap rk LM statistic				188.82 (p-value = 0.0000)
Kleibergen–Paap Wald rk F statistic				168.23 (Critical value = 16.38)
Observations	25,436	25,436	25,436	31,590	31,590	31,590
Adjusted R-squared	0.2686	0.2133	0.2720	0.0725	0.1019	0.1310

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

show that the coefficient estimates of L.OFDI are all significantly negative, consistent with the baseline results. Using a one-period lagged independent variable still has certain limitations because it will cause the sample size to decrease and it cannot totally address endogeneity under certain circumstances.

Second, we drew on the method of Lewbel (1997) [44] and Zhang et al. (2024) [45] and constructed the third power of the difference between the firm’s OFDI and the mean OFDI of other firms in the industry in the given year (Lewbel IV) as our instrumental variable. The reason we used Lewbel IV is that it can address endogeneity concerns when external instruments are absent or weak (Lewbel, 1997) [44]. Since supply chain concentration has a series of important economic consequences and might be affected by various external and internal factors, it is very difficult to find a suitable instrument. In addition, the Lewbel IV approach keeps most of our sample observations. The main limitation of Lewbel IV is that it is relatively sensitive to outliers, which might not be a significant concern because our sample does not include extreme outliers. Using Lewbel IV, we performed the two-stage least squares (2SLS) estimation. The results in columns (4)–(6) show that the coefficient estimates of OFDI are still significantly negative using the instrumental variable approach, consistent with the baseline results. The Kleibergen–Paap rk LM statistic is statistically significant, and the Kleibergen–Paap Wald rk F statistic is greater than the critical value, which indicates that there is no significant under-identification problem or weak instrumental variable problem, respectively. The above results indicate that endogeneity concerns do not significantly affect our conclusion.

4.3.4. Heckman Two-Stage Model Test

Our study is possibly subject to sample selection bias because firms’ OFDI is not a random event, and certain types of firms can be more likely to perform OFDI due to some common characteristics. Those characteristics may also have an impact on supply chain concentration and bias our results. To address this issue, we employed a Heckman two-stage model. In the first stage, we performed a probit regression, with the dummy variable of whether the firm conducts OFDI as the dependent variable and all control variables in the baseline regression as independent variables, because these control variables represent firms’ operating and financial performance and are potentially the main determinants of firms’ OFDI decisions. The first-stage results are reported in column (1) of

Table 7. Heckman two-stage model test.

Variables	(1)	(2)	(3)	(4)
	OFDIDUM	CusCon	SupCon	SCCon
OFDI		−0.0079 **	−0.0201 ***	−0.0140 ***
		(0.0033)	(0.0028)	(0.0024)
IMR		0.0464	0.1465 ***	0.0965 ***
		(0.0456)	(0.0316)	(0.0312)
Size	0.4393 ***
	(0.0184)
Age	−0.0877
	(0.0571)
Lev	0.1377
	(0.1271)
Roa	−0.8234 ***
	(0.2097)
Revgrow	−0.0317
	(0.0232)
Tbq	0.0216 *
	(0.0117)
Opcash	0.8197 ***
	(0.1675)
Nwc	0.0140
	(0.1127)
Soe	−0.5058 ***
	(0.0472)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Observations	31,505	31,505	31,505	31,505
(Pseudo) adjusted R-squared	0.1438	0.2620	0.2085	0.2693

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Column (1) reports the pseudo R-squared because it uses a probit regression. Columns (2)–(4) report the adjusted R-squared.

. In the second stage, we incorporated the inverse Mills ratio IMR obtained in the first stage into model (1) to correct for the sample selection bias and re-estimate the effect of OFDI on supply chain concentration. The results in columns (2)–(4) show that the coefficient estimates of OFDI were all significantly negative after incorporating IMR and the scale of the coefficients was quite close to that of those in the baseline results. The above results show that our main conclusions still hold after considering sample selection bias.

