Learning by Exporting and Importing: Do What and Where You Trade Matter?

1. Learning by Exporting and Importing: Do What and Where You Trade Matter? Chih-Hai Yang National Central University Hsuan-Yu Lin Taiwan Institute of Economic Research 18 May, 2015, 中正大學

2. Outlines of Presentation 1. Introduction – background and aims 2..Literature review 3. Data and empirical specification 4. Empirical results 5. Concluding Remarks

3. Productivity: One of main causes differentiates per-capita income across countries. → promoting productivity !!! Methods or Channels: Advanced countries: R&D first. Developing countries: R&D (little), licensing FDI International trade a. exporting b. technology imports (learning and imitation)

4. Introduction – background (1) ♦ China has become the so-called “World’s Factory”. Share in international trade: increased from 1.7% in 1990 to 12.56% in 2011, overtaking the U.S. as the world’s largest trade economy. → export-led growth ♦ International trade is widely thought to have a positive influence on productivity. -- competition effect -- technological acquisition -- knowledge spillover

5. Introduction – background (2) ♦ China plays the critical role in the Asian integrated production networks, especially for the electronics industry. → assembly exports (fragmentation trade) → importing intermediate goods ♦ China’s electronics industry consists of a high ratio of foreign- owned enterprises (FOEs) -- In the electronics industry (scale-above enterprises) → about 40% in 2002-2007 (firm number)

6. Aims What’s new in this study? 1. Examining the productivity gain of international trade in China’s electronics industry. → existing firm-level studies are few, particularly “imports”. 2. Crucially, this study adopts various measures to capture the trade heterogeneity to implement empirical estimations. → baseline model: total trade → trade heterogeneity: destinations, variety, trade type, unit price 3. Considering both potential productivity-enhancing effects of exports and imports and comparing their relative impacts. 4. We compare the potential difference in productivity effect of trade between domestic firms and FOEs in China.

7. Main Findings 1. Examining the productivity gain of international trade in China’s electronics industry. →YES, both exports and imports 2. Trade heterogeneity. → destinations: advanced economies variety: export (no), import (yes) trade type: processing trade unit price: higher unit price 3. productivity-enhancing effects exports vs. imports → export is stronger 4. productivity effect of trade domestic firms vs. FOEs → a considerable difference, but yes for both → domestic firms benefit more

8. Productivity Gain of International Trade learning-by-exporting hypothesis: Clerides et al. (1988) QJE. →participating in export activity in international markets could help firms acquire technology and knowledge and thereby has a positive effect on promoting firm productivity. direct: local customers and competitors Learning-from-importing: →Imports, on the other hand, might mediate international knowledge transfers and promote firm productivity. 8

9. Game machine Printed circuit board ---- circuit layout 9

10. Literature learning-by-exporting hypothesis: Inconsistent results Wagner (2007, 2012) for a comprehensive survey on the relationship between export and productivity. Learning-from-importing: most studies find a positive link between importing and productivity, e.g., Muuls and Pisu (2009) for Belgium, Kasahara and Lapham (2013) for Chile, Vogel and. Wagner (2010) for Germany, Topalova and Khandelwal (2011) for India, Castellani et al. (2010) for Italy, Andersson et al. (2008) for Sweden, and Bernard et al. (2007) for U.S. 10

11. Trade Heterogeneity and Productivity Gain Exports Andersson et al. (2008), RWE. (Sweden) → productivity premiums increase in the number of markets and the number of products traded, respectively. Silva et al. (2012), (Portugal) → an learning effect among firms exporting to more developed markets, achieving a certain threshold of export intensity. Silva et al. (2013) → the greater the diversification of markets and goods, the better the performance, in terms of productivity Q:how export types (ordinary vs. process trade) and export quality relate to the effect of learning-by-exporting is not yet investigated. 11

12. Trade Heterogeneity and Productivity Gain Imports Sjöholm (1999), JDS. → Capital goods imported from developed economies have higher quality and contain advanced technology. OEM: promote their productivity through spilling over the advanced technologies embodied in the imported intermediate goods Muuls and Pisu (2009), WE → there is a variety effect and a quality effect caused by imported intermediates, as they have quality than local ones. 12

13. China Literature ♦ Learning-by-exporting (Firm-level) Yang and Mallick (2010), WE Sun and Hong (2010), WE Du et al. (2012), CER, domestic firms Dai and Yu (2013), WE Park et al. (2010), RE&S, developed countries → YES ♦ Learning-from-importing (Firm-level) No study. Yu (2015) → reductions in tariffs on imported inputs and final goods affect firm productivity

14. Data and Empirical Specification • Data Our dataset is constructed by linking two Chinese databases: Customs data and the Scale-above Enterprise Survey 1. Customs data: 2000-2006 8-digit HS product code, exporter/importer identity, product unit, quantity, unit value, total value, origin, destination, and type of trade, 2. Enterprise Survey: 2000-2007 → sales of over RMB 5 million basic information: the company name, the entry date, the industry code, the main products, and ownership financial information: to the financial statements of a firm. production information: sales, intermediate goods, and output

15. Sample of Chinese Customs Database

16. Sample: (NBS)

19. Data → Using the Chinese name to match two datasets. → firm-transaction level panel data in the China electronics industry during 2002-2007, → two-digit code of 39, 40, and 41. → unbalanced data containing 15,904 observations → domestic firms: 8,276 , FOEs: 7,628 observations Reasons: 1. in China’s electronics industry, there was a large share of firms being invested by Taiwan since the late 2011. 2. various measures of trade heterogeneity have to enter the empirical specification in a lagged form. 3. using firms with persistent internal trade enables us to simplify this discussion; despite it may suffer a selection bias problem.

