Antecedents and Context of Chinese Firms’ Foreign Exit

Abstract

This paper examines the outward foreign direct investment events of Chinese manufacturing firms from 2008 to 2022. It explores how sunk cost and performance feedback drive firms’ foreign exit strategies based on the behavioral theory of the firm. Additionally, it also examines the contextual factor that may affect the focal relationship. We adopt a panel logistic estimation to test the hypotheses. The conclusions show that firms are more likely to exit overseas markets when faced with sunk costs and negative performance feedback. Conversely, positive performance feedback significantly decreases the probability of firms exiting foreign markets. Additionally, environmental munificence and complexity can weaken the strength of the focal nexus to some extent. These findings hold both theoretical and practical significance for multinational enterprises and the government in the context of the ‘dual-circulation strategy’.

1. Introduction

Since the 1970s, the global landscape has changed dramatically. The trend of economic globalization prevailed as countries began to liberalize trade, breaking down trade barriers by signing free trade agreements. Against this background, companies from various countries have entered overseas markets and embarked on the path of internationalization in order to gain unique advantages in global competition. However, as international markets become increasingly saturated, the dividends brought to companies are no longer fruitful. Instead, they are replaced by huge risks and challenges. Since the financial crisis of 2008, anti-globalization and trade protectionism have been on the rise, and the global economic and political landscape has undergone profound changes unseen in a century. Events such as Brexit, the US’s withdrawal from the Trans-Pacific Partnership, the renegotiation of the North American Free Trade Agreement (NAFTA), the Sino-US trade war, and the outbreak of the Russian–Ukrainian war are becoming more frequent. This turbulence in the external environment poses a major threat to the survival and persistence of multinational enterprises, with some small MNEs being forced to withdraw from foreign operations and large ones reorganizing their international plans by merging or closing under-performing subsidiaries. Therefore, scholars have begun to focus on the phenomenon of foreign exits because the occurrence of repeated internationalization is more in line with the real situation of firms, and the likelihood is greater [1,2].

Since the reform and opening up, China has relied on its labor endowment advantage to achieve rapid economic development. Furthermore, its comprehensive national power and international status have persistently risen to a higher level. The United States and China, as two strong strategic competitors, have been involved in the continuous intensification of the Great Game and increased instability and uncertainty in Sino–US relations [3]. In addition, the global outbreak of the coronavirus pandemic has intensified the pressure to reconstruct the global supply, industrial, and value chain [4]. In light of these circumstances, it is imperative that China’s economic development mode undergoes a substantial transformation so that the traditional international circulation strategy no longer fits the situation of the Chinese economy. In response, the Chinese government has proposed the ‘dual-circulation strategy’ to guide future directions. The dual-circulation strategy refers to the formation of a new development pattern that is primarily based on national circulation, with the national and international circulations mutually reinforcing each other. This strategy posits that with a solid foundation of domestic demand, we should effectively align with international norms to achieve a higher level of international circulation. Eventually, this will facilitate the creation of a new advantage in cooperation and competition, as well as an enhanced international competitiveness. It is no doubt that multinational enterprises play a pivotal role in the advancement of economic globalization. This study focuses on factors affecting Chinese manufacturing multinational enterprises’ foreign exit, serving as an endeavor to help Chinese firms better survive and develop in the overseas market and obtain sustainable competitive advantages. Furthermore, we also incorporate environmental uncertainty to be more in line with the real status of Chinese firms, hoping to provide more targeted advice for the enterprise’s strategic choice and policy introduction.

Existing studies on the antecedents of foreign exit are mainly derived from three dimensions: subsidiary, parent firm, and country-level. Subsidiary level mainly explores the effects of organizational performance, firm characteristics, resources and capabilities, and firm networks on exit strategy. Without a doubt, firm performance has always been a key factor for strategic choice. Poor subsidiary performance will significantly increase the likelihood of exit [5]. However, performance is not the only determinant that other contextual factors should be taken into consideration as well [6,7,8]. Firm age [9], size [10], and irreversibility of entry mode may ultimately influence exit decisions by affecting operating flexibility. Resources and capabilities are the core sources of a sustainable competitive advantage. Thus, possessing abundant resources and superior capabilities undoubtedly improves MNEs’ survival in foreign markets and reduces their exit probability [11]. Different subsidiaries under the same parent company constitute the MNE’s network. It is justified that the location in the network and interaction with other subsidiaries affect the firm’s strategic choice since they reflect how easy or difficult it is to acquire materials needed to survive and to operate flexibly [12,13]. The parent firm level mainly explores the impact of corporate characteristics, international experience, and intangible assets. Existing studies have shown that parent firms with larger scales [10], richer in international experience and intangible assets [14], can provide subsidiaries with resources and knowledge necessary for survival, thus helping them to better overcome risks and challenges. At the country level, research mainly focuses on factors such as geopolitics, the economic attractiveness of the host country, and the development of new technologies. Witt [15] argues that geopolitics is the main driver of foreign exit, while economic forces are the second, albeit with asymmetric effects [16]. Slower economic growth [17], rising labor costs [3], deteriorating business environments, and governance uncertainty all reduce the attractiveness of host markets for multinational enterprises. Finally, the use and growing sophistication of digital technologies enable companies to reduce the need to offshore and shorten global supply chains to reduce risk [18]. In terms of studies exploring the situations of Chinese enterprises, though rather few in numbers, scholars have identified that a firm’s entry mode, international performance, and strategic misfit may exert an influential effect on exit choice [8,19,20,21]. According to Bai’s [19] study, the greenfield investment mode is positively related to foreign subsidiaries’ survival, while the acquisition mode is accompanied by a higher risk of foreign exit. However, it is inappropriate to solely explore the antecedents without taking contextual elements into consideration. Thus, existing literature has explored the internal and external elements serving as contextual variables. A Firm’s international experience, innovation capability [21], and business relatedness [20] between subsidiaries and headquarters can significantly weaken the negative influence of poor organizational performance. In addition, the economic distance and cultural distance, considered external contextual elements, reflect differences between the host and home country. It has been clarified that under the condition of a large economic distance, enterprises are able to flexibly locate and operate in chosen host countries with lower hazard rates and enjoy certain location-specific advantages in both resource exploration and exploitation. However, enterprises operating in a culturally distant host country will face more uncertainty, which could augment the difficulty of expanding abroad [19]. In conclusion, although existing research perspectives are comprehensive, covering both micro and macro levels, the focus of each study is dispersed. In addition, the logical inference based on different theoretical foundations will inevitably produce heterogeneous results, which makes it difficult for us to accurately understand the motivation of foreign exit strategies. To address this shortcoming, this study focuses on the whole stage of development based on behavioral theory and explores the impact of a firm’s resources, capacity, and cost of exit strategy.

