Introduction

The markup of a firm, which is typically calculated as the price of a good or service divided by its cost, is a key indicator of the firm’s pricing and market power, and the level of the markup has a direct bearing on the firm’s competitiveness in the international market and welfare gains along the global value chain (Peters, 2020; De Loecker and Goldberg, 2014; Edmond et al. 2015). For a long time, Chinese exports have been labeled as the “three lows” of low price, low quality, and low profit, the main reason being the low markups of export firms (Sheng and Wang, 2012). According to the classic trade theory of heterogeneous firms, only firms with high productivity can enter the international market to export because of the high fixed costs of entering export markets, and exporters have higher markups than non-exporters (Melitz and Ottaviano, 2008). However, inconsistent with theoretical expectations, studies based on the micro data of Chinese firms showed that Chinese exporters are confronted with a typical “low markup trap” (Zhang and Zhu, 2017). Chinese exporters have lower markups than domestic producers, which can not only reduce the profit level of exporters and affect their long-term sustainable development but also increase the risk of foreign antidumping and lead to a series of problems such as insufficient domestic consumption and resource and environmental conflicts. In response to the “low markup trap” of Chinese exporters, scholars explained it from the reality and policy levels and generally attributed it to the unique characteristics of China’s export trade. On the one hand, extensive processing trade firms exist among Chinese exporters, and such processing trade firms do not have export pricing power. On the other hand, China’s long-standing export tax rebate and subsidy policies and the excessive competition in export firm industries exacerbated the situation of Chinese exporters competing to reduce prices (Sheng and Wang, 2012).

However, it should be noted that paying too much attention to the causes of the “low markup trap” of Chinese exporters may not be meaningful. At present, China is in a critical period of further deepening reform and opening up, and improving the competitiveness of enterprises in the international market and obtaining the substantial benefits of opening up should be the key to creating an upgraded version of China’s opening up to the outside world. Thus, it’s crucial to fully grasp and comprehend how export markets affect the markups of firms. However, relevant studies basically remained in the stage of exploring the impact of transformation, from anon-export to an export status, on firms’ markups (their focus remained on a comparison between export and non-export firms), which may not be a comprehensive perspective. The entry of firms into export markets is only the beginning, and exploring whether the rich export experience accumulated by firms in the export market can contribute to the increase in their markups as their export duration increases is highly important. Meanwhile, not all firms will enter export markets. Therefore, to grasp the impact of export markets on firms’ markups accurately, exploring the impact of the transition from a non-export to an export status on firms’ markups may be inadequate, and further exploring the impact of firms’ export duration on their markups by focusing on those that have entered export markets is necessary. However, research in this area at this stage is lacking. This study’s focus is to export enterprises that have forayed into foreign marketplaces. This study theoretically and empirically examines how to export duration influences firm markups using micro data from Chinese manufacturing firms, offering a fresh viewpoint for a deeply thorough understanding of how export markets affect firm markups.

Two literature categories are important to this study. The first category is the literature on export duration. Export duration measures the degree of firm participation in export markets from the time dimension, which is a micro-level reflection of the dynamic behavior of firm export. Shao et al. (2012) suggested that export duration, which is a concentrated reflection of a country’s comprehensive international competitiveness and ability to cope with external shocks, can effectively reflect the viability of the country’s enterprises in the international market. Early research in this area focused on the measurement of firms’ export duration (Besedeš and Prusa, 2006; Volpe Martincus and Carballo, 2009; Besedeš and Nair-Reichert, 2009; Esteve-Pérez et al. 2013; Socrates et al. 2020). In recent years, an increasing number of studies began to focus on the factors affecting the export duration of firms (or products). Valderrama et al. (2013) examined whether information externalities resulting from various levels of spatial interaction assist new exporters in extending the duration of their trade activity using transaction-level data on Colombian exports between 2004 and 2011. Zhou et al. (2019) found that the significant reduction in import tariffs following China’s accession to the World Trade Organization (WTO) increased the export duration of firm products. The research was based on data from Chinese industrial enterprises, customs import and export, and tariff data for the years 2000 to 2006. Kostevc and Zajc Kejžar (2020) found that a firm’s bilateral inward and outward FDI flow to and from the export destination country has a considerable beneficial impact on the firm’s export survival in the market based on data of enterprises in Slovenia for the period of 2002–2011. Other studies explored its impact on the export duration of firms from multiple perspectives, such as innovation (Chen, 2012), manufacturing servitization (Cui and Liu, 2018), and export tax rebate (Anwar et al. 2019). The pertinent research that has been done in the past, however, has mostly explored the elements that determine a firm’s export duration. As the export duration is an important reflection of a company’s viability in the international market, few studies analyzed how export duration affects the relevant performance of enterprises. Relevant research perspectives need to be further expanded.

The second literature category concerns the impact of exports on firms’ markups. Within the framework of an imperfect competition model, Bernard et al. (2003) examined the interactions between firms’ export behavior, productivity, and markups and demonstrated that firms with high productivity are likely to export and are highly capable of setting high price markups due to their cost advantages. Melitz and Ottaviano (2008), on the other hand, internalized firms’ markups in a model by proposing a linear demand system with product-level differences and theoretically demonstrated that only high-productivity firms can overcome the fixed costs of entering export markets; thus, exporters have higher levels of markups than non-exporters. At the same time, exporters’ markups are positively correlated and increase with their export intensity. In terms of empirical research, most studies focused on the impact of firms’ export behavior on their markups. Görg and Warzynski (2003) examined whether export behavior affects the markups of firms in the UK manufacturing industry and found that exporting significantly increases the markups of firms in the differentiated product industry. Subsequently, the research by Kugler and Verhoogen (2008) on Colombian enterprises, De Loecker and Warzynski (2012) on Slovenian firms, Kato (2014) on Japanese manufacturing firms, and Bellone et al. (2016) on French firms all demonstrated that export firms have greater levels of markups than non-export firms.

