How Trade Liberalization of Intermediates Contributes to China’s Technology Upgrade
ChenWen(陈雯)andMiaoShuangyou(苗双有)
1
School of Economics, Xiamen University, Xiamen, China
2
School of Economics, Zhongnan University of Economics and Law, Wuhan, China
Abstract:
Under a heterogeneous firm analysis framework, this paper creates a theoretical model to investigate how trade liberalization of intermediates influences the technology choice of firms based on the data of Chinese manufacturers during 20002006. According to our empirical result, after China’s WTO entry, trade liberalization of intermediates significantly induced Chinese exporters to apply advanced technologies. In our further consideration of the differentiated productivity of firms, we found that such an effect is related to the initial productivity of firms and only significantly induces mediumproductivity firms to upgrade their technologies. In addition, the technology-promotion effect of trade liberalization of intermediates is the most significant for technology-intensive exporters and the least significant for labor-intensive exporters.
Keywords:
trade liberalization of intermediates, choice of production technology, firm productivity JEL Classification: F10, F14, D24
1. Introduction
Since its official accession to the WTO in 2001, China began to cut import tariffs on a wider scale and completely delivered on its tariff reduction commitments in 2010. Productionoriented import policy has major effects on the production of Chinese enterprises by allowing China to selectively reduce the level of import tariffs for critical components and other intermediate inputs. Technology, which is central to the production process, directly affects product cost and quality and plays a pivotal role on the supply side. The 18th CPC National Congress1 and the 12th Session of the Central Leading Group for Comprehensively Deepening Reforms2 identified technology as a core element that spurs economic growth. Taking the import tariff reductions of the China’s WTO entry in 2001 as a quasi-natural experiment, this paper investigates the effects of trade liberalization on the choice of technology by Chinese manufacturers. In this paper, the quantitative study with the liberalization of trade in intermediate inputs as a typical case not only sheds light on how the
liberalization of trade in intermediates affects the choice of production technology. Our findings may also provide detailed data support and empirical evidence for the future adjustment of trade policies.
Existing studies on trade liberalization on firm behaviors focus on how trade liberalization affects the export and performance of enterprises. Some studies consider that the liberalization of trade in intermediate inputs may enhance firm productivity (Amiti and Konings, 2007; Luong, 2011; Mao and Sheng , 2013), improve the quality of export products (Bas and Strauss-Kahn, 2015; Fan et al., 2015b; Amiti and Khandelwal, 2013) and increase markup percentage ( Fan et al., 2015a). Most studies on the production technology are carried out at the industry level through the lens of factor endowment structure (Tan and Li, 2012), cost of trade (Liu and Zheng, 2013) and environmental regulation ( Zhang et al., 2011). However, based on the assumption of “homogeneous firms”, which departs from the reality, these studies are unable to reveal how homogeneous firms of the same industry respond to the same policy in heterogeneous ways. A few studies attempted to examine production technology at firm level yet most of them followed the methodologies of TFP decomposition (Zhang et al., 2011), concentration of skilled workforce (Liu, 2009) and deviation of capital-labor ratio (Chen and Liu, 2015). While studies on trade liberalization and technology progress are increasingly sophisticated in their respective fields of research, very few have examined these two important topics of research in combination and still less have shed light on how the trade liberalization of intermediate inputs affects the choice of technology. Thus, this paper intends to contribute to this area of research.
By creating a general equilibrium model, Yeaple ( 2005) arrives at the conclusion that trade liberalization encourages firms to adopt higher- technology production processes. By introducing the heterogeneity of manufacturers into the theoretical model and using the data of China’s listed manufacturing companies, Liu ( 2009) discovers that trade liberalization does not help all exporting manufacturers to upgrade technology. By creating a two- country model of heterogeneous firms and using the data of Argentinian firms, Bustos ( 2011) considers that trade integration promotes the upgrade of importers and exporters and is only conducive to the technology upgrade of enterprises with medium-level productivity. However, the above literature studies make no distinction on how trade liberalization in intermediate inputs and trade liberalization in final products affect the choice of technology by firms. Bas and Berthou (2013), who create a partial equilibrium model, believe that trade liberalization for inputs does not induce firms to upgrade their technologies. Based on the data of Indian firms, they arrive at the conclusion that trade liberalization of intermediates only promotes the technology upgrade of medium-productivity firms but makes no distinction on the effects of trade liberalization of intermediates between exporters and nonexporters with different levels of productivity.
