The Impact of Outward Foreign Direct Investment on Product Quality and Export: Evidence from China

Abstract

While China’s outward foreign direct investment (OFDI) is growing rapidly, few studies have considered the impact of this growth on product quality. This study uses microdata at the enterprise level to investigate the effects of OFDI on product quality and export trade, from the perspective of enterprise heterogeneity. The results denote that (1) OFDI improves product quality, which is caused by the direct promotional effect of OFDI and the productivity effect brought by OFDI. (2) As OFDI of low productivity enterprises creates a productivity effect and leads to a promotional effect on product quality enhancement, the magnitude is significantly higher than that of high productivity enterprises. (3) Product quality enhancement significantly increases the promotional effect on exports induced by OFDI of low productivity enterprises, while simultaneously significantly suppressing the promotional effect on exports induced by OFDI of high productivity enterprises. (4) The promotional effect of OFDI on exports decreases as productivity increases, and OFDI significantly promotes exports of low productivity enterprises and has no significant effect on high productivity enterprises. These results suggest that China should continue to expand the space for international cooperation and encourage enterprises to invest overseas, especially those facing export pressures.

1. Introduction

With the advancement of global economic integration and the rapid development of China’s economy, China’s international trade has flourished. According to data from the National Bureau of Statistics of China, the country’s export value has increased from USD 325.6 billion in 2002 to USD 2486.68 billion in 2018, with its percentage share of global total export value rising from 4.9% to 12.8%. Hence, China has remained the largest exporter in the world for numerous years. With the implementation of China’s “Go Out” policy, the scope of outward foreign direct investment (OFDI) by Chinese enterprises has expanded continuously, leading to a full-range enhancement in international market participation. Data from the 2018 Statistical Bulletin of China’s OFDI issued by the Ministry of Commerce revealed that up until 2018, the flow and stock of China’s OFDI ranked second and third in the world, respectively. Domestic investors established 43,000 OFDI enterprises in 188 countries (regions) globally, and more than 80% of the countries (regions) in the world had Chinese investments.

Nonetheless, a glance at export product quality reveals some “hidden threats”. The added value of China’s export products is still low, with foreign trade remaining at the middle and low ends of the global value chain. As global market competition intensifies, enterprises’ export product quality requirements are increasing. Exploring the dynamic mechanisms of export product quality enhancement has become an important topic. The Cournot competition model, which is based on the concept of game theory and explains the decisions of international investments, shows that local companies that lag in technology can directly invest in countries with advanced technology to achieve technological advancement. Furthermore, technology diffusion, demonstration imitation, industry linkage, and employee training effects enhance the productivity and technology level of the home country [1] and, in turn, push export structural adjustments and value chain upgrades [2]. China intends to effectively unleash the power of optimal transnational resource allocation and better utilize the specific comparative advantages of the host country in areas such as technology and efficiency through the “Go Out” policy [3]. However, few studies have examined the impact of China’s OFDI on product quality. Enhancing export product quality is the key to maintaining the continuous development of export trade. This study investigates whether OFDI promotes product quality enhancement and whether it affects exports.

This study has two practical contributions to the research gap. First, it applies micro-matched data from the China Industry Business Performance Database, the Directory of Overseas Investment Enterprises (Institutions), and the China Customs Import and Export Statistics Database, and ensures the accuracy of the research. Second, this study found that OFDI improved export product quality, which was caused by the direct promotional effect of OFDI and the productivity effect prompted by OFDI, and product quality then further promoted exports. These findings provide a new explanation for the relationship between export and OFDI and serve as a reference for advancing China’s foreign trade development. Encouraging OFDI can not only facilitate the upgrading of product quality but can also increase the export scale, especially for enterprises facing export pressure.

2. Literature Review

With the spread of the global epidemic, the economy is subject to increased risks of uncertainty, and China’s foreign trade environment is experiencing many challenges. Improving the business environment at home and abroad [4], strengthening international cooperation [5], engaging in diversified cooperation and exchange [6], and ensuring global economic stability are crucial for the sustainable development of the world [7]. Against this background, actively improving the quality of export products and cultivating the competitiveness of international trade have become essential for China to respond to changes in the international and domestic situation. The impact of OFDI on Chinese enterprise exports has garnered increasing attention.