Contrary to such findings, based on the micro-level data of Hungarian enterprises between 1995 and 2003, Hornok and Muraközy (2019) found that exports do not significantly affect firms’ markups when controlling for imports. Studies based on Chinese firm-level data also suggested a “low markup trap” for Chinese exporters, that is, exporters have smaller markups than non-exporters do (Sheng and Wang, 2012; Zhang and Zhu, 2017). For this reason, some studies further analyzed the micro mechanisms through which exporting affects firms’ markups in terms of innovation (Máñez et al. 2022) and market competition (Garcia-Marin and Voigtländer, 2019; Caselli and Schiavo, 2020). In addition to examining how firms’ export behavior affects their markups, other research evaluated how firms’ export patterns (Feenstra and Hanson, 2004; Yang, 2021) and choice of export destination countries (Manova and Zhang, 2009; Kilinç, 2019; Yang, 2021) affect markups. Although there is a wealth of research on how exporting affects firms’ markups, the majority of these studies concentrated on how export behavior affects the markups of enterprises, and very few examined the influence from the perspectives of export mode and export destination country choice. There isn’t enough research specifically looking at how export duration affects the markups of firms.

Compared with the literature, this study contributes mainly to several aspects. First off, this study focuses on export firms that have entered the export market and explore the impact of the accumulation of export experience on the firm markups from the perspective of export duration, in contrast to the literature, which primarily analyzed the impact of exporting on the markups of export firms, with a comparison of the effect on non-export firms. This study can not only enrich our understanding of the “low markup trap” of Chinese exporters but also provide a new perspective for a highly comprehensive understanding of the impact of export markets on the markups of firms, which would be a useful addition to the literature. Second, under the framework of firm heterogeneity, this study establishes a theoretical model to investigate how export duration affects firms’ markups and theoretically clarifies the mechanism of export duration impacting firms’ markups through the “production efficiency” channel and “market-based pricing” channel. Thus, it can provide a theoretical basis for a deep understanding of how export duration affects the micro performance of firms, especially markups. Third, this paper provides a thorough empirical investigation of the effect of export duration on firm markups and its mechanism using micro data from Chinese manufacturing enterprises. This study confirms that an increase in export duration significantly increases the markups of Chinese firms after carrying out a number of robustness tests, including key indicator substitution, endogeneity treatment, and controlling for other policy changes. This study thus provides fresh empirical proof from large developing countries for further understanding and grasping the effect of export markets on firms’ markups. Fourth, this study explores the effects of export duration on the markups of various firm types by industry, ownership type, region, size, export intensity, and trade mode and provides corresponding policy insights to effectively promote the participation and operation of the different types of firms in export markets.

The rest of this paper is organized as follows: Section 2 creates a model framework for examining how firms’ markups are affected by the duration of their exports, Section 3 sets up the empirical framework, and Section 4 analyzes and discusses the empirical results. The heterogeneous effects of export duration on the markups of various types of firms are further examined in Section 5, and finally, this paper concludes in Section 6.

Theoretical analysis framework

In this section, we establish a model framework to analyze how export duration affects firms’ markups. For simplicity, we consider two symmetric economies, namely, domestic and foreign, that engage in trading and firms that manufacture products only for exportFootnote 1 The domestic export firms have entered the foreign export market and sold their product.

Consumers’ preference

For a continuous commodity consumption bundle Ω, assuming that the representative consumer has a utility function of \(U = \left[{\int_{\omega \in \Omega }} {q(\omega )} ^\rho d\omega \right]^{\frac{1}{\rho }}\), where q(ω) is the quantity of the commodity ω consumed, the elasticity of substitution between any two commodities is σ = 1/(1-ρ) > 1. By constructing the price index as \(P = \left[{\int_{\omega \in \Omega }} {p\left( \omega \right)} ^{1 - \sigma }\right]^{\frac{1}{{1 - \sigma }}}\) and the quantity index as \(Q = \left[ {{\int_{\omega \in \Omega }} {q\left( \omega \right)^\rho d\omega } } \right]^{\frac{1}{\rho }}\), where p(ω) is the price of commodity ω faced by consumers and according to consumer utility maximization, the demand function of commodity ω is determined as \(q\left( \omega \right) = Q\left[ {\frac{{p\left( \omega \right)}}{P}} \right]^{ - \sigma }\).

Firms’ production decisions

We make the assumption, in line with Melitz (2003), that each firm produces a differentiated product ω. A firm uses labor and capital for production and has the following production function:

$$q\left( \varphi \right) = \varphi \left( \chi \right)K^{1 - \alpha }L^\alpha$$
(1)

where K and L stand for the production’s inputs of labor and capital, respectively, and φ(χ) denotes the productivity of the firm.