Then, whom does the China’s policy of trade liberalization of intermediates mainly affect in the choice of technology, exporters or nonexporters3? Is there any particularity in the effects by Chinese firms? In answering these questions, this paper will create a theoretical model on the basis of previous studies and employ China’s highly refined tariff data, customs data and industrial firms’ data of 2000-2006 to carry out our empirical analysis. This paper may offer the following contributions: first, based on previous theoretical studies, this paper also introduces the theoretical models of heterogeneous firms, import of intermediates and choice of production technologies. In this manner, we not only reveal the different effects of trade liberalization of intermediates on technology choice between exporters and non-exporters, but forecast major differences in the effects due to different level of productivity within exporters. Second, this paper systematically elaborates how trade liberalization
of intermediates affects the choice of production technologies by Chinese exporters after China’s accession to the WTO, and employs firm-level data to verify the conclusions of theoretical model; third, technology measurement error exists by using TFP in previous literature due to the unobservable nature of firm production technologies. In addition to the use of capitallabor ratio, this paper creates two proxy variables based on the methodology of Bas and Berthou ( 2013) 4. By identifying whether firms have applied advanced production technologies using detailed customs information, this paper may effectively reduce the disturbance of complex factors such as ownership to the identification of the level of firm technology and serve as a supplement to relevant analysis on this topic.
The rest of this paper is arranged as follows: Section 2 describes the theoretical model; Section 3 creates an econometric model to be tested and explains the choice of variables and data sources; Section 4 conducts a test and analysis on the econometric model based on relevant data of China’s manufacturing industry; Section 5 offers a robustness test; the last section provides conclusions and relevant policy recommendations.
2. Theoretical Model
On the basis of the theoretical models of Melitz ( 2003) and Bustos ( 2011), this paper creates a theoretical model that simultaneously incorporates imported intermediates, tariffs on intermediates and firm production technology to examine how trade liberalization of intermediates affects the choice of technology by firms. Below shows the model specification:
2.1 Domestic Demand and Manufacturing
(1) Domestic demand
The utility function is constant elasticity of substitution (CES) utility function: Where, q is the quantity of consumption goods, ω is the varieties of heterogeneous products, Ω is the consumption set that can be obtained, and σ is the elasticity of substitution between products. The consumption of each variety of heterogeneous commodities q ( ω) and overall price index P are respectively as follows:
(2) Domestic production
In our two- country model, there are two economic sectors in each country: the sector for the production of intermediates and the sector for the production of final goods. The only input factor of the sector for the production of intermediates is labor, which is employed for the production of homogeneous intermediates, yet different countries are engaged in the production of different intermediates with constant return to scale; the production of each unit of intermediate requires a fixed amount of labor input and the market structure is fully competitive. Thus, the price of domestic intermediates is 1. The second sector employs intermediate for the production of continuous differentiated final goods described by Melitz (2003), which requires the use of domestic intermediates and all overseas intermediates. Assuming that all the manufacturers of final products are price recipients in the market of intermediates, they are subject to the import tariff on intermediate inputs τm. Referencing the approach of introducing intermediates into production as followed by Bas and Berthou (2013) and assuming that production function is CES function, we have: Where, φ is productivity and xd and xm respectively denote domestic intermediates and
overseas intermediates; γi is the efficiency of firm i in the use of imported intermediates, which denotes the level of firm production technology. l and h respectively denote low production technology and high production technology, assuming that γh> γl= 1. Firms adopting higher levels of production technology have lower marginal cost of production, i. e. the marginal production cost of firms with high production technology ch is below the marginal production cost of firms with low production technology cl, and relative marginal production cost ch/cl is an increasing function of tariff for intermediates τm.