Compared to dimensions such as the internationalization degree, speed, rhythm, entry mode, and location choice, research on foreign exit is relatively limited. As an important strategic decision, the formulation and implementation of foreign exit will definitely affect the specific behavior of companies. Therefore, this study adopts behavioral theory to explore the antecedents and contextual factors of Chinese firms’ foreign exit. Cyert and March [22] used four constructs to explain the framework of firm behavior, including organizational learning, search, uncertainty avoidance, and bounded rationality. Thus, behavioral theory has been subsequently extended to five dimensions, which are organizational learning, performance feedback, cost perspective, attention-based view, and upper echelons theory [23]. Nowadays, it is widely used to explain risky strategic decisions such as new product launches, innovation, alliance partner selection, M&As, R&D investment, new market entry, and internationalization. In this study, we focus primarily on the cost, performance feedback, and learning dimensions of the behavioral theory and explore their implications for Chinese firms’ foreign exit strategies.

The research contributions of this paper are as follows: (1) Existing studies based on different theoretical perspectives tend to focus on different dimensions of the antecedents and are likely to produce heterogeneous conclusions. This study adopts behavioral theory to study the possible antecedents of foreign exit in an attempt to provide a comprehensive and systematic conclusion. (2) Previous research examined exit phenomena from advanced countries and explored antecedents from multi-dimensions [10,11,13,14]. This was supported by well-established databases such as the Kaigai Shinshutsu Kigyou Souran database, KLCA (Korean Listed Companies Association), and so on. Although there exists research focusing on Chinese firms, it mainly uses primary data [8,21]. It has been recognized that primary and secondary data have both advantages and shortcomings. Primary data may contain rich information providing questions that are carefully set, relying on subjective views that reflect respondents’ perceptions and thus may not be objective to some extent [24]. Secondary data, though they may not surpass the depth and richness of primary data, may present researchers with the temptation to arrive at more extensive conclusions than what the data would support [25]. Therefore, this study attempts to explore Chinese firms’ foreign exit phenomenon by using secondary data, endeavoring to fill the gap in the literature and draw a relatively comprehensive conclusion. (3) Volberda and Lewin [26] contend that strategic choices emerge from both internal organizational forces and external environment of home and host countries. Whether and how environmental uncertainty affects a firm’s exit choice has not been justified yet. Thus, this paper explores the moderating role of environmental uncertainty in order to draw a more comprehensive and accurate conclusion.

The structure of this study is organized as follows. Section 2 introduces the theoretical background and major hypotheses. Section 3 describes our data sources and variable measurements. Section 4 shows our empirical results, including baseline regression and all the robustness tests. Section 5 provides the discussion and conclusion of the article.

2. Theory and Hypotheses

2.1. Theoretical Background

The behavioral theory was developed by Cyert and March in 1963 and is commonly applied in explaining risk-taking decisions, the likelihood of strategic change, and search intensity [27]. In other words, the behavioral theory of the firm, with its emphasis on how firms assess performance according to aspiration levels, selectively learn and update routines, and selectively incorporate the learning of others, is perfectly suited to examine the diversity and change increasingly observed in internationalization decisions [28]. On the basis of bounded rationality, the behavioral theory assumes that decision makers prefer to make strategic decisions based on past experience, cost-saving pressure, and performance feedback under backward-looking logic. Whether foreign exit, as a high-risk strategic choice, can be similarly explained by behavioral theory deserves to be investigated. Although existing research has explored MNE’s survival in overseas markets in terms of resources, knowledge, capabilities, and external environment dimensions, such literature has only emphasized the importance of strategic foresight but ignored the assumption of bounded rationality [29].