The first type of literature related to this article evaluates the impact of OFDI on exports. Most conclusions support the view that OFDI promotes exports [8,9]. Although the disparity in the industrial structures of China and the host countries rises continuously, the export creation effect of OFDI has gradually strengthened [3]. Using data from the Belt and Road Project, Chen et al. [10] proved that Chinese OFDI promoted market expansion. As the effect of OFDI on export trade is related to the invested industries, the developmental stages of the destination countries of investment, and the types of export products, the partial substitution of export trade for OFDI cannot be ruled out [11]. The above empirical analyses were conducted mainly at the macro level. However, Jiang and Jiang [12] adopted the perspective of enterprises at the micro level and divided enterprises’ investment motives into four types (business and trade services, local production, technology research and development (R&D), and resource development). Their research discovered that different investment motivations yield significantly different corresponding effects of OFDI on export trade. Different motivations for internationalization have an important role in this process [13]. Yan et al. [14] further divided export products into six types (intermediate products, consumer goods, capital goods, and labor-intensive, capital-intensive, and technology-intensive products). They found that the effect of OFDI on export trade differs according to the type of export product but is not impacted by the intensity of the export products or the investment method.

The investigation of the relationship between OFDI and export trade is, in essence, an exploration of the optimal approach for a country’s economy to integrate into the international market. To choice to participate in international market competition, choose between OFDI and export trade, or to adopt both approaches depends mainly upon the enterprise’s profit maximization targets. From the perspective of enterprise heterogeneity, Melitz [15] pointed out that exporting enterprises generally have higher productivity than enterprises that focus only on domestic sales. Helpman et al. [16] extended the analysis of Melitz [15] to FDI, showing that the highest productivity enterprises choose FDI, the middle-range productivity enterprises choose to export, and the lowest productivity enterprises serve the domestic market. Whether an enterprise chooses export trade or FDI depends on the comparison between the export cost borne by the enterprise and the cost of the fixed input incurred by FDI. In other words, when the export cost of an enterprise is higher than the fixed input of FDI, the enterprise will choose FDI; otherwise, it will be inclined to choose exports.

The second type of literature related to this article examines the impact of OFDI on enterprise production efficiency. The major motive for OFDI is to improve domestic firms’ productivity levels via the learning effect of OFDI [17]. Seyoum et al. [18] noted that foreign firms are more productive and that their presence has significant spillover effects on the productivity of domestic firms with higher absorptive capacity. Lichtenberg (2001) highlighted that enterprises investing in technology-intensive countries improve the productivity levels of the home countries significantly. As multinational enterprises set up barriers to avoid technology spillovers to competitors rather than the productivity effect among enterprises, the productivity effect within an enterprise is more likely to occur [19]. Kogut and Chang [20] examined the direct investment of the Japanese manufacturing industry in the US and found that Japan usually invested in R&D-intensive industries in the US through joint ventures. After Japanese enterprises made direct investments in the US, the number of patent applications and the level of technological progress in Japan increased significantly. Correspondingly, Tang and Altshuler [21], using U.S. firm-level information, found positive and significant reverse technology spillovers from multinational customers to domestic suppliers. Research in other developed countries has also obtained similar results. Additionally, Chen et al. [22] focused on emerging market multinational enterprises and found that emerging market multinational enterprises that have subsidiaries in host developed markets exhibit stronger technological capabilities at home.

The third type of literature related to this article examines the factors influencing product quality. Yu and Zhang [23] outlined that improved productivity levels can induce parent companies to further improve the production process and flow of enterprises to improve the technical content and quality of products. Some scholars have tested the relationship between productivity and product quality [24,25,26,27]. Haini et al. [28] reported that export diversification contributes to product upgrading, while Zhang and Chen [2] show that accelerating economic development and improving absorptive capacity can promote the contribution of FDI to China’s export maturity. Rehman and Noman [29] found that foreign investment promoted export maturity, and Liu and Wang [30] reached a similar conclusion. Hou et al. [31] pointed out that the deregulation of foreign entry helps enterprises to improve product quality. Compared with the service industry, the deregulation of the manufacturing industry has a greater impact on the quality of downstream products. Anwar and Sun [32] state that FDI improves the quality of export products; however, few studies have gauged the impact of OFDI on product quality.