Under the assumption of CES production function, the marginal production cost of firms is given by ci:
The above equation indicates that the marginal production cost of firms is related to the level of their production technology. Firms with higher technology γi or lower tariffs for imported intermediates boast lower marginal production cost. As can be known from the above equation, the relative marginal production cost ch/cl of hightech manufactures can be expressed as follows: By der i v ing the tar i ff fo r impo r ted intermediates τm through equation (6), we may prove that the relative marginal production cost ch/ cl is an increasing function of tariff for imported intermediates. Trade liberalization of intermediates will reduce the relative marginal production cost ch/cl for high-tech manufacturers. In order to describe this relationship, relative production cost ch/cl is uniformly denoted as λ ( τm) in the following paragraphs.
According to Bustos ( 2011), corporate profitability can be divided into the following four situations depending on a firm’s export status and level of production technology:
Profitability of Type I firms: serving domestic market only and employing low production technology Profitability of Type II firms: serving domestic market only and employing high production technology
Profitability of Type III firms: serving domestic and international markets simultaneously and adopting low production technology
Profitability of Type IV firms: serving domestic and international markets simultaneously and adopting high production technology
Where, ; τx is the home country’s export tariff; f is the firm’s fixed cost of production; fx is the fixed cost when a firm exports; fh is the additional fixed cost incurred to the firm upon its upgrade from low production technology to high production technology.
Figure 1 can be derived by assuming that high-technology firms are exporting firms6:
As shown in Figure 1, firms with productivity below will exit their home market; low-productivity firms ( ) employ low production technology but do not export; medium-productivity firms ( ) adopt low production technology and export; and highproductivity firms ( ) adopt high production technology and export8.
2.2 Firm Decision-making
( 1) Decisions to enter or exit domestic markets
Decisions to enter or exit domestic markets in this paper are only determined by firms serving domestic markets alone and adopting low production technology. Their decision to exit or stay in the domestic market is subject to the critical condition of business profit at zero, i.e.: ( 2) Decisions to export to international markets
Decisions to export to international markets are subject to the critical condition of constant profits irrespective of whether their goods are exported or not, i.e. , which gives us the critical productivity for export .
( 3) Decisions to adopt high production technology
Only when their productivity is sufficiently high will firms adopt high production technology. The critical condition for the adoption of high production technology is the profitability being equal irrespective of whether firms adopt high production technology or low production technology, i.e. , which gives us: In the above equation, r is the firm’s profit. By arranging equations (11) and (12), we may obtain the equation of critical productivity with the adoption of high-technology: The above equation must satisfy the condition in order to ensure . Because relative marginal production cost ch/ cl is an increasing function of tariff τm for intermediates, tariff may directly influence critical productivity for the adoption of high technology through relative marginal production cost. Meanwhile, import duty for intermediates may also indirectly affect critical productivity for the adoption of high-technology through its impact on critical productivity and needs to be analyzed by deriving the specific expression of critical productivity through market
equilibrium.
2.3 Market Equilibrium
Assuming that productivity conforms to Pareto distribution, probability density function is and cumulative distribution function is . Conditions for the firm’s free entry ( FE) and zero profit ( ZCP) are respectively as follows:
Where, δ is depreciation rate and fe is the fixed cost paid by firms upon entry into domestic markets. Probability for firms to adopt high production technology is given
by , where is the probability for Type I firms and is the probability for Type III firms. , and respectively denote the average profitability of Type I, III and IV firms. By combining equations (14) and (15), we may obtain the expression of critical productivity .8 Then, by substituting the expression of critical productivity into equation ( 13), we may obtain the relational expression between critical productivity for the adoption of high technology and tariff for intermediates τm.
Through the above expression, we may demonstrate that critical productivity for the adoption of high technology will decrease with the reduction of tariff for intermediates, i.e.
; critical productivity will increase with the reduction of tariff for intermediates, i. e. . Thus, we may arrive at the comparison before and after the trade liberalization of intermediates, as specifically shown in Figure 2.