According to behavioral theory, firms face challenges in achieving the optimal combination of resources and the lowest cost inputs. As a result, firms may adjust their strategic decision in response to ‘cost-saving pressure’ [30]. When a firm withdraws from foreign markets and re-enters, it will incur significant sunk entry costs. Against this background, firms may exhibit inaction or inertia behavior to stay still in existing markets. Choquette [14] and Dixit [31] also found that the reluctance to change in response to market conditions increases with the sunk cost. The greater the sunk cost, the more patient a firm will become in the face of unfavorable conditions and thus less likely to exit the market.

Behavioral theory can also be applied to explain the relationship between performance and behavior. It assumes that a firm’s past performance will shape its future strategic behavior. Therefore, firms will regulate and adjust their behavior based on relevant performance [20]. Performance feedback serves as a crucial theoretical foundation for comprehending firms’ strategic behavior, decision-making patterns, organizational learning, adaptability, evolution, and development [32]. Due to bounded rationality, organizations may set a level of satisfactory performance expectation based on their past goals, past performance, and the real situation of other similar organizations. The disparity between the actual performance and the expected level can result in either an expectation surplus or expectation deficit, which in turn determines the firm’s strategic decisions [22].

2.2. Sunk Cost and Foreign Exit

Originally, the behavioral theory of the firm was mostly built around three related but largely independent ideas: bounded rationality, imperfect environment matching, and unresolved conflict. Simon’s study [33] emphasized that decision making forms the heart of the organization, and the vocabulary of organizational theory must be derived from the logic and psychology of human choice [34]. However, due to biases in decision-making perspectives as well as limited information, rational actors do not exist in real situations [35]. Thus, firms will set a return expectation based on sunk cost due to bounded rationality [33]. The sunk cost refers to the costs of entering a foreign market and maintaining international operations after entry, including investment in intangible assets; sales and administrative expenses; and costs associated with understanding local laws, regulations, and consumer preferences. The expected level is higher when the sunk cost is greater. In a turbulent and changing external environment, enterprises with large sunk costs may feel more pressure and a greater sense of urgency to survive. Under such circumstances, firms will adopt a mentality of proactively seeking all possible solutions to solve the dilemma. For them, an exit strategy is no longer viewed as a passive business failure but a positive signal of strategic change. Management research on foreign exit has observed a similar shift in perspectives. In the 1970s, exits became an ordinary phenomenon due to the economic recession. At that time, the foreign exit was generally considered a passive strategy to respond to changes in the external environment and deterioration in firm performance. However, since the 1990s, scholars have gradually realized the positive strategic value of exit behavior. They argue that an exit strategy enables firms to seize new business opportunities and expand their capabilities [6,36]. It assists firms in reallocating resources and achieving strategic goals of adjusting international diversification or realizing the scope of the economy [37,38,39,40]. It can also improve a firm’s dynamic capabilities through strategic changes [41]. Based on the findings of strategic exit behavior, enterprises with high sunk costs, in the face of survival threats and competitive disadvantages, are more likely to take a proactive perspective to view exit strategy since it is an effective way to restructure the organization, optimize resource allocation, and improve economic efficiency. In the long run, a strategic exit is more conducive to the healthy and sustainable development of the enterprise.

Hypothesis 1.

Sunk costs increase the possibility of foreign exit.

2.3. Performance Feedback and Foreign Exit

Moliterno and Wiersema [42] proposed that a firm’s performance is influenced by its past operations and shapes its future strategic decisions. Organizations will set satisfactory performance expectations based on historical and social expectations [38,43] and adjust their current behavior and strategic choices in accordance with the performance gap [44]. When the actual performance exceeds the expected level, it is referred to as positive performance feedback. Conversely, it is negative. Positive performance feedback can lead firms to maintain the status quo and make decisions that only satisfy them rather than maximize the benefits [45,46,47]. In addition, it is a signal indicating their existing strategies are effective and feasible, and thus firms prefer to maintain current routines rather than innovate [48]. This is because change and innovation involve risk and uncertainty, and firms are reluctant to make decisions that may negatively affect organizational returns. However, negative performance feedback can deliver a clear and intuitive signal to managers that the existing organizational structure may not meet the needs of the host market [49]. Tough competitive disadvantages will prompt the firm to make upward social comparisons with better-performing organizations to understand its own shortcomings and engage in a problem-searching process, especially for new solutions far from current routines [50]. Simultaneously, the performance gap may increase the operational pressure on the executive team. This, in turn, could motivate the TMT to take risks in the decision-making process to seek new competitive advantages and legitimacy [51,52]. In summary, enterprises often exhibit the phenomenon of ‘Adversity leads to prosperity while wealth leads to persistence’ [53]. This means that firms tend to avoid making changes when positive performance feedback appears but attempt to search for all possible solutions and view foreign exit as a proactive strategy when performance feedback is negative.

Hypothesis 2.

Positive performance feedback decreases the possibility of foreign exit.

Hypothesis 3.

Negative performance feedback increases the possibility of foreign exit.

2.4. The Moderating Role of Environmental Uncertainty

Strategic management studies propose that environmental uncertainty comprises three dimensions, including environmental munificence, dynamism, and complexity [54,55,56]. Environmental munificence refers to the degree to which a firm’s growth and development are supported by its external environment through the availability of necessary resources. If the environment has more available resources, firms can create redundant resources for allocation. This can not only serve as a buffer to overcome risks and threats posed by uncertainty but also aid firms in maintaining partnerships and carrying out innovation [57,58]. Environmental complexity refers to the number and diversity of environmental factors faced by the firm. A complex environment puts pressure on a firm to deal with a wide variety of partners, business activities, and competitors from different fields [59]. As a result, environmental complexity can negatively impact a firm’s survival and development.