The large-scale OFDI of Chinese enterprises began relatively late. Chinese enterprises have no obvious advantages in OFDI compared with those in developed countries in Europe and America. Furthermore, the international competitive advantage in low labor costs is diminishing due to changes in the scarcity of labor supply. Nurturing new advantages in international competition by strengthening innovative activities and technical training is critical to increasing enterprises’ competitive advantages in the international market [33]. Jiang and Jiang [12] utilized data on 761 OFDI enterprises from 2004 to 2006 and found that OFDI enterprises diffused the comparative advantages of the host country’s technology back to domestic enterprises through reverse technological learning. This approach aids in enhancing the sustainable production capacity of Chinese enterprises and improving the international competitiveness of export enterprises [34]. Based on the analysis above, this research proposes the following hypothesis.

Hypothesis 1:

China enterprises’ OFDI contributes to the improvement in product quality.

Some studies have provided much inspiration for the present study. However, most of the existing studies investigate the problem from the perspective of macro-level data, which carry challenges in terms of aggregation biases and omitted variables. Hence, the individual characteristics of different investment subjects were easily concealed, resulting in a certain level of endogeneity, which affects the accurate clarification of the relationship between foreign investment and export trade [35]. As the assimilation of China’s economy into global economic integration accelerates, the connections between OFDI and export trade become closer. OFDI primarily depends on enterprise behavior at the micro level, while the OFDI of different enterprises carries major heterogeneity. In this article, this study employs enterprise-level data more directly and accurately, which helps to clarify and deepen the relationship between OFDI and exports.

3. Empirical Research Design

3.1. Model Specification

Advancing the high-quality development of foreign trade is an essential aim of OFDI. To test the effect of OFDI on product quality and how product quality enhancement contributes to the export effect of OFDI, the following regression model was established.

First, the effect of OFDI on product quality was tested. Based on the above analysis, OFDI not only directly promotes product quality but also indirectly promotes product quality through total factor productivity (tfp). The models are as follows:

quality_it = β₀ + β₁OFDI_it + β₂Z_it + μ_s + μ_r + μ_t + ε_it

(1)

tfp_it = α₀ + α₁OFDI_it + α₂Z_it + μ_s + μ_r + μ_t + ε_it

(2)

$$qualit{y}_{it}={\alpha }_0+{\alpha }_{1}OFDI+{\alpha }_{2}tfp+{\alpha }_{3}OFDI\ast tfp+{\alpha }_4{Z}_{it}+{\mu }_s+{\mu }_r+{\mu }_t+{\epsilon }_{it}$$

(3)

Model (1) tests the direct effect of OFDI on product quality, and Models (2)–(3) test the indirect effect of OFDI on product quality. In these models, i and t represent the enterprise and the sample year, respectively. The quality indicator is utilized to measure the product quality level of an enterprise’s export products. OFDI is set as a binary variable, where 1 indicates that the enterprise chooses OFDI and 0 indicates otherwise. As enterprises do not usually choose OFDI every year, this study follows Li et al.’s [11] method and assumes that an enterprise will choose OFDI each year after the year it chooses OFDI. Regarding tfp, combined with the data characteristics employed in this study and with reference to Lu and Lian [36], the OP (Olley–Pakes) method was chosen for estimation. This study adopted the OP method to adequately address endogeneity issues and sample selection biases, hence generating relatively accurate estimation results. Z_it is the relevant control variable, including enterprise scale, represented by the total assets of the enterprise; capital intensity (kl) is represented by the enterprise’s net fixed assets divided by the enterprise’s number of employees at the end of the year; age is represented by the year of sampling minus the firm’s date of establishment; and we look at whether it is a state-owned enterprise, which is set as a binary variable, where 1 indicates that the enterprise is a state-owned enterprise and 0 indicates otherwise.

Second, to examine how export product quality contributes to the export effect of an enterprise’s OFDI, interaction terms for OFDI and export product quality (OFDI × quality) are introduced to explain the influence of product quality on the export trade effect induced by OFDI. The models utilized are as follows:

$$E{X}_{it}={\beta }_0+{\beta }_{1}OFDI+{\beta }_{2}quality+{\beta }_{3}OFDI\ast quality+{\beta }_4{Z}_{it}+{\mu }_s+{\mu }_r+{\mu }_t+{\epsilon }_{it}$$

(4)

Notably, scholars in China and elsewhere have provided detailed explanations and analyses [23,37]. In this study, the measurement of an enterprise’s export product quality was performed by mainly following the method of Yu and Zhang [23], and the calculation is as follows:

$$\ln (z_{ijgt})={\theta }_g\{\ln [{\alpha }_{jg}{\theta }_g({\sigma }_g-1)/[1+{\alpha }_{jg}{\theta }_g({\sigma }_g-1)]]+\ln (p_{\mathrm{ijgt}})+\ln ({\phi }_{it})-\ln (w_t)\}$$