4.3.5. Controlling for Provincial Factors

Although we included a series of firm-level control variables and industry and year fixed effects in the baseline regression, our model is still possibly subject to omitted-variable bias because some regional-level factors are also likely to have an impact on firms’ supply chain concentration. For example, firms located in regions with higher economic development levels may have lower supply chain concentration because they possess stronger business resources and have stronger bargaining power. Therefore, we included three provincial-level control variables, including GDP per capita (Gdppc), the proportion of the secondary sector to GDP (Secondratio), and the proportion of the tertiary sector to GDP (Thirdratio). To further control for other unobservable provincial factors that do not change over time, we also included the provincial-level fixed effects. The results in

Table 8. Controlling for provincial factors.

Variables	(1)	(2)	(3)
	CusCon	SupCon	SCCon
OFDI	−0.0058 *	−0.0183 ***	−0.0121 ***
	(0.0033)	(0.0028)	(0.0024)
Gdppc	−0.0071	0.0169	0.0049
	(0.0227)	(0.0212)	(0.0172)
Secondratio	0.6291 ***	0.0069	0.3180 *
	(0.2118)	(0.2140)	(0.1656)
Thirdratio	0.4589 **	−0.1630	0.1480
	(0.2041)	(0.2060)	(0.1608)
Controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes
Province FE	Yes	Yes	Yes
Observations	31,702	31,702	31,702
Adjusted R-squared	0.2707	0.2123	0.2772

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

show that the proportions of the secondary and tertiary sectors have significant positive effects on customer concentration. After including provincial controls and fixed effects, the coefficient estimates of OFDI remain significantly negative in columns (1)–(3) and have similar values to those in the baseline results. Therefore, our results are still valid after considering provincial factors.

4.3.6. Excluding Special Samples

The municipalities in China (Beijing, Shanghai, Tianjin, and Chongqing) have better business environments and receive more policy support. As a result, firms in municipalities may have stronger incentives to perform OFDI. Moreover, those firms usually have higher information disclosure standards and, thus, are less likely to have excessive supply chain concentration. Therefore, we excluded the firms that are registered in Beijing, Shanghai, Tianjin, and Chongqing, and we re-estimated the effect of OFDI on supply chain concentration. Columns (1)–(3) in

Table 9. Excluding special samples.

	(1)	(2)	(3)	(4)	(5)	(6)
	Excluding Firms in Municipalities			Excluding Observations from 2008–2010
Variables	CusCon	SupCon	SCCon	CusCon	SupCon	SCCon
OFDI	−0.0074 **	−0.0230 ***	−0.0152 ***	−0.0071 **	−0.0178 ***	−0.0125 ***
	(0.0037)	(0.0031)	(0.0027)	(0.0033)	(0.0028)	(0.0024)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	25,714	25,714	25,714	30,498	30,498	30,498
Adjusted R-squared	0.2670	0.2120	0.2747	0.2638	0.2119	0.2707

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and ** denote significance at the 1% and 5% levels, respectively.

show that the coefficient estimates of OFDI were significantly negative after excluding firms in municipalities, which is consistent with the baseline results. In addition, to rule out the impact of the global financial crisis, which may affect firms’ investment decisions and supply chain management, we excluded observations from 2008–2010 and re-estimated the relationship. Columns (4)–(6) show that the coefficient estimates of OFDI were significantly negative after excluding observations from 2008–2010, which further confirms the robustness of our results.

5. Channel Analysis

Based on the previous theoretical analysis, we propose three potential channels through which OFDI reduces supply chain concentration, namely, the size effect, the innovation effect, and the reputation effect. In order to validate hypotheses H2a–H2c, we built the following mediation effect model on the basis of model (1):

$$Me{d}_{it}={\beta }_0+{\beta }_{1}OFD{I}_{it}+{\beta }_2{X}_{it}+{\delta }_j+{\mu }_t+{\epsilon }_{it}$$

(2)

$$Concentratio{n}_{it}={\gamma }_0+{\gamma }_{1}OFDI+{\gamma }_{2}Me{d}_{it}+{\gamma }_3{X}_{it}+{\delta }_j+{\mu }_t+{\epsilon }_{it}$$

(3)

where Med is the mediating variable, which includes relative market share RMS, firm innovation Pat, and firm reputation Rep. The specifications of the other variables are consistent with model (1).