The greater the environmental munificence, the more resources available from the external environment. This is a positive signal for enterprises with high sunk costs or negative performance feedback because they are able to improve their business situation without withdrawing from existing overseas markets. However, firms with positive performance feedback may experience a relatively small or even negative marginal utility of environmental munificence. This is because positive performance feedback indicates that the current structure aligns with the needs of foreign markets. It also satisfies the need for self-satisfaction and recognition. Therefore, firms may become complacent and resistant to change due to long-term subconscious self-identification [60], which ultimately results in organizational inertia. This will limit a firm’s action to reproducing past successes and reduce motivation to innovate. In addition, positive feedback may also negatively affect a firm’s dynamic capabilities, reducing its ability to adapt to risk and uncertainty [61,62]. The absorptive capacity is hindered by the difficulty in identifying, analyzing, assimilating, and internalizing novel knowledge and information in the competitive market due to path and time dependence. In the end, this results in reduced efficiency of knowledge absorption, increased cost and time of resource conversion, and reduced strategic flexibility. Such hindrances are detrimental to the realization of economies of scope and scale [63,64]. In a word, limited motivation and core rigidity decrease the probability of survival and thus weaken the negative effect of positive performance feedback on foreign exit.

In a market with low environmental complexity, the number of competitors is relatively stable, making it difficult for new firms to enter and compete. Such a stable environment is advantageous to maintaining the survival of enterprises facing high sunk costs or negative performance feedback. However, for firms with positive performance feedback, the effect is similar to that of environmental munificence. In an environment with low complexity, the firm experiences greater self-satisfaction and identification, which ultimately results in a reluctance to innovate and core rigidity. In the long run, this will gradually weaken the firm’s comparative advantage and increase the likelihood of exiting foreign markets.

Hypothesis 4a.

Environmental munificence negatively moderates the positive effects of sunk costs on foreign exit.

Hypothesis 4b.

Environmental munificence positively moderates the negative effects of positive performance feedback on foreign exit.

Hypothesis 4c.

Environmental munificence negatively moderates the positive effects of negative performance feedback on foreign exit.

Hypothesis 5a.

Environmental complexity negatively moderates the positive effects of sunk costs on foreign exit.

Hypothesis 5b.

Environmental complexity positively moderates the negative effects of positive performance feedback on foreign exit.

Hypothesis 5c.

Environmental complexity negatively moderates the positive effects of negative performance feedback on foreign exit.

3. Research Design

3.1. Sample and Data Sources

Based on the availability and objectivity of Chinese OFDI data, this study selected A-share listed manufacturing multinationals in Shanghai and Shenzhen stock markets and their OFDI events from 2008 to 2022 to explore the drivers of foreign exit strategy. The OFDI and firm-level financial and governance structure data of Chinese MNEs were obtained from the China Stock Market Accounting Research (CSMAR). This study also matched the CSMAR database with the Directory of Overseas Investment Enterprises (Institutions) of the Ministry of Commerce of the People’s Republic of China in order to ensure reliability and accuracy. We eventually obtained data on 1425 manufacturing enterprises during the period of 15 years.

3.2. Definitions and Measurements of Main Variable

Foreign exit. In this study, we used a binary variable to measure firms’ OFDI exits. If a firm engages in normal OFDI in year t-1 but investment in year t is zero, then the firm is considered to have conducted a foreign exit in year t, which is defined as 1 and is 0 otherwise.

Sunk cost. Clark and Wrigley [65] classified sunk costs into three types: set-up sunk cost, accumulation sunk cost, and exit sunk cost. This study mainly focuses on exit sunk costs and employs a binary variable to measure the construct. If a firm engages in OFDI in year t-1 and chooses to exit the market in year t, then the investment amount in the previous year is considered as the firm’s sunk cost, which it is unable to recover in the future. The existence of a sunk cost is assigned a value of 1 and is 0 otherwise.

Performance feedback. Firm performance is multidimensional and multilevel by nature, and it is particularly difficult to select suitable indicators from the diverse and complex operationalization of the performance [66]. According to Hult’s study, financial performance centers on outcome-based indicators assumed to reflect economic goals, inclusive of accounting-based and market-based metrics. The measures listed under financial performance appear to be commonly used measures often found on financial statements. Thus, this study adopts financial performance and selects return on assets (ROA) to measure this construct. In addition, performance feedback is measured as the difference between the actual performance and the industry’s average level. Positive performance feedback is given a value of 1 when the actual firm performance in year t is higher than the industry’s average level, while negative performance feedback is given if the opposite is true.

Environmental munificence. Environmental munificence examines the extent to which the external environment supports firm growth and development by providing necessary resources. Following the mainstream methodology [54,67], this study uses munificence to denote a five-year average growth ratio of the amounts of industry gross revenue. We first regress the natural logarithm of industry gross revenues and an index variable of time in years, with time serving as an independent variable. Then, we divide the regression slope coefficient by the mean gross revenues of the past five-year period to measure the construct. For example, we tracked back to obtain the industry-level gross revenues during the 2003–2007 period to compute environmental munificence in 2008.