We assume that a firm’s productivity is a function of its export duration χ. Generally, firms may face intense international competition in the export market and can obtain access to the latest product design, the most advanced production technology and management mode, and so on. Firm productivity improvement can be promoted directly or indirectly through the “learning by exporting” effect (Greenaway and Kneller, 2007; De Loecker, 2013). The longer the export duration, the richer the export experience accumulated by the firm, and the greater the “learning by exporting” effect. Yang and Mallick (2010) found that firms have improved production efficiency after entering the export market, especially in the second year after entering. To this end, we further assume that a firm’s productivity increases with the extension of its export duration, that is, \(\partial \varphi \left( \chi \right)/\partial \chi > 0\).

Assume that the factor markets are all perfectly competitive, and every firm is a price taker in the factor markets. Let W and R stand for labor and capital prices, correspondingly. A firm’s cost minimization implies the following function of the firm’s marginal cost:

$$mc\left( \chi \right) = \frac{{\alpha ^{ - \alpha }\left( {1 - \alpha } \right)^{\alpha - 1}R_{}^{1 - \alpha }W^\alpha }}{{\varphi \left( \chi \right)}}.$$
(2)

Suppose that when a firm sells products to foreign consumers, it will face not only variable trade costs γ (in the form of iceberg costs, meaning that units of export commodities must be loaded onto a ship in order for one unit to reach a foreign market γ) but also distribution costs η(χ). Some empirical studies assumed the introduction of distribution costs. For instance, Feenstra (1998) pointed out that transportation costs and distribution costs account for nearly 90% of the retail price of Barbie dolls exported from Asia to the United States. Burstein et al. (2003) found that distribution costs account for as high as 40% of the average retail price in the United States and as high as 60% of that in Argentina. Generally, after entering the export market, firms will face relatively high distribution costs owing to their lack of understanding of export market regulations and the sales environment. However, as the export duration increases, firms will become increasingly familiar with the export market environment, and their rich accumulated export experience will undoubtedly give them a considerable advantage in the establishment of marketing networks and the discovery of potential customers. For example, Rakhman (2010) found that firms with a long export duration are highly capable of adjusting their business modes to changes in market demand. Therefore, we further assume that a firm’s distribution costs will decrease as its export duration increases, that is, \(\partial \eta \left( \chi \right)/\partial \chi < 0\).

According to the aforementioned settings, for every unit of product manufactured by a firm, the actual price faced by foreign consumers will be

$$p = \gamma p_0 + \eta \left( \chi \right),$$
(3)

where P0 is the firm’s ex-factory price.

On the basis of Eq. (3) and consumers’ preference, when a firm’s ex-factory price is P0, the demand function of foreign consumers for the firm’s products is \(q = Q\left[ {\frac{{\gamma p_0 + \eta \left( \chi \right)}}{P}} \right]^{ - \sigma }\), from which we easily obtain the profit function of the firm in the export market.

$$\pi \left( \chi \right){{{\mathrm{ = }}}}\left( {p_0 - \frac{{\alpha ^{ - \alpha }\left( {1 - \alpha } \right)^{\alpha - 1}R_{}^{1 - \alpha }W^\alpha }}{{\varphi \left( \chi \right)}}} \right)Q\left[ {\frac{{\gamma p_0 + \eta \left( \chi \right)}}{P}} \right]^{ - \sigma }.$$
(4)

We take the first derivative of Eq. (4) and obtain the optimal p0 under the profit maximization, as follows:

$$p_0\left( \chi \right){{{\mathrm{ = }}}}\frac{\sigma }{{\sigma - 1}}\left( {\frac{{\alpha ^{ - \alpha }\left( {1 - \alpha } \right)^{\alpha - 1}R_{}^{1 - \alpha }W^\alpha }}{{\varphi \left( \chi \right)}} + \frac{{\eta \left( \chi \right)}}{{\sigma \gamma }}} \right).$$
(5)

By defining the markup of a firm with an export duration χ (denoted as markup(χ)) as the ratio of its ex-factory price to its marginal cost, that is, \(p_0\left( \chi \right) = mc\left( \chi \right)markup\left( \chi \right)\), and further combining Eqs. (2) and (5), the markup of the firm can be easily obtained as:

$$markup\left( \chi \right){{{\mathrm{ = }}}}\frac{\sigma }{{\sigma - 1}}\left( {\frac{{\varphi \left( \chi \right)\eta \left( \chi \right)}}{{\sigma \gamma \alpha ^{ - \alpha }\left( {1 - \alpha } \right)^{\alpha - 1}R_{}^{1 - \alpha }W^\alpha }} + 1} \right).$$
(6)

Equation (6) shows that with the increase in the export duration of a firm, the firm’s markup may increase or decrease owing to \(\partial \varphi \left( \chi \right)/\partial \chi > 0\) and \(\partial \eta \left( \chi \right)/\partial \chi < 0\). On the one hand, as the export duration of a firm increases, firms can obtain more productivity improvement through “learning by exporting”. An increase in the productivity of a firm is accompanied by an reduction in the marginal cost of firm production (see Eq. [2]), thereby improving the firm’s markup. We note how the export duration affects a firm’s markups as the “production efficiency” channel. On the other, as χ increases, the distribution costs of a firm in the export market will decrease accordingly, which will reduce the ex-factory price of the firm’s products (see Eq. [5])Footnote 2, thereby decreasing the firm’s markup. We note how the export duration affects the markup as the “market-based pricing” channel. Therefore, under our established analytical framework, we propose the following hypotheses:

Hypothesis 1: The impact of the increase in export duration on firms’ markups is uncertain, that is, it may be positive or negative.