(5)

where $i$ represents the export enterprise, $j$ signifies the export destination, $g$ denotes the export product category, $t$ outlines the export time, $p_{ijgt}$ is the offshore unit price of the exports, ${\phi }_{it}$ is the productivity of the enterprise, $w_t$ is the cost of the input, and ${\alpha }_{ig}$, ${\theta }_j$, and ${\sigma }_g$ are parameters at the national HS6-level. The export product quality $z_{ijgt}$ can be derived from Equation (5). Due to space limitations, the detailed derivation and analysis process is not explained here in detail. Interested readers can refer to Yu and Zhang [23].

3.2. Data Sources and Processing

The data utilized in this study were mainly combined and retrieved from the China Industry Business Performance Database and the Directory of Overseas Investment Enterprises (institutions). The China Industry Business Performance Database provides an enormous amount of microdata at the enterprise level; however, it also suffers from problems such as fuzzy definitions of variables, missing and omitted indicators, and difficulties in identifying sample codes. Notably, abnormal samples must first be eliminated before the actual matching and combination. Similarly to the method proposed by Yu [38], the following approaches were employed to perform the primitive processing of the China Industry Business Performance Database: (1) duplicate or incorrect records were removed; (2) samples that omitted important indicators, such as the net value of fixed assets and gross industrial output value, were eliminated; (3) samples of smaller-scale enterprises, such as those with fewer than eight employees, were eliminated; (4) samples wherein the enterprise codes could not be identified were eliminated; and (5) according to the Generally Accepted Accounting Principles, samples with fixed assets larger than total assets and enterprises with current assets larger than total assets were removed.

As the enterprise names in the Directory of Overseas Investment Enterprises (Institutions) and the China Industry Business Performance Database do not correspond, a careful matching of enterprise names in the two databases was performed to generate a matched result that included as much sample data as possible. This approach was similar to that employed by Tian and Yu [39]. First, matching was performed with enterprise names. If the same enterprise name appeared in the two databases, it was considered to be the same enterprise. Second, because both databases had missing enterprise names, some enterprises were not successfully matched. For the samples with missing enterprise names, the postal codes of the enterprises’ locations and their phone numbers were utilized as enterprise identities. Finally, the matched results were tested with the year of enterprise founding. If the same enterprise name appeared in the two databases concurrently in the same year, the matching was considered to be successful.

Moreover, additional indicators were required to measure product quality. Based on the aforementioned matched data, the China Customs Import and Export Statistics Database was introduced to refine the data to the level of the Harmonized System (HS) six-digit commodity code (for the detailed process of combining the China Industry Business Performance Database and China Customs Import and Export Statistics Database, please refer to Tian and Yu [38]). The export product quality was mainly derived from weighted calculations of export quality at the product level. Before calculating the export product quality at the enterprise level, an estimation of the export product quality at the product level should first be conducted, followed by the corresponding normalization of the export product quality at the product level and aggregation to obtain the export product quality at the enterprise level.

Table 1. Descriptive statistics.

Variable	Mean	Std. dev	Min	Max
OFDI	0.002	0.048	0.000	1.000
Quality	0.728	0.112	0.367	0.897
Export	0.135	0.247	0.000	0.681
Scale	10.105	1.423	4.537	20.168
Kl	3.851	1.322	−6.683	14.371
Age	10.158	10.109	0.000	407.000
State	0.045	0.189	0.000	1.000
Tfp	4.251	0.945	−5.681	9.627

displays the descriptive statistics of variables.

3.3. Estimation Method

First, the Lagrange multiplier test results rejected the hypothesis that individual random effects do not exist. In the empirical testing, the regression was run using a random-effects model. Second, a mixed ordinary least squares model was utilized to estimate the effect of FDI on export trade. From the results of the White test, Breusch–Pagen test, and Durbin–Watson test, heteroscedasticity and autocorrelation exist in the regression model. To overcome the between-group heteroscedasticities and contemporaneous correlations, feasible general least squares was employed to perform regression fitting. Additionally, data matching was performed according to prior information during data processing, which provided better control of the endogeneity caused by bilateral causality. The calculation process controlled for fixed effects, such as individual and time, which relieved the endogeneity due to omitted variables to a certain extent. Moreover, a propensity score matching (PSM) analysis was conducted prior to the regression analysis. From the results, the difference between the treatment group and the control group was 0.157, and it passed the t-test. In the PSM analysis, a covariate imbalance test was also conducted. The PSM matching results balanced the data well, and the deviation of the standardized coefficient of the selected control variables was significantly reduced after matching, which showed that the sample’s propensity score match was ideal and suitable for subsequent empirical analysis.