Environmental complexity. Environmental complexity refers to the number and diversity of environmental factors that enterprises face. Existing measurements typically focus on two aspects: question-item scales [59,68] and the industry concentration index [54,67]. This study adopts the second approach by implementing the Lerner Index to indicate industry concentration. The Lerner Index, also referred to as the Lerner Monopoly Power Index, reflects the level of monopoly power in a market. The index ranges from 0 to 1, with higher values indicating greater monopoly power, less competition in the market, and, therefore, less industry complexity.

Control variables. This study controls for several variables that may affect a firm’s foreign exit strategy, including firm size, R&D intensity, advertising intensity, asset–liability ratio, and board structure. We operationalize firm size as the natural logarithm of total assets. R&D intensity and advertising intensity are measured as the ratio of R&D investment to business revenue and sales expenditure to total sales revenue, respectively. The asset–liability ratio is measured as the ratio of total liabilities to total assets. The board structure is measured by dividing the number of independent directors by the total number of board of directors. Panel logistic estimation was adopted to test the hypotheses due to the binary nature of the dependent variable.

4. Results

4.1. Descriptive Analysis

Table 1. Descriptive statistics and correlation.

Variables	1	2	3	4	5	6	7	8	9
1. Foreign exit	1.000
2. Sunk cost	0.464	1.000
3. Positive performance feedback	−0.077	0.012	1.000
4. Negative performance feedback	0.077	−0.012	−1.000	1.000
5. Advertising intensity	−0.010	−0.005	−0.026	0.026	1.000
6. R&D intensity	−0.017	−0.007	−0.040	0.040	0.002	1.000
7. Asset–liability ratio	0.024	−0.014	−0.262	0.262	−0.054	−0.048	1.000
8. Firm size	0.047	0.003	0.029	−0.029	−0.017	−0.018	0.398	1.000
9. Board structure	−0.022	−0.023	−0.021	0.021	−0.024	−0.021	0.007	−0.003	1.000
Mean	0.139	0.035	0.490	0.510	0.476	9.001	0.413	22.550	0.382
SD	0.346	0.184	0.500	0.500	14.950	73.770	0.191	1.258	0.056
VIF	1.07	1.07	1.07	1.07	1.15	1.15	1.09	1.09	1.15

Source: Compiled by the author based on data from CSMAR database.

shows pairwise correlations of all variables included in this study. All correlations are fairly low except for that of positive performance feedback and negative performance feedback. It is important to note that the performance feedback of a company in the focal year can only be positive or negative, and the relationship between the two key variables is exactly the opposite. This study also calculates the average value of the variance inflation factors of all variables, which are all substantially below the acceptable level of 10, indicating that multicollinearity is not a serious issue.

4.2. Baseline Results

Table 2. Antecedents’ baseline results.

	Model 1	Model 2	Model 3	Model 4
Sunk cost		17.974 *** (11.806)
Positive performance feedback			−0.268 ** (−2.243)
Negative performance feedback				0.268 ** (2.243)
Controls	Yes	Yes	Yes	Yes
Observations	9010	9010	2887	2887

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

presents the antecedents’ baseline results. Model 1 includes all control variables, while Model 2 to Model 4 examine the effect of sunk costs and positive and negative performance feedback on foreign exit, respectively. The coefficient of the sunk cost is positive and significant at the 1% level in Model 2 (β = 17.974, SE = 11.806, p = 0.000), suggesting that sunk costs significantly increase the possibility of foreign exit. This result corroborates Hypothesis 1. Model 3 tests the relationship between positive performance feedback and exit choice. The result shows that the coefficient of positive performance feedback is negative and significant at the 5% level (β = −0.268, SE = −2.243, p = 0.025). Such a result provides support for Hypothesis 2, which proposes that positive performance feedback can effectively reduce the likelihood of firms’ foreign exit. Next, regarding Hypothesis 3, we find that the coefficient of negative performance feedback is positive and significant at the 5% level (β = 0.268, SE = 2.243, p = 0.025), thus lending support for Hypothesis 3.

Table 3. Moderating effect of environmental uncertainty.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	18.500 *** (11.804)			20.184 *** (11.407)
Positive performance feedback		−0.215 * (−1.744)			−0.208 * (−1.658)
Negative performance feedback			0.215 * (1.744)			0.208 * (1.658)
Environmental munificence	−4.351 *** (−4.146)	−1.626 ** (−2.172)	−1.626 ** (−2.172)
Environmental complexity				19.416 *** (9.113)	19.652 *** (9.601)	19.652 *** (9.601)
Sunk cost × munificence	−0.649 (−0.034)
Positive performance feedback × munificence		2.688 ** (1.978)
Negative performance feedback × munificence			−2.688 ** (−1.978)
Sunk cost × complexity				−25.306 * (−1.701)
Positive performance feedback × complexity					6.447 *** (2.673)
Negative performance feedback × complexity						−6.447 *** (−2.673)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9010	2887	2887	8998	2876	2876