As shown in Figure 2, critical productivity for the adoption of high-technology reduces from
to after tariff reduction for intermediates and firms with productivity in the range between
and will upgrade from low production technology to high production technology.
Through the above analysis, we may arrive at the proposition of this paper’s theoretical model.
Proposition: Trade liberalization of intermediates will promote the adoption of high production technologies by firms and such an effect of promotion is subject to the initial productivity of firms and only those exporting firms with medium productivity will be induced to upgrade their technologies.
In the interest of length, the process of demonstration will not be elaborated in this paper 10.
3. Econometric Model, Measurement Indices and Data Explanation
3.1 Specification of Econometric Model
In order to test whether trade liberalization of intermediates after China’s WTO entry induced Chinese exporters to upgrade their technologies,
this paper creates the following empirical model based on the above-mentioned theoretical model.
Where, subscripts i, j and t respectively denote firm, sector and year. Yit is the level of production technology for firm i in year t; intertariff is tariff for intermediates; vj, vp and vt reflect the effects of industry (two-digit code), province ( two- digit code) and year on the production technology of firms; εit is stochastic disturbance term. Control variable Z includes firm size, firm age, level of firm production technology in previous year, weighted firm tariff for final goods, as well as Herfindahl index.
In order to test whether the technology promotion effect of trade liberalization of intermediates after China’s WTO entry is subject to the original level of productivity for firms, this paper divides firms into five groups based on their TFP in 2002 to investigate how trade liberalization of intermediates affects production technologies of firms with different initial productivity. The model is as follows:
Where, ρ is the group of firm i; Q is group dummy variable for different groups of productivity; the interaction term is weighted firm tariff for intermediates times grouping dummy variable. The significance of this interaction term and the value of coefficient can be used to assess whether tariff reductions for intermediates affects the choice of production technology by firms in various groups and the extent of such effect.
It needs to be noted that when the proxy variable of firm production technology is dummy variable ( firm_cap1 and firm_cap2), non-linear probit model should be employed; when the proxy variable of the level of firm production technology is capital-labor ratio, the fixed effect model should be employed.
3.2 Explanation of Indicators
( 1) Proxy indicator of firm production technology
In this paper, a number of proxy variables are employed to indirectly reflect the level of firm production technology as no indicator exists in China’s database of industrial enterprises that can be used to directly measure the level of firm production technology. Our approach is explained as follows:
Bas and Berthou ( 2013) employed the dummy variable of whether capital goods are imported to measure firm production technology. Their approach is modified in this paper to obtain proxy variable 1 and proxy variable 2:
a. Proxy variable 1 ( firm_cap1): Referencing the approach of Bas and Berthou (2013) 11, during the period of 2000- 2006, the value of proxy variable firm_cap1 is 0 before firms first import capital goods; upon the first import of capital goods, the value of proxy variable firm_cap1 is 1; after the first import of capital goods by firms, the proxy variable firm_cap1 is 1 throughout the following years.
b. Proxy variable 2 ( firm_ cap2): Different from the selection of proxy variable 1, this paper sets the standard of proxy variable 2 as whether the total value of imported capital goods exceeds US$1,000, which aims to reduce the interference to the effectiveness of proxy variable caused by the import of a few capital goods.
This paper identifies the import of capital
goods as the proxy variable of firm’s production technology for the following reasons: there are significant technical and quality superiority of overseas capital goods and the dependency of Chinese manufacturers on high- tech products such as imported equipment. Capital goods refer to durable goods employed by firms for the manufacturing of machinery and equipment, reflected as fixed assets in accounting information. Compared with non-capital goods, capital goods contain a higher level of technology and serve as an important medium for technology diffusion (Xu andWang, 1999). Eaton and Kortum (2002) and Mutreja (2014) found that the production of most capital goods in the world are concentrated in a few R&D intensive countries, which export capital goods, while other countries (including China) are importers of capital goods. Hence, we have reasons to believe that high-tech Chinese manufacturers are more likely to import capital goods. Moreover, through the analysis of typified facts, this paper discovers that firms importing capital goods are more productive than those that do not import any capital goods (in terms of labor productivity and TFP) and that the import of capital goods demonstrates not much difference across sectors13. In addition, it has been found through a regression analysis that after importing capital goods, firms have greatly increased their productivity and TFP, which also supports the rationality of our choice of proxy variable.
c. Proxy variable 3: capital- labor ratio (KL): given the effect of replacement between production technology and labor, high- tech firms within an industry often exhibit higher capital- labor ratios. Thus, this paper also uses the indicator of capital-labor ratio as the proxy variable of firm production technology.
(2) Trade liberalization of intermediates Existing li t e r a t u r e normally adopts industry tariff for intermediates as firm’s tariff for intermediates, which results in certain measurement errors and cannot reflect the tariff differences on intermediates of heterogeneous firms in the same industry. Hence, following the practice of recent literature ( Bas and StraussKahn, 2015; Tian and Yu, 2013), this paper calculates the weighted average tariff at firm level in order to better reflect the effects of trade liberalization on the import cost of heterogeneous firms through the following equation: Where, intertariffit is intermediate import tariff for firm i in year t ( HS six- digit code); intertariffct is intermediate c’s import tariffthrough general trade in year t, reflecting product-level import tariff13; θict is the share of product c in the total import of intermediates by firm i.
(3) Weighted firm tariff for final products Firm tariff for final products reflects the level of competition from overseas manufacturers caused by changes in the level of tariff for final goods calculated through contariffit= Σθictcontariffit, where contariffit is the import tariff for final goods f in year t and θict is the share of f import in the total import of final goods for firm i in year t. Lower tariff of final goods means higher level of competition from overseas manufacturers.
(4) TFP
TFP measurement methodologies include OLS, OP, LP and GMM. OLS methods simultaneously has the problems of endogenous deviation and sample selection deviation; OP method overcomes the problem of endogenous deviation and sample selection deviation caused by the mutual determination of variables (Olley and Pakes, 1996), yet great sample losses are incurred, while LP method effectively resolves the problem of data loss (Levinsohn and Petrin, 2003); GMM method avoids the disturbance of endogeneity ( Lu and Lian, 2012). This paper will select OP and LP methods to respectively estimate the TFP of Chinese manufacturers. In the process of estimation, this paper will measure labor input by the average number of workforce employed by various firms and employ industrial value-added and annual average fixed assets to measure the output and capital input of firms and adopt the deflator with 1998 as base period provided by Brandt et al. (2012) for adjustment. (5) Firm size
New trade theory underscores the impact of economy of scale on the pattern of international
trade and geographic distribution of firms. Meanwhile, new trade theory also considers that firm size is a major source of firm heterogeneity and has a major impact on firm behaviors. Using the data of US industrial firms, Swamidass and Kotha (1998) investigate the relationship among the size, high production technology and market performance of firms. Lee (2009) also found that firm size exerts major effects on firm choice of technology. Hence, this paper needs to control the impact of firm size on the choice of technology. Larger firm size means that the same upgrade of production technology entails a lower fixed cost in proportion to the firm’s total cost. Thus, the estimation coefficient of firm size is expected to be a positive value. This paper adopts the logarithm of firm’s workforce employment to measure firm size.
(6) Firm age
According to the life-cycle theory, a firm’s business turnover and profits will grow with its age in the early stage of its development. The longer the firm operates, the more likely its production equipment and other facilities will age and become replaced with new and more advanced technology ( Mao and Sheng, 2013). Thus, older firms are more likely to adopt advanced production technologies. However, when the firm starts to decline in its late stage of development, it becomes less likely to renew its technology. An empirical study by Bas and Berthou (2013) also uncovered the major effects of firm age on the choice of technology. In this paper, firm age is the difference between current year and the year of a firm’s founding plus 1. (7) Herfindahl index
Equation for calculating Herfindahl index is , where saleit is the business turnover of firm i in year t; salejt is the total business turnover of two-digit sector j in year t; Sit is the market share of firm i in sector j in year t. Higher Herfindahl index suggests higher market concentration.