4. Results

4.1. Product Quality Enhancement and its Contributions to the Export Effect

Table 2. Tests of the effect of OFDI on product quality.

Variable	(1)	(2)	(3)	(4)
	Product Quality	Enterprise Productivity	Product Quality	Export
OFDI	0.101 ***	0.090 ***	0.0705 **	0.177 ***
	(0.028)	(0.026)	(0.029)	(0.064)
Scale	0.215 ***	0.213 ***	0.148 ***	0.351 ***
	(0.000)	(0.001)	(0.000)	(0.002)
Kl	−0.001 ***	−0.122 ***	−0.0003 ***	−0.0004 ***
	(0.000)	(0.001)	(0.000)	(0.000)
Age	−0.005 ***	−0.003 ***	−0.004 ***	−0.010 ***
	(0.000)	(0.000)	(0.000)	(0.0002)
State	−0.106 ***	−0.489 ***	−0.147 ***	−0.107 ***
	(0.003)	(0.010)	(0.016)	(0.003)
Tfp			0.239 ***
			(0.001)
OFDI × tfp			0.123 ***
			(0.008)
Quality				1.845 ***
				(0.008)
OFDI × quality				−0.0194 *
				(0.011)
Constant term	−0.278 ***	2.511 ***	−0.637 ***	3.308 ***
	(0.005)	(0.012)	(0.006)	(0.012)
Industry	Controlled	Controlled	Controlled	Controlled
Region	Controlled	Controlled	Controlled	Controlled
Year	Controlled	Controlled	Controlled	Controlled
Observed value	5,557,116	4,058,898	4,058,898	4,058,898
R-squared	0.197	0.086	0.137	0.232

Remarks: (1) ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. The values in parentheses are the standard deviations of the regression in all tables. (2) R² is within R-squared in all tables.

reports the estimation results for Models (1)–(4). Column (1) reports the direct effect of OFDI on export product quality, where the parametric regression result of OFDI is significant and positive, indicating that OFDI promotes product quality directly. In the context of China’s “Go Out” policy, OFDI creates reverse pressure for enterprises to engage in R&D and upgrading, which, in turn, promotes the technical content of products and product quality enhancement to expand development and competitive advantages in the international market. Column (2) shows the effect of OFDI on enterprise productivity: the parametric regression result of ODFI is significant and positive, specifying that OFDI significantly promotes enterprise productivity. The results are consistent with the findings of Jiang and Jiang [12] and Seyoum et al. [18] and match the theoretical expectations of game theory’s Cournot competition model. Improvements in enterprise productivity help to further improve product quality; therefore, Hypothesis 1 is verified.

The regression results of the control variables meet expectations. Enterprise scale has a positive impact on product quality, suggesting that with an expansion in enterprise scale, product quality continues to increase, showing the economies of scale effect. According to the theory of the optimal production of enterprises, in the early stage of enterprise development, owing to the small scale of enterprises, the inputs of various production factors did not reach the optimal production of enterprises (i.e., the production cost of enterprises is not the lowest). With a gradual expansion in the enterprise production scale, the factor ratio improved towards the Pareto optimal direction, the production cost gradually decreased, and enterprises gradually showed the advantage of economies of scale in the international market. The capital intensity coefficient is significantly negative, highlighting that enterprises with the fastest improvement in product quality are not necessarily those with higher capital intensity. Currently, a large proportion of China’s foreign trade enterprises are engaged in processing trade. Such enterprises process and assemble products and invest less in the production of large machinery and high-tech products. The business life coefficients of the enterprises are significantly negative. The duration for which Chinese enterprises typically engage in foreign trade is often short and lacks continuity, which harms the improvement in product quality. The coefficient of state-owned enterprises is significantly negative, which is related to the low participation of state-owned enterprises in foreign trade. In reality, the participation of Chinese state-owned enterprises in the international market is low. Hence, private- and foreign-funded enterprises have a more important role in foreign trade. As the export commodity structure of state-owned enterprises is relatively similar, when facing international market competition, they lack the flexibility to realize the differentiation strategy adjustment, which easily leads to a reduction in the scope of export products. Moreover, state-owned enterprises enjoy preferential policies and better credit mechanisms, and their technology R&D levels are also at a greater level. In comparison to making higher fixed investments to penetrate the international market, state-owned enterprises can also achieve greater profits in domestic sales.