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

presents the results of the moderating effect of environmental uncertainty. Models 1 to 3 examine the moderating role of environmental munificence on the relationship between sunk costs, positive and negative performance feedback, and firms’ foreign exit, while Models 4 to 6 correspondingly test the effect of environmental complexity. Model 1 shows that the coefficient of the interaction term between environmental munificence and the sunk cost is not significant (β_moder = −0.649, SE = −0.034, p = 0.973). Thus, Hypothesis 4a is not supported. The coefficient of the interaction term between positive performance feedback and environmental munificence is positive and significant at the 5% level in Model 2 (β_moder = 2.688, SE = 1.978, p = 0.048). Such results provide support for Hypothesis 4b, which proposes that as the environmental munificence increases, the negative effect of positive performance feedback on foreign exit will be attenuated. According to Model 3, the interaction term between negative performance feedback and environmental munificence is negative and significant at the 5% level (β_moder = −2.688, SE = −1.978, p = 0.048), corroborating Hypothesis 4c. Differing from Model 1, the interaction term of the sunk cost and environmental complexity is significant and negative at the 10% level in Model 4 (β_moder = −25.306, SE = −1.701, p = 0.089). These results supports Hypothesis 5a, which proposes that the positive effect of the sunk cost on firms’ foreign exit diminishes when environmental complexity is lower. However, identical to environmental munificence, the results of Models 5 and 6 show that the moderating effect of environmental complexity on the nexus between positive performance feedback and foreign exit is significantly positive (β_moder = 6.447, SE = 2.673, p = 0.008), while that of negative performance feedback is significant and negative at the 1% level (β_moder = −6.447, SE = −2.673, p = 0.008). These results provide support for both Hypothesis 5b and 5c.

4.3. Robustness Checks

4.3.1. Robustness Check by Using Different Estimation Methods

This study adopts different estimation methods to test the robustness and reliability of the findings. On the basis of the characteristics of the data, we used the panel Probit random effect model and the Cox proportional hazard model to run the robustness check.

Table 4. Robustness check by changing estimation model.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	10.123 *** (11.557)			9.271 *** (18.255)
Positive performance feedback		−0.227 *** (−3.733)			0.639 *** (−2.693)
Negative performance feedback			0.227 *** (3.735)			1.565 *** (2.693)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9010	9000	9000	1398	1397	1397

The coefficients of Models 1 to 3 are probit regression results, with negative (positive) coefficients interpreted as decrease (increase) in foreign exit risk rate; the coefficients of Models 4 to 6 are the results of the risk ratios in Cox proportional hazard model, with coefficients greater than 1 (less than 1) interpreted as an increase (decrease) in foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

shows the regression results, where Models 1 to 3 present the Probit model, and Models 4 to 6 are the Cox proportional hazard model. It is found that the results in Table 4 are consistent with those of baseline research, which indicates that the findings are less dependent on the estimation method.

4.3.2. Robustness Check by Adding Additional Control Variables

In baseline regression, we include five control variables that are closely related to firms’ foreign exit strategy. To further test the robustness of our findings, this study also adds variables related to governance structure and development capability, i.e., equity balance and income growth rate.

Table 5. Robustness check by adding control variables.

	Model 1	Model 2	Model 3	Model 4
Sunk cost		17.377 *** (10.856)
Positive performance feedback			−0.265 ** (−2.216)
Negative performance feedback				0.265 ** (2.216)
Controls	Yes	Yes	Yes	Yes
Observations	9010	9010	2887	2887

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

presents the results of antecedent exploration, while

Table 6. Robustness check of moderating effect by adding control variables

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	18.218 *** (11.594)			19.364 *** (10.655)
Positive performance feedback		−0.211 * (−1.715)			−0.204 (−1.623)
Negative performance feedback			0.211 * (1.715)			0.204 (1.623)
Environmental munificence	−4.341 *** (−4.128)	−1.625 ** (−2.164)	−1.625 ** (−2.164)
Environmental complexity				19.042 *** (8.815)	19.674 *** (9.594)	19.674 *** (9.594)
Sunk cost × munificence	−0.720 (−0.038)
Positive performance feedback × munificence		2.738 ** (2.012)
Negative performance feedback × munificence			−2.738 ** (−2.012)
Sunk cost × complexity				−24.304 * (−1.656)
Positive performance feedback × complexity					6.435 *** (2.666)
Negative performance feedback × complexity						−6.435 *** (−2.666)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9010	2887	2887	8998	2876	2876

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

shows the robustness check of the moderating effect. The results are consistent with baseline regression, indicating that our findings are robust and reliable.

4.3.3. Robustness Check by Changing the Measurement of Moderator

In baseline regression, we chose industry sales to calculate environmental munificence. To further test the robustness of the results, this study switched to the total industry revenue indicator to measure environmental munificence.

Table 7. Robustness check by changing measurement.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	18.502 *** (11.806)			20.184 *** (11.407)
Positive performance feedback		−0.214 * (−1.737)			−0.208 * (−1.658)
Negative performance feedback			0.214 * (1.737)			0.208 * (1.658)
Environmental munificence	−4.383 *** (−4.168)	−1.628 ** (−2.172)	−1.628 ** (−2.171)
Environmental complexity				19.416 *** (9.113)	19.652 *** (9.601)	19.652 *** (9.601)
Sunk cost × munificence	−0.591 (−0.031)
Positive performance feedback × munificence		2.703 ** (1.987)
Negative performance feedback × munificence			−2.703 ** (−1.987)
Sunk cost × complexity				−25.306 * (−1.701)
Positive performance feedback × complexity					6.447 *** (2.673)
Negative performance feedback × complexity						−6.447 *** (−2.673)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9010	2887	2887	8998	2876	2876