5.1. The Size Effect

As described by Li et al. (2021) [14], we constructed the relative market share RMS, which is the ratio of the firm’s revenue to the revenue of the largest firm in the industry, to measure the firm’s market size. The higher the relative market share, the larger the market size and the stronger the bargaining power the firm has in the industry.

Table 10. Channel analysis: increasing market share.

Variables	(1)	(2)	(3)	(4)
	RMS	CusCon	SupCon	SCCon
OFDI	0.0092 ***	−0.0069 **	−0.0181 ***	−0.0125 ***
	(0.0032)	(0.0033)	(0.0028)	(0.0024)
RMS		−0.0461 ***	−0.0163	−0.0312 ***
		(0.0136)	(0.0132)	(0.0104)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Observations	31,718	31,718	31,718	31,718
Adjusted R-squared	0.5263	0.2632	0.2085	0.2700

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and ** denote significance at the 1% and 5% levels, respectively.

reports the test results of the size effect. In column (1), the coefficient estimate of OFDI is significantly positive, indicating that OFDI increases the firm’s relative market share. In column (2), the coefficient estimate of RMS is significantly negative, indicating that the increase in relative market share significantly reduces customer concentration. In column (3), the coefficient estimate of RMS is also negative but not significant, indicating that the increase in relative market share does not significantly reduce the supplier concentration. In column (4), the coefficient estimate of RMS is once again significantly negative, suggesting that the increase in relative market share significantly reduces the firm’s overall supply chain concentration. The above results indicate that OFDI broadens the firm’s sales channels and leads to an increase in its market share, thus reducing the customer concentration. In comparison, the effect of OFDI-induced market share increases on reducing the supplier concentration is relatively limited. In general, the increased market share reduces the overall supply chain concentration. The above results show that OFDI can reduce supply chain concentration through increasing market share. Hence, H2a is verified.

5.2. The Innovation Effect

As described by Zhang et al. (2024) [45], we used the number of patent applications Pat, which is the natural logarithm of 1 plus the firm’s total number of patent applications, as the measure of firms’ innovation capacity. The higher the number of patent applications of a firm, the higher the innovation capacity it owns.

Table 11. Channel analysis: enhancing innovation capacity.

Variables	(1)	(2)	(3)	(4)
	Pat	CusCon	SupCon	SCCon
OFDI	0.1267 ***	−0.0058 *	−0.0152 ***	−0.0105 ***
	(0.0212)	(0.0033)	(0.0028)	(0.0024)
Pat		−0.0124 ***	−0.0236 ***	−0.0180 ***
		(0.0019)	(0.0017)	(0.0014)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Observations	31,680	31,680	31,680	31,680
Adjusted R-squared	0.5127	0.2664	0.2289	0.2852

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and * denote significance at the 1% and 10% levels, respectively.

reports the test results of the innovation effect. In column (1), the coefficient estimate of OFDI is significantly positive, indicating that OFDI increases the firm’s innovation capacity. In columns (2)–(4), the coefficient estimates of Pat are all significantly negative, indicating that the increase in the firm’s innovation capacity significantly reduces its customer concentration and supplier concentration, and therefore also reduces the overall supply chain concentration. This result is consistent with the previous theoretical analysis. OFDI enhances firms’ innovation capacity through the reverse technology spillover effect and other potential means, helping the firm attract more diversified customers and suppliers for cooperation, and thus reducing the supply chain concentration. In other words, OFDI can reduce supply chain concentration through enhancing firms’ innovation capacity. Hence, H2b is verified.

