Heterogeneity among Intermediaries in China
Our main data comes from the Chinese Customs Trade Statistics (CCTS) database, as used by Ahn et al. (2011) and Tang and Zhang (2012). This is compiled by the General Administration of Custom of China, and includes firm-level export values and quantities at the 8-digit HS product level by country of destination. For each individual export flow, we have both the quantity exported and the corresponding free on board (f.o.b.) value in U.S. dollars. We can then calculate the unit value of exports for each firm, product, and destination. The database also records the destination of exports and contains firm-specific information such as ownership (foreign, state or private), name and address. We collapse the data to the annual level and aggregate product data to the 6-digit HS level.
We adopt the common practice in the literature of identifying intermediary firms based on the Chinese characters that have the English equivalent meaning of “importer,” “exporter,” and/or “trading” in the firm’s name (Ahn et al., 2011; Tang and Zhang, 2012). In particular, we follow the approach in Tang and Zhang (2012) and search for the following pinyin (Romanized Chinese) phrases: “jin4chu1-kou3,” “jing1mao4,” “mao4yi4,” “ke1mao4,” “wai4jing1,” “wai4mao4,” and “gong1mao4.”
We would like to differentiate between intermediaries that export a variety of products spanning unrelated sectors and those with a core competence in a single line of business. The former correspond to the type of traders that appear in the empirical literature, where intermediaries have consistently been found to export more products to more destination markets and more varieties per country than direct firms (Ahn et al., 2011; Bernard et al., 2010a; Crozet et al., 2013). This aspect of trading firms suggests that part of the role of intermediaries is to help firms send products to destination markets. On the contrary, intermediaries with a restricted core competence, which we will refer to as specialized traders, conform to the image of intermediaries in Dasgupta and Mondria (2012): they screen product quality and then reveal this to consumers.
We will distinguish between the two types of intermediaries according to their distribution of export sales over products: we calculate for each intermediary firm f the share of exports in each product p, s fp . We then compute the firm’s Herfindahl index by aggregating the squares of the shares of all the products exported by firm f: 15 HI = ( s p )2 f f , (1) p∈s f.
where Sf is the set of (Nf) products that firm f exports, and s fp is the export-value share of product p over the total export value of firm f. A higher value of HIf means that the firm’s export basket spans a narrower range of varieties. Firm-level product scope is expected to rise with firm size and productivity (Bernard et al., 2010a; Bernard, Redding, and Schott, 2011). To control for those mechanical associations in our analysis of the heterogeneity of product concentration across intermediaries, we regress the HI measure on a quadratic polynomial in firm size (proxied by export value) with fixed effects for ownership,16 and then take the residual, ∈ HI f .
Quality dispersion in China
We exploit the variation in the scope for quality differentiation across products and space to see whether intermediaries, or a subset of them, mitigate adverse-selection problems by guaranteeing product quality. We will show that there is substantial heterogeneity in terms of quality differentiation (i.e., the dispersion of qualities) across Chinese cities for a given product. This heterogeneity determines the prevalence of export intermediaries and, more importantly, the importance of the role that specialized intermediaries play in overall intermediation.
Our estimates of quality differentiation follow Khandelwal (2010) by calculating quality dispersion for each city–product pair as the standard deviation of the estimated ln Kfpc across all (firm–product– destination) flows.26 We use data for 2004, as our empirical strategy relates 2005 intermediary prevalence to the one-year lagged quality dispersion at the city–product level.
Our concentration on product–city-, as opposed to product-level, variation in quality dispersion reflects that quality dispersion varies across both space and products in our data. Table OA11 in the online Appendix (for access details see Supporting Information at the end of this paper) reveals substantial variation in quality dispersion across Chinese cities, even for fairly homogeneous goods (garlic and silicon). Following Khandelwal (2010), we treat quality dispersion as an exogenous product characteristic.
Our work however differs in that we also measure quality dispersion at the city level. What we call cities here correspond to the first administrative division of the 31 Chinese provinces.27 Given China’s large population and area, the 321 cities in our sample are anything but small. We further only retain city–product pairs with over 10 (firm–product–destination) export flows to ensure that there are enough observations for a reliable measure of quality dispersion.
The Empirical Analysis of Intermediation
Our regression estimates the share of intermediary exports in city–HS6 observations, which is correlated with a proxy for the scope of vertical differentiation.
While firm-level customs data is available for 2000 to 2006, the Chinese system of trading licenses was not entirely dismantled until 2005. Following the literature, we consider the single year 2005 as the baseline as export licenses had been removed by 2005, and any firm that wished to trade directly with foreign partners was free to do so (Ahn et al., 2011). We show in the robustness checks in Subsection 4.2.2 that our results continue to hold in a panel specification appealing to variation over time in a given city–product pair of the relationship between quality dispersion and intermediary prevalence.
Accounting for the destination country
Our empirical strategy has so far mostly exploited variations in the need to screen quality by the source of the exports. The capacity of buyers to deal with information asymmetry and identify the quality of Chinese exports also depends on their nationality. Buyers may better be able to verify the quality of their imports if they are not too far away from and share linguistic and cultural ties with China. By way of contrast fixed export costs or import tariffs imposed by the destination country are not expected to affect the difficulty of quality assessment. Table 7 shows the moderating role of country characteristics in the correlation between the intermediation export share and vertical differentiation. The dependent variable is the share of intermediary exports in city–product– country observations in 2005. The key parameter of interest is the interaction between our quality-dispersion measure for a city–product pair and proxies for information asymmetries between China and the destination country. Fixed effects at the city–product, city–country and product–country levels are introduced.
The first three columns introduce bilateral variables to capture the particular links between China and its partner countries: distance 50 and the share of ethnic Chinese population in 1990 and 1980 respectively.51 Column (4) considers the sunk cost of exporting to a partner as measured by the number of import procedures in the World Bank’s Doing Business Report (Djankov et al., 2006), while column (5) uses the tariff imposed by the partner.52 Our findings are fairly intuitive. The link between specialized intermediary prevalence and quality differentiation is stronger for more distant exports and falls when the destination country has more ethnic Chinese. By way of contrast, our proxies of fixed trade costs and the tariffs imposed on Chinese exports at destination do not affect the quality-verification role.
Table of contents :
Chapter 1 Aid for Trade and the Quality of Exports
II. Methodology and Data
Chapter 2: Quality screening and trade intermediaries: Evidence from China
III. Intermediaries and Quality
IV. The Empirical Analysis of Intermediation
V. Empirical Results: Intermediation and Quality
Chapter 3: Export Orientation, Demand Uncertainty and Innovation Premium: Evidence from Chinese firms
II. Export Orientation and R&D investment in China
III. Empirical Analysis
IV. The Role of Demand Uncertainty
V. Responses of Heterogeneous Firms under the Demand Uncertainty