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

shows that the results are consistent, indicating that our findings are robust and reliable. In addition, following Goll and Rasheed’s [69] study, environmental munificence can also be measured by exploring the ten-year period prior to the survey. Thus, we re-ran the regression process of moderating effect by adopting this measurement. The result is shown in

Table 8. Robustness check by changing measurement for 10-year period.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	19.246 *** (11.713)			20.184 *** (11.407)
Positive performance feedback		−0.157 (−1.220)			−0.208 * (−1.658)
Negative performance feedback			0.157 (1.220)			0.208 * (1.658)
Environmental munificence	−14.756 *** (−7.272)	−6.639 *** (−4.598)	−6.638 *** (−4.598)
Environmental complexity				19.416 *** (9.113)	19.652 *** (9.601)	19.652 *** (9.601)
Sunk cost × munificence	−1.879 (−0.055)
Positive performance feedback × munificence		4.023 * (1.777)
Negative performance feedback × munificence			−4.023 * (−1.777)
Sunk cost × complexity				−25.306 * (−1.701)
Positive performance feedback × complexity					6.447 *** (2.673)
Negative performance feedback × complexity						−6.447 *** (−2.673)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9000	2887	2887	8998	2876	2876

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

, further clarifying the robustness of our study.

4.3.4. Robustness Check by Shortening Time Span

The sample included in the baseline regression contains the period of the global financial crisis of 2008–2009 and the coronavirus pandemic of 2020–2022. Given that such exogenous shocks may exert an additional impact on a firm’s exit strategy, we also shortened our time span to 2010–2019 to test the robustness of our results.

Table 9. Robustness check by shortening time span.

	Model 1	Model 2	Model 3	Model 4
Sunk cost		19.952 *** (11.490)
Positive performance feedback			−0.106 (−0.633)
Negative performance feedback				0.106 (0.633)
Controls	Yes	Yes	Yes	Yes
Observations	5139	5139	1582	1582

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

and

Table 10. Robustness check of moderating effect by shortening time span

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	18.918 *** (9.736)			18.418 *** (9.851)
Positive performance feedback		−0.030 (−0.175)			−0.136 (−0.788)
Negative performance feedback			0.030 (0.175)			0.136 (0.788)
Environmental munificence	−5.051 ** (−3.521)	−1.814 ** (−2.010)	−1.813 ** (−2.010)
Environmental complexity				11.288 *** (4.080)	15.901 *** 5.346)	15.902 *** (5.346)
Sunk cost × munificence	3.795 (0.200)
Positive performance feedback × munificence		2.847 (1.640)
Negative performance feedback × munificence			−2.847 (−1.640)
Sunk cost × complexity				−15.234 (−0.592)
Positive performance feedback × complexity					9.197 *** (2.533)
Negative performance feedback × complexity						−9.197 *** (−2.533)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	5139	1582	1582	5130	1574	1574

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

show the results of antecedent exploration and the moderating effect. We can notice that there are differences in the significance of the coefficient, but the effect remains in the same direction as the baseline analysis. A possible explanation for this difference is that such antecedents explored in this study may affect foreign exit more significantly under complex and uncertain environments than in stable environments. In other words, the influence of sunk costs and performance feedback appears more obvious if worldwide exogenous shocks happen.

4.4. Heterogeneity Analysis

The international business literature has long considered a large firm size to be a particularly advantageous resource for firms [70], which is likely to influence firms’ strategic decisions. Compared to SMEs, large firms have more internal and external resources that can be used for OFDI, and at the same time, large firms have more privileged learning channels to reduce the uncertainty and risk of internationalization. In summary, they are significantly different from SMEs in terms of resource endowments and dynamic capabilities. Therefore, this study examines the impact of different firm sizes on the focal relationships.

Table 11. Baseline results by differential firm size.

	Model 1	Model 2	Model 3
Sunk cost × size	18.250 *** (10.187)
Positive performance feedback × size		−0.198 (−1.629)
Negative performance feedback × size			0.198 (1.629)
Controls	Yes	Yes	Yes
Observations	9010	2887	2887

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

and

Table 12. Moderating effect by differential firm size.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Sunk cost	18.509 *** (11.697)			20.186 *** (11.388)
Positive performance feedback					−0.196 (−1.562)
Negative performance feedback						0.196 (1.562)
Environmental munificence	−4.389 *** (−4.189)
Environmental complexity				19.425 *** (9.117)	19.344 *** (9.502)	19.344 *** (9.502)
Sunk cost × munificence × size	−3.302 (−0.169)
Positive performance feedback × munificence × size		2.687 ** (2.108)
Negative performance feedback × munificence × size			−2.687 ** (−2.108)
Sunk cost × complexity × size				−26.173 * (−1.798)
Positive performance feedback × complexity × size					5.589 ** (2.297)
Negative performance feedback × complexity × size						−5.589 ** (−2.297)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9010	2887	2887	8998	2876	2876

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

present corresponding results. According to Table 8, the sunk cost differs significantly across firm sizes, and the positive effect is more pronounced for large firms than for small- and medium-sized firms. However, there is no significant difference between the effect of positive and negative performance feedback on firms of different sizes. Table 9 demonstrates that the moderating effect of environmental munificence and complexity on foreign exit is more significant for large firms than for SMEs. Both moderators can significantly weaken the strength of focal nexuses. The results are consistent with those of the full sample, further indicating that our findings are robust and reliable.