5.3. The Reputation Effect

As described by Li et al. (2023) [15], we used the natural logarithm of firms’ intangible assets Rep as the measure of their reputation. Firms’ intangible assets generally include trademarks, goodwill, copyrights, patent rights, leasing rights, franchises, etc. The more intangible assets a firm has, the better its reputation.

Table 12. Channel analysis: improving reputation.

Variables	(1)	(2)	(3)	(4)
	Rep	CusCon	SupCon	SCCon
OFDI	0.0752 ***	−0.0060 *	−0.0168 ***	−0.0114 ***
	(0.0169)	(0.0033)	(0.0028)	(0.0024)
Rep		−0.0174 ***	−0.0153 ***	−0.0163 ***
		(0.0024)	(0.0022)	(0.0018)
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes
Observations	31,444	31,444	31,444	31,444
Adjusted R-squared	0.6523	0.2663	0.2108	0.2756

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and * denote significance at the 1% and 10% levels, respectively.

reports the test results of the reputation effect. In column (1), the coefficient estimate of OFDI is significantly positive, which indicates that OFDI improves the firm’s reputation. In columns (2)–(4), the coefficient estimates of Rep are all significantly negative, indicating that the improvement of a firm’s reputation significantly reduces its customer concentration and supplier concentration, and therefore also reduces the overall supply chain concentration. In sum, OFDI can improve firms’ reputation, helping to gain more customers and suppliers, and reducing the supply chain concentration. That is, OFDI can reduce supply chain concentration through improving firms’ reputation. Hence, H2c is verified.

6. Cross-Sectional Analysis

6.1. Equity Nature

Although OFDI can play a positive role in market development, technology upgrades, cost savings, etc., it is also important to recognize its risk attributes, such as the long period of completion and high level of uncertainty. The risks involved in the OFDI projects may weaken the willingness of firms’ potential upstream suppliers and downstream customers to cooperate, because they will worry about the unanticipated losses in OFDI firms’ investment projects, which may impact or even terminate the cooperation and result in supply chain disruptions. In this regard, the large scale and strong risk resistance ability of state-owned enterprises (SOEs) can serve as an invisible guarantee. In addition, SOEs play the role of “forerunner” in OFDI, effectively leading private firms to enter the international market [46], and possibly generating stronger size effects and innovation effects. These two factors will make customers and suppliers more inclined to cooperate with OFDI SOEs. Therefore, we expect that OFDI has a more prominent effect in reducing supply chain concentration for SOEs.

Table 13. Cross-sectional analysis: equity nature.

	(1)	(2)	(3)	(4)	(5)	(6)
	SOEs	Non-SOEs	SOEs	Non-SOEs	SOEs	Non-SOEs
Variables	CusCon	CusCon	SupCon	SupCon	SCCon	SCCon
OFDI	−0.0208 ***	0.0007	−0.0173 ***	−0.0185 ***	−0.0191 ***	−0.0089 ***
	(0.0059)	(0.0039)	(0.0048)	(0.0034)	(0.0043)	(0.0029)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	10,939	20,778	10,939	20,778	10,939	20,778
Adjusted R-squared	0.3297	0.2533	0.2917	0.1798	0.3497	0.2484
Between-group difference	p-value = 0.000		p-value = 0.130		p-value = 0.000

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** denote significance at the 1% level.

reports the results of cross-sectional analysis based on firms’ equity nature. Columns (1)–(2) show that OFDI significantly reduces customer concentration for SOEs, while OFDI does not have a significant effect on customer concentration for non-SOEs. Columns (3)–(4) show that OFDI significantly reduces supplier concentration for both SOEs and non-SOEs. The absolute values of their coefficient estimates are relatively close, and the between-group difference is not statistically significant. Columns (5)–(6) show that OFDI significantly reduces the overall supply chain concentration for both SOEs and non-SOEs. In addition, the absolute value of the coefficient estimate is larger in SOEs, and the between-group difference is statistically significant. The above results indicate that OFDI has a more prominent effect in reducing supply chain concentration for SOEs, which is consistent with our expectation, and this difference is mainly reflected in customer concentration.