To further analyze how enterprise productivity affects export product quality and how product quality enhancement contributes to the export effect of FDI, columns (3) and (4) report the findings of the estimations to Models (3) and (4). To control for endogeneity between variables and to solve the problem that a single equation ignores the relations between equations, a two-stage least squares analysis was conducted for the estimation. In column (3), the coefficient of the interaction term of OFDI and tfp is significant and positive at the 5% level of significance, which indicates that, as the productivity level increases, the promotional effect of OFDI on product quality enhancement increases, and the promotional effect of OFDI on product quality relies on productivity enhancement. In other words, the promotional effect of OFDI on productivity contributes to a further enhancement in firms’ export product quality. These outcomes match the motivations of OFDI, where OFDI can be considered to be a dynamic mechanism in China to promote product quality. Overall, when the parametric regression result of OFDI is also considered, enterprises’ OFDI leads to both direct and indirect promotional effects on export product quality enhancement through productivity.

In column (4), the parametric regression results for quality are significant and positive at the 1% level of significance, specifying that product quality enhancement contributes to the promotion of exports. Additionally, the parametric coefficient of the interaction term of OFDI and tfp is significant and negative at the 10% significance level, which means that as export product quality increases, the promotional effect of OFDI on exports decreases. As export product quality increases, the reverse technology spillover effects of OFDI gradually decrease. In this situation, if enterprises hope to further increase their export product quality, they will have to increase their R&D input into their products and enhance their self-dependent innovation abilities. Insufficient levels of innovation may hinder the further promotion of product quality, which, in turn, suppresses exports to a certain extent. However, continuous product quality enhancement strengthens the competitive advantages of enterprises in the domestic market, which increases the probability that enterprises will choose domestic sales and decrease their exports. This reasoning corresponds to the export characteristics of Chinese enterprises, namely that low productivity enterprises choose to export and high productivity enterprises choose to sell domestically. This finding illustrates the paradox of Chinese enterprises’ exports.

4.2. Heterogeneity Analysis

To clarify the differences in the relationships between FDI, product quality, and export trade, enterprises were divided into low productivity enterprises and high productivity enterprises according to their mean value of productivity. New estimations were performed for Models (3) and (4), and the regression results are presented in

Table 3. Regression results of enterprises by productivity level.

Variable	High Productivity Enterprises		Low Productivity Enterprises
	(5)	(6)	(7)	(8)
	Product Quality	Export	Product Quality	Export
OFDI	0.818 ***	−0.066	0.445 ***	0.677 ***
	(0.172)	(0.053)	(0.0739)	(0.203)
Quality		0.764 ***		3.905 ***
		(0.004)		(0.063)
OFDI × quality		−0.034 ***		0.0713 ***
		(0.010)		(0.0168)
Scale	0.164 ***	0.523 ***	0.0614 ***	0.369 ***
	(0.0009)	(0.0010)	(0.0009)	(0.0058)
Kl	−0.0005 ***	−0.0004 ***	0.0004 ***	−0.0033 ***
	(0.000)	(0.000)	(0.000)	(0.000)
Age	−0.0043 ***	−0.0146 ***	0.0007 ***	−0.0191 ***
	(0.0001)	(0.0002)	(0.0002)	(0.0006)
State	−0.159 ***	−0.138 ***	−0.134 ***	−0.141 ***
	(0.031)	(0.017)	(0.033)	(0.034)
Tfp	0.109 ***		0.498 ***
	(0.0017)		(0.0021)
OFDI × tfp	0.049 ***		0.168 ***
	(0.0146)		(0.0273)
Constant term	−2.387 ***	3.845 ***	0.679 ***	−0.425 ***
	(0.0114)	(0.0098)	(0.0107)	(0.0750)
Industry	Controlled	Controlled	Controlled	Controlled
Region	Controlled	Controlled	Controlled	Controlled
Year	Controlled	Controlled	Controlled	Controlled
Observed value	2,117,038	2,117,038	1,941,860	1,941,860
R-squared	0.073	0.282	0.008	0.055

Remarks: (1) *** represents significance at the 10% level. The values in parentheses are the standard deviations of the regression in all tables. (2) R² is within R-squared in all tables.

.