3.3 Data Explanation
Firm information in this paper is taken from China Industrial Enterprise Database ( 20002006). Referencing the practice of Xie et al. ( 2008) and Cai and Liu ( 2009), we deleted samples that meet any of the following criteria: ( 1) the number of employees is fewer than 10 persons or is missing; ( 2) total assets are smaller than liquid assets; ( 3) total assets are smaller than the net value of fixed assets; ( 4) cumulative depreciation is smaller than currentphase depreciation; ( 5) gross industrial output is smaller than zero; ( 6) the year of founding precedes the year of record or is missing. In addition, the discrepancy of industry codes before and after 2003 caused by China’s amendment of Classification of Economic Sectors in 2003 is adjusted in this paper following the method of Brandt et al. ( 2012). All nominal variables in this database have been adjusted using the deflator with 1998 as base period.
Tariff data in this paper are taken from the WTO’s Tariff Download Facility Database. Due to the inconsistencies of harmonized versions of HS six-digit codes14, we have unified product tariffs into HS1996 version based on the table of conversion between HS1986 and HS2002 versions provided by the United Nations Statistics Division and thereby classified imported products into intermediate goods, final consumer goods and capital goods according to theHS1996 and BEC15 comparison table. Analytical samples in this paper are firms that import intermediates through general trade since tariff for intermediates is only applicable to firms that import intermediates through general trade.
Product- level information in this paper is taken from China’s customs database, which records each and every entry of data from the monthly import and export customs declaration
of firms during 2000-2006, which includes the tax codes of firms, the HS eight- digit import/ export products, method of transport, as well as other information. This paper needs to match this database with the database of industrial enterprises that contains corporate information through the following method: (1) firm name is directly used as matching field; (2) the last seven digits of postcode and telephone number are used as matching fields. As long as any field is successfully matched, the firm will be included into the consolidated data.
4. Empirical Analysis
4.1 Effect of Intermediate Trade Liberalization on Technology Choice of Chinese Manufacturers
(1) Analysis by the status of export
In our theoretical model, we assume that the critical productivity of high- technology is superior to the critical export productivity of firms. Under this assumption, trade liberalization will only induce exporters to upgrade their technology. In order to verify the rationality of this critical assumption, we divide firms into non-exporting firm samples and exporting firm samples by the value of their export delivery and respectively conduct empirical analysis using equation (16). In our models, we have controlled the effects of time, industry and region. Where, the dependent variable of columns ( 1) and ( 2) is proxy variable firm_cap1 and the dependent variable of columns (3) and (4) is proxy variable firm_cap2; the dependent variable of columns (5) and (6) is capital-labor ratio KL and the samples of columns ( 1), ( 3) and ( 5) are non- exporters; samples in columns ( 2), ( 4) and ( 6) exporters. Table 1 lists the estimation results.
This paper mainly deals with the estimation coefficient of weighted tariff for intermediates. Fo r no n - e x p o r ter s am p les , th e w ei g h ted coefficients of tariff for intermediates in Model (1), (3) and (5) are all positive but not significant and we did not find any evidence suggesting that trade liberalization of intermediates induced China’s non-exporters to apply higher technology. This result contradicts with Bas and Berthou’s ( 2013) conclusion that trade liberalization of intermediates induced all Indian firms (including
non-exporters) to apply higher technology. For exporter samples, the coefficients of weighted tariff for intermediates in Models (2), (4) and (6) using different proxy variables are all negative and the coefficients of the first two models are significantly negative, indicating that trade liberalization of intermediates significantly induced China’s exporters to adopt higher technology, which coincides with Bustos’s (2011) conclusion.
Control of variables: Coefficient of tariff for the firm’s final goods is significantly positive, indicating that competition from overseas competitive firms will undercut the profitability of Chinese manufacturers, making it less profitable for them to apply high technology; Herfindahl index coefficient is positive for some variables and negative for others, meaning that domestic market structure has an uncertain effect on the choice of firm technology; the coefficient of firm size is positive, indicating that larger firms are more likely to apply higher technology, which is consistent with common sense; the coefficient of firm age is mostly negative, meaning that older firms are less likely to apply high technology: older firms are normally more competitive and less motivated to upgrade their technology or more likely to be in the declining stage16, which restricts their technology upgrade. In addition, another possible explanation is that new firms tend to adopt more advanced equipment (Zhang et al., 2011).