6.2. Industry Competition

When firms are facing strong industry competition, they usually have lower profit margins and greater survival pressures. Therefore, they have stronger motivations to improve product quality, reduce product costs, or implement differentiated strategies by OFDI. These firms are more likely to alter their distribution channels and production modes and, thus, are more likely to seek new customers and suppliers. In addition, strong market competition leads to lower switching costs for customers and suppliers [6], and it increases the probability of firms entering into a new cooperative relationship. Therefore, we expect that OFDI has a more prominent effect in reducing supply chain concentration for firms under high industry competition.

We split the sample into firms under high industry competition (industry Lerner index less than the sample median for the given year) and firms under low industry competition (industry Lerner index greater than or equal to the sample median for the given year) based on the median of the industry Lerner index.

Table 14. Cross-sectional analysis: industry competition.

	(1)	(2)	(3)	(4)	(5)	(6)
	High Competition	Low Competition	High Competition	Low Competition	High Competition	Low Competition
Variables	CusCon	CusCon	SupCon	SupCon	SCCon	SCCon
OFDI	−0.0125 ***	−0.0013	−0.0241 ***	−0.0106 ***	−0.0183 ***	−0.0059 **
	(0.0043)	(0.0041)	(0.0036)	(0.0037)	(0.0032)	(0.0030)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	16,370	15,342	16,370	15,342	16,370	15,342
Adjusted R-squared	0.2713	0.2525	0.2163	0.2064	0.2711	0.2694
Between-group difference	p-value = 0.000		p-value = 0.000		p-value = 0.000

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and ** denote significance at the 1% and 5% levels, respectively.

reports the result of cross-sectional analysis based on industry competition. The results in columns (1)–(2) show that OFDI significantly reduces customer concentration for firms under high industry competition, while the effect is not significant for firms under low industry competition. The results in columns (3)–(4) show that OFDI significantly reduces supplier concentration for firms under high industry competition and firms under low industry competition. In addition, the absolute value of the coefficient estimate is larger for firms under high industry competition, and the between-group difference is statistically significant. The results in columns (5)–(6) show that OFDI significantly reduces the overall supply chain concentration for firms under high industry competition and firms under low industry competition. Furthermore, the absolute value of the coefficient estimate is larger for firms under high industry competition, and the between-group difference is statistically significant. The above results indicate that OFDI has a more prominent effect in reducing supply chain concentration for firms under high industry competition, which is consistent with our expectation, and this difference is reflected in both customer concentration and supplier concentration.

6.3. Technological Property

In the current competitive environment, product novelty and technology levels play an increasingly important role in determining product competitiveness. High-tech firms are more capable of developing new products with higher technological content, so their products are more likely to be accepted by international customers, and their possibilities for expanding their customer base through OFDI are higher. In addition, the strong R&D capability of high-tech firms makes them more capable of obtaining reverse technological spillovers from OFDI [23] and more likely to change their modes of production. Consequently, high-tech firms will have stronger incentives to seek new customers and suppliers. Based on these two factors, we expect that OFDI has a more prominent effect in reducing supply chain concentration for high-tech firms.

We split the sample into high-tech firms and non-high-tech firms based on the technological property of the firms’ industry [47].

Table 15. Cross-sectional analysis: technological property.

	(1)	(2)	(3)	(4)	(5)	(6)
	High-Tech 1	Non-High-Tech	High-Tech	Non-High-Tech	High-Tech	Non-High-Tech
Variables	CusCon	CusCon	SupCon	SupCon	SCCon	SCCon
OFDI	−0.0148 ***	0.0028	−0.0191 ***	−0.0185 ***	−0.0169 ***	−0.0078 **
	(0.0044)	(0.0049)	(0.0036)	(0.0045)	(0.0032)	(0.0037)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	19,896	11,822	19,896	11,822	19,896	11,822
Adjusted R-squared	0.1868	0.3513	0.1739	0.2583	0.2012	0.3531
Between-group difference	p-value = 0.000		p-value = 0.404		p-value = 0.000