The resulting coefficients of OFDI in columns (5) and (7) are significant and positive at the 1% significance level. This result outlines that regardless of whether firms have low or high productivity, OFDI brings about an indirect promotional effect on product quality. Therefore, it leads to product quality enhancement through the productivity effect of OFDI. These analyses further proved the robustness of the conclusions presented in Table 2. However, when the magnitudes of the coefficients are considered, the coefficient of the interaction term of OFDI and tfp in column (5) is 0.049 and the coefficient of the interaction term of OFDI and tfp in columns (7) is 0.168, suggesting a more significant promotional effect of OFDI’s productivity effect on product quality for low productivity enterprises. Technology sourcing is a major source of motivation for Chinese enterprises to perform OFDI. In particular, it is easier to obtain the host country’s advanced production experience and technology by investing in R&D-intensive industries in host countries. Compared to high productivity enterprises, low productivity enterprises have a high potential for growth in this aspect and can enhance their product quality more rapidly.

Furthermore, the parametric regression results for quality in columns (6) and (8) are significant and positive at the 1% level of significance, relaying that regardless of whether the firm has low or high productivity, product quality enhancement contributes to the promotion of exports. However, regarding the promotional effect of product quality enhancement on exports, a degree of heterogeneity exists between enterprises. For low productivity enterprises, the promotional effect on exports induced by product quality enhancement is significantly stronger than for high productivity enterprises (3.905 > 0.764). The parametric regression results for OFDI in column (6) are negative but insignificant. The parametric regression results of OFDI in column (8) are significant and positive, specifying that OFDI has negative effects on the exports of high productivity enterprises to a certain extent and has significant positive effects on the exports of low productivity enterprises. Likewise, the coefficient of the interaction term of OFDI and product quality in column (6) is significant and negative at the 1% significance level, while the coefficient of the interaction term of OFDI and product quality in column (8) is significant and positive at the 1% significance level. This outcome denotes that for high productivity enterprises, as the firm’s product quality increases, OFDI suppresses the increase in export trade, and for low productivity enterprises, as product quality improves, the promotional effect of OFDI on exports increases.

The enhancement effects of product quality and product quality enhancement have significant differences in their contributions to the export effect of OFDI. Ranking enterprise productivity from low to high, the promotional effect of OFDI on export trade, along with product quality enhancement, increases at first and decreases thereafter, thereby exhibiting an inverted-U changing trend. These conclusions indicate that an export paradox exists among Chinese enterprises.

4.3. Further Analysis

The previous analysis illustrates the paradox in the exports of Chinese enterprises, which conveys that an enhancement in enterprise productivity levels suppresses enterprise exports to a certain degree, consistent with the findings of Li [40]. The contrast demonstrated by the relationship between Chinese enterprises’ exports and productivity has attracted strong interest from scholars, and relevant studies are gradually emerging. For the “productivity paradox” demonstrated by the exports of Chinese enterprises, one possible explanation is that lower productivity enterprises are subject to higher costs in domestic sales, which can be higher than the export cost. To avoid competitive pressure from fellow domestic enterprises, such enterprises choose to export. In contrast, with technological advantages, higher productivity enterprises have better positions in domestic sales and prefer domestic sales for greater profits rather than enduring the higher fixed input and export costs required for in-depth development in the international market.

To deepen the understanding of this idea, based on Model (4), an interaction term of OFDI and tfp (OFDI × tfp) was introduced, which replaced the interaction term of OFDI and export product quality (OFDI × quality). According to the end-use and factor intensity of the products, the export products are divided into six categories: intermediate products, consumer goods, capital goods (since the commodity code in the China Industry Business Performance Database is International Standard Industrial Classification of all Economic Activities (ISIC) Rev4, which cannot be converted directly into the Broad Economic Category (BEC) codes that categorize commodity types, the ISIC Rev4 codes are first converted into Standard International Trade Classification codes, which are then converted into BEC codes. According to the BEC classification standard, the commodities in the sample database are divided into intermediate goods, consumer goods, and capital goods), labor-intensive, capital-intensive, and technology-intensive products. For further observation of the impact of enterprise productivity on the export effect of OFDI (following Jiang et al. [41]), fourteen industries are classified as labor-intensive (food processing; food manufacturing; beverage manufacturing; tobacco processing; textile; garments and other fiber products; leather, furs, down, and related products; timber processing; bamboo, cane, palm fiber, and straw products; furniture manufacturing; papermaking and paper products; printing and recorded media; cultural, educational, and sporting goods; rubber products; and plastic products), eight industries are classified as capital-intensive (petroleum processing and cooking; processing of nuclear fuel; manufacturing of non-metallic mineral products; smelting and processing of ferrous metals; smelting and processing of non-ferrous metals; manufacturing of metal products; manufacturing of general purpose machinery; and manufacturing of special purpose machinery, instruments, meters, and cultural and office equipment), and six industries are classified as technology-intensive (medical and pharmaceutical products; manufacturing of chemical raw materials and chemical products; manufacturing of chemical fibers; transport equipment; manufacturing of electrical machinery and equipment; manufacturing of communication equipment; and computers and other electronic equipment).