Hypothesis 2: An increase in export duration may increase firms’ markups via the “production efficiency” channel or reduce the markups via the “market-based pricing” channel.

Empirical analysis framework

Econometric model

We create the following baseline regression model, based on Lu and Yu (2015), to empirically investigate the effect of export duration on the markups of Chinese manufacturing enterprises.

$$lmp_{it} = \beta _1Export\_T_{it} + \delta X_{it} + \lambda _i + \upsilon _t + \mu _{it},$$
(7)

where i represents the firm; t represents the year; impit is the logarithm of the markup of firm i in period t; Export_Tit is the core independent variable of interest, which measures the length of firm i’s export duration in period t; Xit represents a set of other control variables; λ1 and υt represent firm fixed effects and year fixed effects, respectively; and μit is the random error term.

We first define markup as the proportion of the product price to the marginal cost before calculating it for a firm. Referring to De Loecker and Warzynski (2012), we know that markup μit of firm i in period t has the following form:

$$\mu _{it} = \theta _{it}^v\left( {\alpha _{it}^v} \right)^{ - 1},$$
(8)

where \(\alpha _{it}^v\) is the proportion of the total expenditure of variable input v in the total sales revenue of the firm, which can generally be calculated directly from the firm’s production data, and \(\theta _{it}^v\) is the output elasticity of variable input v in the firm’s production, which can be obtained by estimating the firm’s production function.

To estimate the subsequent translog production function, we use the semi-parametric estimation method (i.e., ACF) created by Ackerberg et al. (2015):

$$\begin{array}{l}y_{it} = \beta _ll_{it} + \beta _kk_{it} + \beta _mm_{it} + \beta _{ll}l_{it}^2 + \beta _{kk}k_{it}^2 + \beta _{mm}m_{it}^2 + \beta _{lk}l_{it}k_{it}\\ \qquad\quad+ \,\beta _{lm}l_{it}m_{it} + \beta _{km}k_{it}m_{it} + \beta _{lkm}l_{it}k_{it}m_{it} + \omega _{it} + \varepsilon _{it}\end{array}$$
(9)

where yit is the log of the total output of firm i in period t; lit, mit, and kit represent the log of its labor input, intermediate input, and capital stock, respectively; ωit represents the total factor productivity of the firm, and εit is the random error term. To determine the markups of firms, we use the output elasticity of the intermediary inputs since labor is not a variable input in Chinese firms (Lu and Yu, 2015), whereas capital is a dynamic input.

In this study, firm export duration (Export_Tit) is the core independent variable of concern in the empirical analysis. Drawing on Besedeš and Prusa (2006) and other related studies, we define the export duration of a firm as the period of time when the firm first enters the export market to the time when the firm exits the export market. However, when constructing a measure of a firm’s export duration, special attention must be paid to two issues. The first issue is that some firms may have multiple export duration periods. For example, if a firm enters a foreign export market in a certain year, it may stop exporting to the market after a certain period of time and then reenter the export market once again after a few years. In the relevant literature, this situation is referred to as “multiple spells”. According to Besedeš and Prusa (2006), though some firms may have multiple spells during a sample period, the approach of regarding the first spell of a firm as the only spell during the sample period will not have a substantial impact on the distribution of the firms’ spell length during the sample period. Therefore, only the first export duration period will count when calculating the export duration of a firm. The second issue is the possible existence of left truncation in the data. In this study, the data sample period is 2000–2007. Knowing the export status of some firms outside the sample period would be impossible. For example, for firms that engaged in exporting before 2000, their specific export duration cannot be determined, which will lead to an obvious underestimation of their export duration. To avoid interference from the left-truncated data on the subsequent empirical results, we remove the left-truncated observations in the subsequent regressions.

In Eq. (7), we add the following control variables based on Bellone et al. (2016) and Liu et al. (2019): ① firm size (Size), measured by the logarithm of firm sales; ② factor intensity (Kl), measured by the logarithm of the capital-labor ratio; ③ average wage (Wage), measured by the logarithm of the ratio of the total wages payable by the firm to the number of employees in the firm; ④ firm age (Age), measured by the logarithm of the difference between the current year and establishment year of a firm; and ⑤ government subsidies (Subsidy), measured by the ratio of the number of subsidies received by a firm to the sales of the firm.

Data

The Chinese Annual Survey of Industrial Firms (CASIF) database provided the data for the empirical study. We chose only the data of the manufacturing firms in the CASIF database from 2000 to 2007, which constitutes our research sample, keeping in mind the continuity of the pertinent indicators and the validity of the research findings. Referring to Feenstra et al. (2014), Yu (2015), Xiang et al. (2017), and other related studies, the original data of the Chinese industrial firms is first cleaned and screened, with sample firms lacking variables and having information contradicting accounting knowledge being deleted. Second, we use the customs data from 2000 to 2007 in the subsequent empirical analysis. Drawing on the literature, such as Yu (2015) and Xiang et al. (2017), we match the industrial enterprise data with the customs data by using the Chinese name of the firms and their zip code and telephone numbers.

