Does Ethnicity Pay? Evidence from Overseas Chinese FDI in China

Yasheng Huang，MIT Sloan School of Management

Li Jin，Harvard Business School

Yi Qian

Kellogg School of Management, Northwestern University

摘要：利用所有在华外商直接投资公司的综合样本，探讨了种族是否提高经营绩效的问题。虽然已经有相当数量的理论文献研究种族和外国投资，预测种族对企业盈利能力的影响还远不清楚。本文的实证证明，在整个样本中，少数民族企业并不比非少数民族企业拥有经营优势。进一步的测试表明，虽然少数民族企业在初期具有比非少数民族企业更大的经营优势，但这种优势会随着时间的推移而下降。然后，探究了可能导致民族优势下降的原因，结果表明，缺乏对无形资本和人力资本的投资是驱动本文观察到的模式。总的来说，研究结果表明，种族并不能给公司带来永久的经营优势。

数据：设法得到了一个非常全面的数据，1998-2001年中国工业普查(CIC)，由中国国家统计局(NBS)编制。据作者所知，CIC是关于中国工业企业最详细的数据库，非常适合于探索民族差异的盈利能力。

We managed to get a very comprehensive data, the Chinese Industry Census (CIC) for the years 1998-2001, compiled by National Bureau of Statistics (NBS) in China. CIC is, to our knowledge, the most detailed database on Chinese industrial firms and is well suited to exploring the ethnic variance in profitability.

The dataset contains detailed financial data for each of the entire population of those FIEs in China with sales in excess of 5 million yuan (roughly 600,000 USD) for each of the census years between 1998 and 2001. All the detailed company information and the panel structure allow us to properly handle the potential endogeneity of the FDI by origin. To give a few examples, certain industries are more productive and may have attracted more ECE investors because ECE investors may possess more innate knowledge about these industries. This correlation between the productivity of the industry which a company belongs to and whether the company adopts ECE FDI could cause upward bias in the coefficient estimate on the ECE-FDI treatment variable. We mitigate this potential bias by controlling for industry dynamics at a very disaggregated level. The industry codes in CIC are at the four-digit level, detailed to the level of product groupings, such as leather shoes, as opposed to just shoes. The panel structure helps to eliminate any time-invariant firm-specific effects and the study of a large number of firms and sectors make our results generalizable.

Similarly, whether a company receives ECE FDI could also be endogenous to the company’s own productivity. The best we can do is then to control for the crucial company characteristics, such as the number of years since incorporation, size, leverage and labor-intensiveness, in the regression models. In particular, these variables help to capture the observed and the correlated latent company experience and productivity advantages. In the set of benchmark regression models with only four key variables and the industry and province dummies, there is enough flexibility to add non-linear terms of the key covariates to overcome the limitations of the OLS linear assumptions, and we are re-assured that the main findings of our paper are robust to these considerations.

3.1. 测试ECE(ethnically Chinese economies)的静态效应

The benchmark model to test the two camps of theories on ethnicity and firm performance is an industry- and province-fixed effects regression model carried out in the samples of joint venture (JV) companies and non-JV companies separately: