import uses and in what can we learn from chinese micro
TRANSCRIPT
Import Uses and Domestic Value Added in Chinese Exports:What can we learn from Chinese micro data?
Shunli Yao, Hong Ma, Jiansuo PeiUIBE Tsinghua UIBE
October 2013
Motivation• WTO/OECD launched “Made in the World” initiative in 2011• http://www.oecd.org/industry/ind/measuringtradeinvalue‐
addedanoecd‐wtojointinitiative.htm
Motivation• Estimate DVA in exports
an important topicbut lack of data
• Ahmad et al. (2011) pointed out that “the key challenges in the immediate future concern the quality of trade statistics and the assumptions made to allocate imports to users”
• OECD’s exercise with Turkish micro data is an attempt to reveal the patterns of firm heterogeneity in trade and production and based on that, to improve trade in value added measures (Ahmad and Araujo, 2011).
Literature• Input‐output tables are the appropriate tools to meet the needs
(compared with product‐level analysis, e.g. the iPhone value chain, and trade‐data based analysis)
• Preferably international input‐output tables (e.g., GTAP, WIOD, EORA, and OECD)
• Problems: China’s trade structure does not properly represented biased estimates for China vis‐à‐vis its trade partners
• Solution: develop China’s DPN (or DP) framework
pioneered by Chen et al. (2001) followed by Lau et al. (2007), Dean et al. (2011), Chen et al. (2012), Koopman et al. (2012)
Literature (cont.)
• Key is to spilt‐up the import use from intermediate deliveries• Proportionality assumptions and firm survey• Problems:
Over‐simplified, Feenstra & Jensen (2012) biased estimates, Bernard et al (2007).
• Solution: explore existing firm level data
better surveyto narrow the scope of assumptions
Import uses in China 2007 IO
• Competitive IO tables
42/135 sectors, public release
• Competitive & Non-competitive IO tables
42/135 sectors, internal use
• DPN frameworkw/ processing & normal exp, domestic prod
42 sectors, internal use
Intermediate use Final use D P N DFD EXP TOT
D DDZ DPZ DNZ Df 0 Dx
P 0 0 0 0 Pe Px
N NDZ NPZ NNZ Nf Ne Nx
IMP DM PM NM Mf 0 Mx
VA )( ′Dv )( ′Pv )( ′Nv
TOT )( ′Dx )( ′Px )( ′Nx
A sketch of the DPN framework
Source: Lau et al. (2007), Chen et al. (2012)
Import use in China 2007 IO
• How to determine the shares of import use by sector?
• Semi-survey by NBS, first time in Chinese IO tables
10,000 large firms, sampled from enterprise dataset
Yangtze Delta 3500, Guangdong 1300(Shanghai 1100, Zhejiang 1100, Jiangsu 1300)
Focus given to FIEs
Import use in China 2007 IO
• No use of firm level trade data
No prior knowledge of trade patterns
• Biased in regional and size distribution
small firms behave differently from large ones
roughly 1/3 in Yangtze1/3 in Guangdong1/3 in Northern China (w/ Japan & Korea)
Customs survey
• Targets trading companies• Selection of companies and goods
Based on 2010 import volume.• For non‐processing imports
top 60% percentile, with imported goods 1,734 traders and 343 8‐digit HS goods
• For processing imports, roughly 10,000 firms and 300 8‐digit HS goods
• No use of firm level production data• No prior knowledge of production patterns
The Customs Imports Uses Survey Questionnaire Commodity code: XXXX.XXXX Imports (thousand USD): XXXX Proportion: XX.X%
Primary classification
Secondary classification (Input-output sector) Amount Ratio (%)
Note
Intermediate use
01 Agriculture, forestry, animal husbandry and fishery
02 Coal mining and washing industry 03 Oil and gas exploration industry … … … … … … … … … … … … … … … … 65 Public administration and social
organizations
Final use Final consumption expenditure Capital formation
Aggr. estimates
China micro data: available, but usable? (1/2)
More problems:• Imports resold to other firms• Imported inputs not used in current year (inventory)• Firm produces in multiple sectors
But, never mind, make assumption and use it!
Upward, Richard, Zheng Wang and Jinghai Zheng, 2012, “Weighing China’s export basket: The domestic content and technology intensity of Chinese exports,” Journal of Comparative Economics
China micro data: available, but usable? (2/2)
Only a subset of exporting firm
Firm heterogeneity calls for more effort in survey design
Firm heterogeneity are more pronounced at industry level (exp share distribution)
Firm heterogeneity are more pronounced at industry level (int. import /tot output)
Firm heterogeneity are more pronounced at industry level (int. import / tot input)
More on firm heterogeneity
DVS can be any number within [38.9, 69.7]
Conclusion & discussion• Current micro data are not adequate to give precise DVS
estimates– non‐representative samples – trading agency problems
• A wide range of DVS estimates are given in our paper
• Suggestions for future firm survey work • 1) identify the small production firms from firm level trade
data. Then, the dataset L&M can be expanded to include large, medium and small firms (LMS).
• 2) select a sample of firms from LMS to be covered by the survey. Other aspects, such as ownership, sector, location and trading partners should also be considered.
Thanks for your attention!