We employ the ACF approach to estimate the firm production function by industry, which is in accordance with the two-digit industry code in the CASIF database, in order to accurately estimate the markups of the firms because the production technology may differ greatly among industries. We calculate the firm-level markups using Eq. (8) and the output elasticity of the intermediate inputs. The corresponding outcomes are displayed in Table 1. The mean value of the firms’ markups is within the range of 1.2–1.4 for all the industries. Table 2 further reports the descriptive statistics of the main variables used in the empirical analysis.

Table 1 Average markups and quantiles for each industry.
Full size table
Table 2 Descriptive statistics of main variables.
Full size table

Empirical results and discussion

Baseline regression results

Table 3 reports the estimation results of Eq. (7). In columns (1) and (2), we add only the firm fixed effects. The estimated coefficient of export duration is significantly positive regardless of whether or not the relevant control variables at the firm level are included, thereby indicating that as the firms’ export duration increases, their markups increase accordingly. We further control for the year fixed effects in columns (3) and (4), where we obtain similar estimated coefficients to those in columns (1) and (2). The estimated coefficients of export duration remain significantly positive, that is, the increase in export duration is conducive to the increase in the firms’ markups. Overall, we observe no substantial change in the sign and significance of the estimated coefficient of export duration regardless of whether or not the year fixed effects and firm-level control variables are added, which indicates that to a certain extent, the favorable effect of export duration on firms’ markups won’t alter as much as the control variable changes. The estimated coefficient of export duration, which is 0.0035 and passes the 1% significance test, means that the markup of a firm will rise by 0.35% for each extra year of export duration, according to the findings in column (4).

Table 3 Baseline regression results.
Full size table

The estimated coefficients in column (4) provide information on the control variables. The significant positive estimated coefficient of Size suggests that large enterprises will have high markups. In reality, large-scale firms tend to have monopoly power and can maintain relatively high levels of markups. The estimated coefficient of Kl is notably negative, showing that the markup of the firm decreases as the capital-labor ratio rises. The estimated coefficient of Wage is significantly positive, indicating that the markup of the firm increases as the average wage rises. A probable explanation for this outcome is that firms that provide high average wages can attract highly skilled employees, whose participation will contribute to the firms’ productivity improvement, which in turn will increase their markups. The estimated coefficients of Age and Subsidy do not pass the significance test, suggesting that the markups of the firms are not significantly impacted by firm age or government subsidies.

Robustness tests

Key indicator substitution

Previously, we estimated the firms’ markups based mainly on the framework of De Loecker and Warzynski (2012) by estimating the firm’s production function, which is also referred to as the production approach, as it relies mainly on the estimation of a firm’s production function. The accounting method, which primarily employs financial indicators to quantify firms’ markups including industrial value added, wage inputs, and cost of intermediate inputs, is another approach frequently employed in the literature to estimate firm markups. The accounting approach is easier to measure firm markups than the production method, and the requisite financial indicator data are more generally available and easier to get in related investigations. In this part, as a robustness check, we re-estimate the firms’ markups using the accounting method. The results of the estimation of Eq. (7) using the freshly calculated firm markups are shown in Column (1) of Table 4. The estimated coefficient of export duration remains to be significantly positive, indicating that an increase in export duration is advantageous to an increase in the firms’ markups and that the results of the previous analysis are generally reliable. These results are similar to the previous baseline regression results.

Table 4 Robustness test results.
Full size table

Lag effect of export duration

Rakhman (2010) pointed out that export duration can affect a firm’s export behavior in the subsequent year, so the impact of export duration on firm performance (e.g., markups) may also have a time lag. If a time lag exists in the impact of export duration on firms’ markups, then using the value of the firms’ markups and export duration in the same year in the previous regression may not be appropriate. Out of caution, taking into account the lag effect of export duration on firms’ markups, we re-estimate Eq. (7) by substituting the export duration of one lag period for the current period. The corresponding outcomes are detailed in Table 4’s column (2). Similar to the outcomes of the earlier baseline regression, the estimated coefficient of export duration of one lag period is still significantly positive. This result shows that the major finding of the previous analysis—that export duration will have a considerable positive influence on the firms’ markups—will stay true when we take into account the lag effect of export duration.

Endogeneity

In this study, we mainly talk about how export duration affects the markups of firms. Yet, there may be a two-way causal relationship between firm markups and the duration of exports. On the one hand, previous theoretical analysis shows that export duration may have an impact on firms’ markups through the “production efficiency” channel and “market-based pricing” channel. On the other hand, a firm with high markups will generally have a high profit margin and will likely survive in the export market, which will lead to an extended export duration. That is, a firm’s markups may also affect its export duration. To prevent possible endogeneity issues arising from two-way causality from interfering with the previous regression results, we re-estimate Eq. (7) in this section using the system GMM for the sake of robustness. According to the estimation results in column (3) of Table 4, which are consistent with the prior baseline regression findings, the estimated coefficient of export duration is still significantly positive. While the firms’ export duration may be endogenous, we create the industry-level mean of the endogenous explanatory variable as its instrumental variable (IV), following on a similar approach from Fisman and Svensson (2007). To conduct the IV regression of Eq. (7), we utilize the mean of the export duration of all the enterprises in each industry (three-digit industry level). The predicted coefficient of export duration is still significantly positive, according to the estimation results in column (4) of Table 4. These results suggest that the outcome of the previous analysis may not be confounded considerably by the endogeneity problem.

Exclusion of influence of China’s inclusion in the WTO

The sample period for this study runs from 2000 to 2007. Towards the end of 2001, China formally joined the WTO, thereby starting a new round of trade liberalization in the country. China’s inclusion in the WTO provided convenience and new development opportunities for firms to effectively explore overseas markets, which may have affected the results of the preceding investigation. To increase robustness, we attempt to eliminate the impact of China’s WTO membership on the earlier regression results in this section by creating a new subsample. To re-estimate Eq. (7), we specifically choose just the data from 2002 and later to generate the subsample following China’s participation in the WTO. According to the results in column (5) of Table 4, which are similar to those from the previous baseline regression, the estimated coefficient of export duration is still significantly positive. As a result, in general, the conclusions of the previous analysis will be less likely to be affected by the policy change brought about by China’s accession to the WTO and will show enough robustness.

Balanced panel

Previously, we estimated Eq. (7) using unbalanced panel data, in which considerable firm entry and exit behaviors may exist. The findings of the previous estimation may be skewed due to firms’ frequent entry and exit, particularly their transient presence. As a robustness check, in this section, we build a balanced panel dataset to re-estimate Eq. (7) in order to avoid the effect of the firms’ entrance and exit behaviors on the estimation results. We re-estimate Eq. (7) specifically by keeping only the companies that were present throughout the sample period. The findings in column (6) of Table 4 show that the estimated coefficient of export duration is still substantially positive, demonstrating that export duration has a significant positive impact on the firms’ markups and supporting the validity of the earlier analysis’s findings.

Mechanism analysis

Previous empirical research based on data from Chinese manufacturing firms reveals that an increase in export duration greatly raises markups. However, export duration may improve the firms’ markups via the “production efficiency” channel on the one hand, but may decrease the firms’ markups via the “market-based pricing” channel on the other, according to the theoretical analysis in Section 2. However, does export duration impact the markups of the Chinese manufacturing firms through the above two channels? In this section, we empirically test the “production efficiency” channel and “market-based pricing” channel.

We defined firm markup in the prior analysis as the proportion of a firm’s product price to its marginal cost (Markup=P/MC). The marginal cost of production is primarily determined by the firm’s production efficiency, and the price level of the product is heavily influenced by its market pricing power. These two factors together decide a firm’s markup. However, the CASIF database, which served as the primary source of data for the empirical analysis in this research, does not directly contain data on product prices and marginal costs. As a result, we are unable to explicitly assess how export duration affects the firms’ marginal costs and product prices. However, by adopting a similar strategy to De Loecker and Warzynski (2012), we can indirectly investigate the “production efficiency” and “market-based pricing” channels that underlie the effect of export duration on the firms’ markups. De Loecker and Warzynski (2012) claim that the marginal cost of production depends primarily on the productivity of a firm. Therefore, we can pinpoint the “production efficiency” channel by which export duration affects the firms’ markups from the standpoint of firm productivity. At the same time, De Loecker and Warzynski (2012) showed that \(lmp = \ln Markup = \ln P - \ln MC\). Next, by controlling for firm productivity in the regression model, we control for firm variability in marginal costs and thus indirectly identify the variability in firm product prices. Therefore, we can add the firms’ total factor productivity as a control variable to the baseline regression Model (7) to indirectly identify the “market-based pricing” channel through which export duration affects the firms’ markups. Here is the corresponding econometric model:

$$TFP_{it} = \beta _1Export\_T_{it} + \delta X_{it} + \lambda _i + \upsilon _t + \mu _{it},$$
(10)
$$lmp_{it} = \beta _1Export\_T_{it} + \beta _2TFP + \delta X_{it} + \lambda _i + \upsilon _t + \mu _{it},$$
(11)

where TFPit is the logarithm of firm i’ total factor productivity in period t, and the meanings of the other variables are identical to those in Section 3. For the total factor productivity of the firms, we use the ACF method for the estimation.

Table 5 reports the results of the estimation of regression Eqs. (10) and (11). We use the firms’ total factor productivity as the explained variable in columns (1) and (2). We can see that whether or not we use the ordinary least squares (OLS) estimation or IV estimation (we utilize the mean of the export duration of all the enterprises in each industry as IV), the export duration will greatly boost firm productivity, i.e., the longer the export duration, the higher the firm productivity. The marginal cost of production decreases, which is conducive to the increase in firms’ markups, with the increase in production productivity. This, to a certain extent, verifies the “production efficiency” channel of export duration affecting the firms’ markups. In columns (3) and (4), we establish the regression equation for the firms’ markup change from export duration and total factor productivity, which controls for the influence of firm productivity change on the firms’ markups. We can see that the estimated coefficient of export duration, is significantly negative, indicating that export duration has a significant negative impact on the firms’ market pricing power. Specifically, an increase in export duration results in a significant decrease in the firms’ product price level, lowering their markups. This result confirms the “market-based pricing” channel of export duration affecting the firms’ markups to some degree.

Table 5 Influencing mechanism analysis.
Full size table

In general, the regression results in Table 5 indicate that hypothesis 2 proposed in the previous theoretical analysis is valid. On the one hand, the increase in export duration will improve the firms’ production efficiency, which will be conducive to the improvement of their markups (the “production efficiency” channel). On the other hand, the increase will weaken the firms’ market pricing power (that is, the price level of the firms’ products), which will impose a negative effect on the firms’ markups (the “market-based pricing” channel).