4.5. Endogeneity Test

Shaver [71] stated that firm strategic choices, based on their past experience and forecasting of future performance, are often endogenous rather than exogenous. In addition, the theory of firm-specific advantages suggests that there is a causal relationship between a firm’s intangible resources and capabilities and its ‘going global’ strategy [72]. Thus, this study lags both independent and control variables by one year and uses a dynamic logistic random effects model to address the endogenous problem arising from omitted variable bias and reverse causation. This approach has been widely applied in existing research [72,73,74,75].

Table 13. Results of endogeneity test.

	Model 1	Model 2	Model 3	Model 4
Sunk cost		1.724 *** (10.960)
Positive performance feedback			−0.462 *** (−4.163)
Negative performance feedback				0.462 *** (4.166)
Controls	Yes	Yes	Yes	Yes
Observations	8524	8524	8513	8513

The coefficients in the table are logistic regression coefficients; negative (positive) coefficients are interpreted as a decrease (increase) in the foreign exit risk rate; standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1; source: compiled by the author based on data from CSMAR database.

presents the regression results. Identical to the baseline results, sunk costs and negative performance feedback can negatively affect the likelihood of foreign exit, while positive performance feedback significantly reduces the risk of exit. This thus clearly justifies that the endogeneity problem does not influence our findings at all.

5. Conclusions

Based on behavioral theory, this study empirically examines the relationship between sunk costs and performance feedback on Chinese firms’ foreign exit using data from A-share-listed manufacturing firms from 2008 to 2022. We also explore the moderating effect of environmental munificence and complexity. The results show that positive performance feedback can significantly reduce the probability of exiting, while sunk costs and negative performance will push firms to exit from overseas markets. However, environmental munificence and complexity may, to some extent, attenuate the driving effects of sunk costs and performance feedback on the foreign exit strategy.

There are two interesting findings in this study. The first one is the positive driving effect of sunk costs. Due to the upfront sunk cost of entering a foreign market, firms are prone to stay in the market and make long-term resource commitments, which ultimately constitute state dependence or the long-term stability of international operations. However, our study depicts the opposite result on the basis of a strategic perspective on foreign exit behavior. Firms with higher sunk-cost inputs will face greater survival pressure and are at a competitive disadvantage in the foreign market. Under such circumstances, firms are more likely to make riskier decisions. Despite the sunk cost of exit, firms pay more attention to the strategic value of resource reallocation and firm restructuring, which is conducive to long-term sustainable development. Therefore, our findings provide new evidence of Chinese firms in the strategic exit field. Foreign exit is no longer a sign of failure as in the 1970s but rather an act of strategic change undertaken by firms after careful consideration. The second interesting phenomenon is that firms with positive performance feedback are more likely to exit overseas markets in high-munificence and low-complexity environments. This is strange because such an environment is supposed to be favorable for survival due to the abundance of resources and reduction in uncertainty. Based on our study, the opposite effect mainly comes from the psychological reluctance to change, a lack of innovation incentives, and rigidity of core capacity. As a result, firms’ comparative competitive advantage will gradually be lost, thus increasing the likelihood of exit. This finding is also a wake-up call for decision makers to avoid falling into the ‘capability trap’ and bear in mind that development is a long-term process with persistent organizational learning and innovation.

Our findings provide practical suggestions for both enterprises and the government. Enterprises should try various ways to improve international performance because it can significantly reduce the likelihood of foreign exit. This can be realized by persistently enhancing their innovation capacity because innovation represents the primary catalyst for economic advancement. In addition, this is conducive to the cultivation of comprehensive dynamic capability, enabling enterprises to better adapt to the complex environment. On the other hand, the government should provide enterprises with beneficial policies so as to help them better survive in foreign markets. Firstly, it is necessary to build a network or platform for firms to freely exchange knowledge and share international experience, which is conducive to reducing sunk costs before entering new overseas markets. In our results, it has been justified that sunk costs are positively related to foreign exit. That is, enterprises with lower sunk costs are less likely to exit from host countries. Meanwhile, facilitating the negotiation of bilateral and plurilateral free trade agreements is also a key initiative that the government needs to focus on. China has long followed a foreign policy of treating its neighbors as good friends. Thus, we should maintain this advantage to promote industrial integration and trade flows as well. The pilot free-trade zones, ports, and inland open centers can be used as the new frontiers of opening up to the overall world. They can also provide new channels of opening up in greater quantity and quality for multinational enterprises to develop. Secondly, according to the strategic perspective of foreign exit, exit is no longer a failure but a strategic chance. Therefore, it is also a good choice for the government to attract Chinese multinationals back to the domestic market by fostering a sound business environment. In this way, the government can not only benefit from solid national circulation but also gradually reduce our dependence on foreign resources and technologies as well.

This study has certain limitations that could be addressed in future research. First, we focus on parent-firm drivers and ignore the examination of subsidiary and country-level factors due to data limitations. Future research can enrich the database in order to explore the joint effect of multilevel factors. Second, although we introduce a proactive strategic exit perspective, we fail to measure and test whether the motivation of strategic change is positive or negative. Future studies may analyze the psychological motivation of exit decisions through a more detailed mechanism design. Finally, this study only explores the moderating effect of environmental uncertainty on foreign exit. However, there are many other contextual factors that may affect a firm’s exit decision, such as formal and informal institutions. Future research can expand this field by taking more contextual factors into consideration.