Table 4. Tests of the export productivity paradox.

Variable	(9)	(10)	(11)	(12)	(13)	(14)	(15)
	Overall	Intermediate Product	Consumer Goods	Capital Goods	Labor-Intensive	Capital-Intensive	Technology-Intensive
OFDI	0.104 ***	0.056	0.130 *	0.167	0.114 ***	0.138 ***	0.067 *
	(0.020)	(0.048)	(0.071)	(0.135)	(0.037)	(0.036)	(0.036)
OFDI × tfp	−0.015 ***	−0.004	−0.024	−0.036	−0.018 **	−0.023 ***	−0.006
	(0.004)	(0.010)	(0.016)	(0.029)	(0.008)	(0.008)	(0.007)
Constant term	0.090 ***	0.558 ***	0.708 ***	0.452 ***	0.123 ***	0.041 ***	−0.047 ***
	(0.003)	(0.012)	(0.016)	(0.029)	(0.006)	(0.005)	(0.006)
Controlled variable	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Industry	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Region	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Year	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Observed value	3,564,040	496,402	321,286	36,875	1,304,473	961,594	1,023,482
R-squared	0.133	0.112	0.105	0.067	0.098	0.085	0.093

Remarks: (1) ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. The values in parentheses are the standard deviations of the regression in all tables. (2) R² is within R-squared in all tables.

presents the regression results. In column (9), wherein a population regression is run, the coefficient for the interaction term exceeds the statistical significance level of 1%. In columns (13) and (14) for labor- and capital-intensive products, respectively, the coefficients of the interaction terms exceed the 5% and 1% levels of significance, respectively. In the other regression equations, the coefficients of the interaction term were negative but insignificant. This finding specifies that enterprise productivity enhancement does indeed weaken the export effect of OFDI, and the suppression is more obvious for exports of labor- and capital-intensive products. This outcome further confirms the existence of the export productivity paradox.

5. Conclusions

Chinese enterprises’ OFDI brings about overall advantages to exports, but the promotional effect of OFDI on exports decreases as the level of productivity increases and shows significant disparities at different levels. Specifically, OFDI significantly promotes exports of low productivity enterprises and has no significant effect on high productivity enterprises. The relationship between Chinese enterprises’ OFDI and productivity contradicts the expectations of the theories of heterogeneous firms and trade and proves that a productivity paradox exists in the exports of Chinese enterprises.

Moreover, OFDI enhances product quality. The expansion of OFDI creates reverse pressure on enterprises to engage in transformation and upgrades, which promotes product quality enhancement to a certain extent. However, OFDI produces a significant productivity effect and diffuses the advanced production technology of the host country back into the home country through reverse technological learning. This effect contributes to the promotion of the technical content of products and enhances product quality. Additionally, the promotional effect on product quality induced by the OFDI productivity effect shows a significant disparity among enterprises. The promotion effect on product quality induced by the productivity effect of OFDI is significantly stronger for low productivity enterprises than for high productivity enterprises.

Notably, quality enhancement undermines the promotional effect of OFDI on exports to a certain degree. When considering the level of firms, product quality enhancement significantly strengthens the promotional effect of OFDI on the exports of low productivity enterprises and significantly suppresses the promotional effect of OFDI on the exports of high productivity enterprises. This result also points to the existence of an export paradox in Chinese enterprises.

This study further clarifies the effect of Chinese enterprises’ OFDI on exports and simultaneously tests and verifies the promotional effect of OFDI on product quality. Therefore, encouraging enterprises to “go out” and increase OFDI helps to elevate China’s position in the global value chain. In the context of increasing economic uncertainty, China should continue to expand the space for international cooperation and encourage enterprises to invest overseas. Enterprises facing export pressure should actively seek opportunities for international investment and improve their export scales through flexible investment strategies. In a complex international situation, future research could focus on the sustainability of OFDI and product quality improvement.