Heterogeneity analysis

Analysis by industry type

Differences exist among different industries; thus, the effects of export duration on firms’ markups may vary by industry. Therefore, drawing on Yue and Wang (2020), we divide all the firms into two categories, that is, labor-intensive industries and capital-technology-intensive industries, to see if there are any notable differences in the effects of export duration on the markups of the firms in the two industry categories. The results of the estimation of Eq. (7) using the subsamples of firms in the capital-technology-intensive industries and firms in the labor-intensive industries, respectively, are reported in Columns (1) and (2) of Table 6. We can observe that for the labor-intensive industries, the estimated coefficient of export duration does not pass the significance test, indicating that markups of firms in the labor-intensive industries are not considerably impacted by export duration. On the other hand, the predicted export duration coefficient for capital-technology-intensive industries is notably positive, suggesting that an increase in export duration will help enterprises in these sectors boost their markups. A possible reason for this outcome is that in general, firms in capital-technology-intensive industries tend to pay considerable attention to R&D investment and technological upgrading and thus will likely obtain direct or indirect productivity gains from the “learning by exporting” effect as export duration increases. The positive impact of export duration on the markups of such enterprises through the “production efficiency” channel will likely be relatively large.

Table 6 Regression results by industry type and firm size.
Full size table

Analysis by firm size

The effects of export duration on the markups of firms of various sizes may differ greatly because of the substantial differences in the production efficiency, anti-risk ability, market monopoly power, and other characteristics of different sized enterprises. Thus, we use the sample’s median firm size as the threshold to split the sample into two groups, namely large-scale firms and small-scale firms. Using the subsamples consisting of the two types of firms, Eq. (7) is re-estimated. According to the regression findings in columns (3) and (4) of Table 6, a longer export duration can significantly raise the markup levels of both small- and large-scale firms. However, in terms of the magnitude, the increase in export duration is relatively conducive to the increase in the markups of the large-scale firms. This result may be explained by the fact that small-scale enterprises typically have lower productivity and weaker market pricing power than large-scale ones. As a result, an increase in export duration may have a smaller positive impact on the markups of small-scale enterprises via the “production efficiency” channel than that on large-scale enterprises’ markups, whereas the negative impact via the “market-based pricing” channel on small-scale enterprises’ markups is larger than that on large-scale enterprises’ markups. Ultimately, the increase in export duration is relatively conducive to the increase in the markups of large-scale firms.

Analysis by ownership type

China is currently in a time of transition to a market economy, and enterprises with various ownership types operate and produce in very different ways. Hu and Liu (2014) noted that the ownership structure is a significant element influencing the performance of Chinese firms due to China’s distinctive institutional setting. As a result, the effect of export duration on markups of firms with various ownership kinds may vary. Drawing on Ding et al. (2013), we divide all the sample firms into three categories—state-owned enterprises (SOEs), private firms, and foreign-funded firms—according to the percentage of the registered capital invested by the firms to investigate such heterogeneous effects. The regression results of the freshly estimated Eq. (7) is shown in columns (1)–(3) of Table 7. We can see that export duration only substantially influences the markups of foreign firms and has no discernible effect on the markups of private companies. A probable explanation for this may be that foreign companies will likely achieve productivity improvement from the “learning by exporting” effect with the support of technology, capital, and sales channels provided by their powerful parent companies. The estimated export duration coefficient for SOEs is significantly negative, which means that a rise in export duration will result in a significant reduction in the level of SOE markups. This is perhaps because SOEs frequently benefit from preferential government policies and funding and lack of incentives to improve productivity. The increase in export duration may impose a large negative effect on the SOEs’ markups through the “market-based pricing” channel, which exceeds the positive impact through the “production efficiency” channel.

Table 7 Regression results by ownership type and region.
Full size table

Analysis by region

Owing to the different natural endowment conditions arising from China’s vast territory and historical factors, substantial differences exist in economic development and openness among regions. The eastern coastal area has been leading the way in opening up when compared to non-coastal regions, with convenient means of transportation and quality infrastructure development. These variations will probably result in heterogeneous impacts of export duration on the markups of enterprises in different regions. For this reason, we first categorize all of the provinces into eastern coastal and non-coastal areas based on their distance from the sea, and we then re-estimate Eq. (7) using subsamples made up of the firms in the two regions. The estimation findings in columns (4) and (5) of Table 7 show that while the markups of the firms in the coastal area definitely benefit from the increase in export duration, the markups of the firms in the non-coastal area are not considerably impacted. This result may be explained by the fact that export companies are primarily concentrated along the eastern coast due to the region’s efficient transportation capabilities and good infrastructure development and that business competitiveness there is generally fierce. The strong competition effect may result in the relatively strong positive effect of the increase in export duration on the markups of the firms in the coastal areas through the “production efficiency” channel. Thus, the increase in export duration mainly increases the markups of the coastal firms significantly.