(2) Categorized analysis by factor intensity
Production technology of firms is normally related to the type of their products and the industry in which they operate. The question is whether trade liberalization of intermediates affects the technology upgrade of Chinese exporters of different products and industries in different ways. To answer this question, referencing the classification method of Yuan et al. (2015), this paper classifies exporters into labor- intensive, capital- intensive, resourceintensive and technology-intensive firms to carry out empirical study respectively.
The result indicates that after the WTO entry, the effect of China’s intermediate trade liberalization on exporters’ technology choice is related to the factor intensity of firms. The estimation coefficient of weighted intermediate tariff of resource-intensive exporters is positive but not significant. A possible explanation is that the sample size for this type of firms is relatively small and error tends to be more significant. For the other types of firms, the coefficient of weighted intermediate tariff is significant at 1% and negative, meaning that intermediate trade liberalization of induced these three types of firms to apply higher technology. Moreover, judging by the absolute value of coefficient, this effect is the smallest for laborintensive exporters and the most significant for technology-intensive exporters. This conclusion is consistent with our expectation.
4.2 Differentiated Effects on Exporters
In the above paragraphs, it has been verified that trade liberalization of intermediates induces manufacturing exporters to apply higher production technology. In this part, we will verify that such an effect is subject to the initial productivity of firms. In order to verify this proposition, we apply equation (17) for an analysis on China’s manufacturing exporters17, with results shown in Table 2.
Obviously, no matter control variable is included or not, the estimation coefficient of weighted tariff for intermediates is significantly negative for medium- productivity firms and insignificant for firms at the two ends of productivity spectrum. This indicates that
trade liberalization of intermediates only induces China’s medium- end manufacturing exporters to upgrade their technology, which is consistent with the hypothesis of this paper and also consistent with the conclusion of Bustos ( 2011) that trade liberalization induces medium-productivity exporters to upgrade their technology.
5. Robustness Analysis
In order to test the credibility of above empirical results, this paper employs different methods for robustness analysis.
( 1 ) Group testing of the effect of intermediate trade liberalization on China’s nonexporting manufacturers
Non- exporting firms are classified into different groups to further test the hypothesis that intermediate trade liberalizations of China’s WTO entry does not significantly induce Chinese non- exporting manufacturers to apply high production technology. Empirical results are shown in Table 3.
As can be known from Table 3, no matter control variable is included or not, the estimation coefficient of weighted tariff for intermediates with a multitude of proxy variables is not significant and only a small part of coefficients is significant but significantly positive. All coefficients are not significantly negative. We have uncovered no evidence that intermediate trade liberalizations of the China’s WTO entry significantly induced non-exporters with different productivity to apply high technology, which robustly supports the conclusions of Table 1. (2) Different trade liberalization indicators In order to avoid the impact of tariff estimation for intermediates on the empirical result, this paper employs non-weighted tariff for intermediates to re-estimate the grouped firms. The results shown in Table 4.
I n Ta bl e 4 , Co l u m n s (1 ) - (3 ) ar e th e empirical results of non- weighted tariff for intermediates, which is generally consistent with the empirical results of weighted tariff for intermediates in Table 2, i.e. the estimation coefficient of intermediates is significantly negative for medium- productivity firms and insignificant for firms at both ends of the productivity spectrum, which robustly supports the empirical result of weighted tariff for intermediates and verifies the expectation of this paper’s theoretical model.
(3) Grouping of different firms
The above section employs OP method to estimate the TFP of exporting firms but the error of estimation may cause firms to be mistakenly grouped and thus affect the empirical results. In order to exclude the disturbance of productivity estimation errors to the empirical conclusions, we employ LP method to re- estimate the TFP, regroup firms and re-conduct empirical analysis, with results shown in Columns ( 4) and ( 5) in Table 4.