![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
This article examines the impact of exchange rate changes on Chinese firms’ decisions on export market entry and export share. Using a large dataset for Chinese firms in 2000–2006, we find that changes in exchange rate levels play a significant role on both export extensive and intensive margins of Chinese firms. Compared to studies using macro data, our firm-level analysis allows us to control for firm and industry heterogeneity. Firm size and location matter. We do not find a difference between foreign and domestic firms in responding to exchange rate changes. Industry heterogeneity is also found to be important.
Acknowledgements
We gratefully acknowledge financial support from the Leverhulme Trust under Programme Grant F114/BF.
Notes
1. Note that the estimation should be viewed as reduced form regressions for trade in nominal values.
2. Modelling export entry decisions with an export dummy is common practice in the literature, for example, Clerides et al. (Citation1998), Das et al. (Citation2007) and Kneller et al. (Citation2008).
3. Firm level controls include firm size, average wage, labour productivity, age, ownership dummies, and coastal region dummy. Definitions of variables are in Online Appendix 1.
4. According to Chinese regulations, foreign firms in China include firms owned by other countries and those owned by Hong Kong, Taiwan and Macao. Here we separate foreign firms into two groups: foreign and Chinese ethnic, because Chinese ethnic firms use the same language and have more networks in China.
5. As yearly China GDP growth and yearly world GDP growth are included, some of the year dummies have to be dropped to avoid multicollinearity. We also tried year dummies only to control for fixed effects common across time to capture the cyclical and macroeconomic factors across time, and results remain similar (available on request).
6. Wooldridge (Citation2002: Chapter 16) considers two cases in Tobit models. Corner solution outcomes refers to the situation that ‘y takes on the value zero with positive probability but is a continuous random variable over strictly positive values'. The censored regression model arises due to data censoring. Export share can be regarded as the former.
7. Kneller et al. (Citation2008) adopt a similar methodology but in a cross-section framework proposed by Papke and Wooldridge (Citation1996).
8. The Bernoulli probit quasi-log-likelihod estimators are consistent and asymptotically normal regardless of the distribution of y conditional on X: y could be a continuous variable, a discrete variable, or have both characteristics (Papke and Wooldridge, Citation2008).
9. The data to construct the ratio is Chinese trade data compiled by China Customs Statistics. The trade data is commodity data which is classified according to HS8 with the information about whether the export is processing trade or not. We therefore converted HS8 commodity data to industry level data according to Chinese National Economic Industry Classification 2003 for each 3 digit sector.
10. See Hubbard (Citation1998) for a survey.
11. Our firm dataset covers the period of 2000–2006. As we always lag one period of exchange rate indices in our regressions, the REER indices used in this article are from 1999 to 2005. Online Appendix 3 presents REERS for all the other 3-digit industries not shown in .
12. We use the natural logarithm of REER index point in the regression. So the partial derivative should be d[P(export=1)|X]= 0.22*d[REER]/REER. As the mean of REER is 95.9, d[P(export=1)]/d[REER]=0.22/95.9=0.0023.
13. The results for Tobit models are similar to those for QMLE models (available on request).
14. See notes for for the calculation of the measures. Summary statistics are: processing trade (mean: 0.376; standard deviation: 0.245); sunk cost (mean: 0.701; standard deviation: 0.248); competition (mean: 0.949; standard deviation: 0.55); leverage (mean: 0.737; standard deviation: 0.07).