Technical barriers to trade and China’s exports: firm-level evidence

ABSTRACT

This paper studies the trade-restriction and trade-promotion effects that technical barriers to trade (TBTs) have on export performances of Chinese firms over the period 2001–2006. Different from tariffs and most traditional non-tariff measures, TBTs can impose both positive and negative impacts on exports. On the one hand, TBTs increase the compliance costs and have the trade-restriction effect. On the other hand, TBTs improve product quality and mitigate uncertainty, which enhances demand and promotes trade. We investigate how TBTs influence the extensive and intensive margins of a firm’s exports and how the impact varies with the size, productivity, and ownership of a firm. Specifically, we look into a firm’s entry and exit decisions, export value, unit price, and quantity. The estimated results show that TBTs decrease both the extensive and intensive margins of small firms, and they improve the export performances of large firms. The finding indicates that while small firms that cannot meet the technical standards will leave their markets, large firms will stay in their markets and expand their export volumes. The trade effects of TBTs also change by characteristics of the importing country and sector.

KEYWORDS:

• Technical barriers to trade
• intensive margin
• extensive margin
• firm-Level data
• panel data
• unit price

JEL CLASSIFICATION:

• F13

Acknowledgement

We are grateful to conference participants at the 93^rd Annual Conference of the Western Economics Association International. Wei-Chih Chen acknowledges the financial support from the Ministry of Science and Technology, Taiwan. (106-2410-H-194-099). Xiaohua Bao acknowledges the Natural Science Foundation of China (71673177), the Social Science Foundation of China (18ZDA069), and Shanghai Municipal Education Commission (2019-01-07-00-07-E00031).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ In the case of correcting market failure, the trade effects of TBTs are unintended. For instance, labelling requirements mitigate the information asymmetry problem between producers and consumers. Regulations of environmental protections reduce a negative externality. The adoption of common standards will increase consumers’ utility in industries with network externalities. In circumstances without market failure, the trade effects of TBTs are the aim of the policies. Technical regulations are placed in order to restrict imports, expand domestic firms’ market share, and benefit specific interest groups.

² Melitz and Ottaviano (2008) constructed a heterogeneous-firm model with endogenous markups. The model shows that in the short run, unilateral trade liberalization decreases the cost cut-off of entry, which intensifies the competition (i.e. only firms whose costs are less than the cut-off enter the market). This pro-competitive effect increases the number of product varieties and the productivity of firms, and it lowers markups.

³ Hu and Lin (2016) and Hu et al. () use the firm-level data of China to examine how the EU’s policy of child resistance on cigarette lighters affects the export performance of Chinese firms. While other papers estimate the average effect of TBTs across sectors and countries, Hu and Lin (2016) and Hu et al. () focus on one specific TBT and explore its influence thoroughly.

⁴ Fontagné et al. (2015), El-Enbaby, Hendy, Zaki (2016), Fontagné and Orefice (2018) and Kamal and Zaki (2018) focus on the heterogeneous effect over firm size, which is proxied by the value of exports.

⁵ Not all variables appear every year in the dataset, and variable names are not consistent in different years. There are also many mistakes in variable values, mostly due to firms misreporting. Following the methods in Cai and Liu (2009) and Feenstra, Li, and Yu (2014), we drop observations based on the criteria below: (1) Observations with missing key variables (such as id number, total assets, number of employers, etc.); (2) Observations where liquid assets are greater than total assets; (3) Observations where total fixed assets are greater than total assets; (4) Observations where net values of fixed assets are greater than total assets; (5) Observations where sales are less than 5 million RMB.

⁶ Among the 8490 TBTs raised between 1995 and 2006, 6108 of them do not have explicit corresponding product HS codes. We only calculate the TBT measures based on those notifications with HS codes.

⁷ For example, a TBT with an HS4 code of “2010” will apply to all products with code of “2010” will apply to all products with HS6 codes beginning with “2010”, such as “ 201010” and “ 201020”.

⁸ The sample size in the benchmark estimation is much smaller than the original merged sample of the customs data and the production data because of two reasons. First, since we include the firm-country-HS6 fixed effect in the benchmark estimation, a firm-country-HS6 with only one year of exports will be dropped. Second, we lagged the firms’ explanatory variables by one year, which makes observations from 2000 be dropped in the estimations.

⁹ The sample size of the exit regression is smaller than that of other trade margins due to right censoring. In the last year (2006) of the sample, we cannot observe whether or not a firm stops exporting in the next year, so we drop observations of 2006 in the exit margin. On the contrary, the entry margin does not have a data censoring problem since the customs data of 2000 is available and our sample starts from 2001 (2001–2006).

¹⁰ To mitigate concerns about endogeneity, a firm’s productivity is measured by the initial value at the year the firm starts exporting to a market.

¹¹ Fontagné and Orefice (2018) use a binary variable to instrument the TBT STC. The binary variable equals one if two conditions hold: (1) country c has a TBT STC on at least one other product, and (2) at least one other country has a TBT STC on product p.

¹² This method is the same as the exit dummy defined in Fontagné and Orefice (2018).

¹³ The agricultural sectors are represented by HS01-HS15. The manufacturing sectors are represented by the other 2-digit HS codes except for HS25–27 and HS98.

¹⁴ We also classify products into differentiated and homogeneous (including reference priced) ones based on the classification of Rauch (1999). The estimated results show that there is no significant difference regarding the influence of TBTs between these two groups of products.

¹⁵ The initial size of a firm is measured in the first year it starts exporting to a country-HS6.

¹⁶ The F-statistics are reported in the bottom of Table 6.