20. Empirical Specification TFP: Levinsohn-Petrin TFP Total Trade : lnEXPORT lnIMPORT Trade destination: lnEX_Adv lnEX_Dev lnIM_Adv lnIM_Dev Trade variety: lnEX_Variety lnIM_Variety (6-digit HS code) Trade types: lnEX_Process lnEX_Ordinary lnIM_Process lnIM_Ordinary Trade quality: lnEX_Price lnIM_Price (6-digit HS code) KL: capital-labor ratio RD: R&D expenditure Size: firm size measured by number of employee Ownership: Private, SOEs, FOEs

21. Table 3 Variable Definitions and Summary Statistics Note: All statistics are calculated using the firm-level data, except for the lnUV which is calculated using the firm-transaction-level data.

22. Export and Import Heterogeneity in China’s Electronics Industry Figure 1a Figure 1b Distribution of exporting variety Distribution of importing variety the elasticities by which the number of firms falls of as number products increase are -1.764 and -1.960 for exports and imports, respectively. They are higher than the case of Sweden, reaching only -1.4 in the case of both exports and imports (Andresson et al., 2008).

23. Figure 2a Figure 2b Distribution of exporting destination Distribution of importing destination the elasticity by which the number of firms falls of as number countries increase is about -1.660 for exports and -2.056 for imports which are similar as those found in Andersson et al. (2008) that the corresponding figure are -1.7 and -2.0 in Sweden, a small open economy. Eaton et al. (2004) report an elasticity of -2.5 based on French firm-level export data.

24. Figure 3a Figure 3b Distribution of exporting price Distribution of importing price

25. Empirical Results Table 4 Learning by Trade – Total Exports and Imports

26. Discussions 1. The coefficient for export variable is significantly positive, supporting the hypotheses of learning-by-exporting. → consistent with findings in previous studies. 2. the import variable is also associated with a significant positive coefficient, indicating that firms with more import previously experience higher productivity. 3. the productivity-enhancing effect of exports seems to be stronger than that of imports without controlling for regional effects (FE models)

27. Trade Heterogeneity and Learning Effect Table 5 Trade Heterogeneity and Productivity – Destination and Variety

28. Discussions 1. Either exporting to or importing from advanced economies has a significantly positive impact on productivity, whereas the hypotheses of learning-by-exporting and learning-from-importing are not supported trading with developing and less developed countries 2. this learning-by-exporting effect is stronger than that of learning- from importing. 3. variety of exporting products have no significant influence on promoting productivity. Foxconn?? Apple product 4. increase in variety of importing goods has a productivity gain for Chinese importers

29. Trade Heterogeneity and Learning Effect Table 6 Trade Heterogeneity and Productivity – Trade Type and Price

30. Discussions 1. confirm the productivity-enhancing effect brought about by processing trade 2. We have to note that processing export is probably the consequence of processing import. 3. As the value-added of assembly exports of is low, it leads to a little lower productivity-enhancing effect compared with that of processing imports.. 4. the learning effect exhibits also for ordinary exports but not for ordinary imports. 5. Export and import price (lnEX_Price, lnIM_Price) are positive and significant in all model specifications →indicating firms export and import product with higher quality obtain a higher productivity premium

31. Table 7 Time structure of the learning effect

32. Discussions 1. the learning-by-exporting effect is stronger if firms export to advanced countries, while the magnitude decays sharply since the third year 2. the productivity effect of importing from developing countries turns to be significantly negative in the third years. 3. learning-from-importing in the electronics industry is mainly contributed by importing from advanced economies 4. importing more variety of products lead to a larger productivity – enhancing effect compared with exporting the same variety of products. 5. process trade is the main channel of international knowledge transfer. 6. the quality of trading commodities indeed matters to the learning effect of international trade

33. Table 8 Comparison between Domestic and Foreign Firms

34. Discussions 1. the learning-by-exporting hypothesis holds for both domestic and foreign firms, but domestic firms are associated with a large magnitude of productivity gain than their foreign-owned counterparts 2. both processing and ordinary exports are significantly positive for domestic exporters, while neither import types enhance their productivity. 3. the influences of trade types on productivity show a diverse pattern that processing import rather than processing export act a mediating role to promote productivity 4. the quality of trading commodity in terms of unit price remains to have a strong influence on productivity for both domestic firms and foreign affiliates.

35. Concluding Remarks • Both learning-by- exporting and learning-from-importing hypotheses are overall supported. • trade heterogeneity matters to this effect →Exporting (Importing from) to advanced countries, processing exports (imports), and exports (imports) with a higher unit price can lead to a productivity gain mediated by the learning effect. →Variety of exporting products is less matter to productivity • Exporting overall tends to result in a higher productivity premium than that of importing, while the effect of individual dimensions of trade heterogeneity varies a little. • The productivity effect of learning-by-exporting is stronger for domestic firms and it is mainly attributed to the processing exports. • Learning-from-importing is more relevant to firm productivity for foreign affiliates, as they import key components and technologies from their parent firms in advanced countries.

36. Thank You!! Comments are welcomed