Analysis by export intensity

Firms generally are active in the domestic market first, and only those with high efficiency in production will choose to further serve the foreign market (Melitz, 2003). So, certain differences will exist in export intensity among different firms. To explore the heterogeneous effect of export duration on the markups of firms of different export intensities, we classify all the sample firms into two groups of firms with low and high export intensity, using the median export intensity of the firms (defined as the share of a firm’s export value in its total output value) as the threshold value. Equation (7) is then re-estimated, and columns (1) and (2) of Table 8 present the corresponding outcomes. We can see that while the markups of the enterprises with low export intensity are not much impacted by export duration, the markups of the firms with high export intensity are greatly influenced. A possible reason for this outcome is that firms with high export intensity, whose products are sold overseas and development centers are inclined toward export markets, are relatively sensitive to changes in the sales environment of export markets. Such firms are also sensitive to changes in market competition and updates in the production technology of similar firms, and their careful cultivation of export markets may make it easy for them to increase productivity through the “learning by exporting” effect as export duration increases (Esteve-Pérez et al. 2007). Thus, the increase in export duration may have a relatively strong positive effect on the productivity of such enterprises through the “production efficiency” channel, thereby significantly increasing their markups.

Table 8 Regression results by export intensity and trade mode.
Full size table

Analysis by trade mode

In the early stage of development, most developing countries have relatively abundant cheap labor resources but are relatively short of capital; thus, they generally choose to develop processing trade vigorously. Processing trade can not only help solve a large number of employment problems, but it can also make up for the lack of domestic capital. Since the reform and opening up, the Chinese government has implemented various preferential policies to facilitate the development of processing trade, which developed quickly in China and has grown to be a significant component of the nation’s foreign trade. We divide the sample firms into two categories: general trade firms and processing trade firmsFootnote 3, and then re-estimate Eq. (7) to explore the differential effects of export duration on the markups of firms of distinct trading modes. From columns (3) and (4) in Table 8, we can see that for the general trade firms, the increase in export duration can significantly increase the level of their markups. The estimated coefficient of export duration, however, is not significant for the processing trade firms, showing that the markups of these enterprises are not much impacted by export duration. This result is in line with what we anticipated. Processing trade enterprises essentially import raw materials or intermediate goods from international markets, process, assemble, and export them taking advantage of China’s inexpensive labor. Export duration has little effect on the markups of processing trade enterprises through the “production efficiency” and “market-based pricing” channels due to the “two ends are outside, big in and big out” nature of this industry.

Conclusion

In this study, we focus on export firms that have entered the export market and conduct an in-depth analysis of how export duration influences firms’ markups from theoretical and empirical perspectives. To understand the inner mechanism of how export duration affects firms’ markups, we first build a theoretical model to investigate the effect of export duration on firms’ markups within the context of firm heterogeneity. According to the theoretical analysis, there is some uncertainty regarding how an increase in export duration will affect firms’ markups. On the one hand, it will increase markups via the “production efficiency” channel, but on the other, it will decrease markups via the “market-based pricing” channel. Second, using the micro data of Chinese manufacturing enterprises, we empirically test the findings of previous theoretical analyses. The results of the empirical analysis show that, in line with the results of previous theoretical analyses, the increase in export duration contributes to the increase in the markups of Chinese manufacturing firms by increasing their production efficiency (“production efficiency” channel). Yet, by decreasing Chinese manufacturing firms’ market pricing power (i.e., the price level of their goods; “market-based pricing” channel), the lengthening of export duration also has a considerable adverse effect on the markups of those enterprises. Overall, the extension in export duration greatly raises the markups of Chinese manufacturing firms because the positive effects on markups via the “production efficiency” channel outweigh the negative effects on markups via the “market-based pricing” channel. After performing a number of robustness tests, such as key indicator replacement, endogeneity treatment, and controlling for other policy changes, this result still holds.

Finally, we further explore the heterogeneity in the effects of export duration on the markups of various types of firms by industry type, size, ownership, region, export intensity, and trade mode and obtain several results. First, while the markups of the enterprises in capital-technology-intensive industries increase as export duration increases, the markups of those in labor-intensive industries are not much affected. Second, the lengthening of export duration will probably result in larger markups for large-scale businesses than for small-scale ones. Third, the extension of the export duration has little impact on private firms’ markups and only greatly increases foreign firms’ markups while significantly lowering SOE markups. Fourth, while the markups of the enterprises in coastal regions benefit from the rise in export duration, those in non-coastal areas are not much impacted. Fifth, whereas the markups of firms with low export intensity are not greatly impacted by export duration, those with high export intensity may witness a marked increase in their markups. Sixth, the increase in export duration only encourages an increase in the markups of general trade companies; it has little impact on the markups of processing trade companies.

Finally, we present the policy implication of our study. Although the markups of Chinese exporters are relatively lower than those of non-exporters, in which a “low markup trap” exists, in this study, we find that increasing the duration of export trade relations can promote the increase in the markups of Chinese manufacturing exporters. This finding can provide a good idea to Chinese exporters to get out of the “low markup trap” and get rid of the “three low” labels of low price, low quality, and low profit. Policymakers should actively encourage exporters to continue to work hard in export markets, deepen existing export trade relationships, and develop new export trade relationships at the right time. By lengthening the duration of their export trade relationships, firms can obtain considerable productivity gains brought about by the “learning by exporting” effect, thereby promoting an increase in their markups. This is unquestionably crucial for raising the competitiveness of Chinese enterprises in the global market and enhancing welfare gains along the global value chain. At the same time, there is significant heterogeneity in how the lengthening of the export duration affects the markups of various kinds of Chinese manufacturers. Therefore, when formulating policies to encourage export firms to continue to cultivate the export market, the government should pay considerable attention to the heterogeneous impact of export duration on the markups of different types of firms to improve policy implementation accuracy and effectively bring the effects of the corresponding policies into play.