Notes: Robust standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *** and ** denote significance at the 1% and 5% levels, respectively. ¹ Based on the industry classification standards by the China Securities Regulatory Commission (2012), firms with industry classification codes of C25 to C29, C31 to C32, C34 to C41, I63 to I65, and M73 are identified as high-tech firms.

reports the results of cross-sectional analysis based on technological property. The results in columns (1)–(2) show that OFDI significantly reduces customer concentration for high-tech firms, while the effect is not significant for non-high-tech firms. The results in columns (3)–(4) show that OFDI significantly reduces supplier concentration for both high-tech firms and non-high-tech firms. The absolute values of their coefficient estimates are relatively close, and the between-group difference is not statistically significant. The results in columns (5)–(6) show that OFDI significantly reduces the overall supply chain concentration for both high-tech firms and non-high-tech firms. Moreover, the absolute value of the coefficient estimate is larger for high-tech firms, and the between-group difference is statistically significant. The above results indicate that OFDI has a more prominent effect in reducing supply chain concentration for high-tech firms, which is consistent with our expectation, and this difference is mainly reflected in customer concentration.

7. Conclusions

Currently, Chinese firms have a strong reliance on major customers and suppliers. Excessive supply chain concentration can lead to a series of potential risks. As a strategy to seek high-quality business partners and improve competitiveness, firms’ OFDI is closely related to their supply chain configuration. However, the impact of OFDI on supply chain concentration has not received much attention. This paper aims to fill this gap. Using the data of Chinese A-share listed firms from 2008 to 2022, we examined the impact of OFDI on supply chain concentration through a two-way fixed-effects model. We found that OFDI significantly reduces customer concentration, supplier concentration, and overall supply chain concentration, and this conclusion still holds after a series of robustness tests. We inspected the channels through the mediation effect model and found that the size effect, the innovation effect, and the reputation effect are three plausible channels through which OFDI reduces supply chain concentration. We performed cross-sectional analysis through subsample regressions and found that OFDI has a more prominent effect in reducing supply chain concentration for state-owned firms, firms under high industry competition, and high-tech firms.

This paper finds that OFDI is an effective way for firms to reduce supply chain concentration, which has important policy implications for guiding firms’ OFDI behavior, strengthening the sustainability of firms’ supply chains, and enhancing supply chain resilience, especially for emerging markets. Firstly, the government should further improve the disclosure standards of firms’ supply chain information, strengthen the supervision of firms with excessively high supply chain concentration, and prevent the risk of supply chain disruption. For example, the government can promote the establishment of digital supply chain management systems and supply chain risk prediction systems, thus strengthening the opaqueness and sustainability of the supply chain. Secondly, firms should make reasonable use of OFDI as a corporate strategy to utilize the advantageous resources in both domestic and international markets and enhance their international competitiveness and bargaining power. In this way, they can be better positioned to attract more competent upstream and downstream partners and build high-quality supply chain relationships. In addition, firms can also learn the cutting-edge supply chain management methods through OFDI and further promote their sustainable supply chain development. Thirdly, the government can provide targeted incentive policies and support services to help firms improve their OFDI performance and reduce their OFDI risks. Many small firms in emerging markets possess comparative advantage but lack the capital and experience to carry out OFDI. To enable those firms to optimize their supply chain configuration, the government can provide targeted policies or leverage the leading role of SOEs.

This paper extends the research on the determinants of supply chain concentration and the microeconomic consequences of OFDI. In general, we enrich the understanding of the relationship between firms’ investment decisions and supply chain relationships. However, this study has certain limitations. We only considered the impact of OFDI on supply chain concentration, but this is only one aspect of the overall supply chain relationship and configuration. There are many other aspects, such as supply chain quality, safety, and resilience, which are also important for firms’ sustainable development. We recommend that future research should consider those dimensions.