What do we need besides trade?

Abstract

Trade has been shown by many authors to have strong positive impact on productivity. However, we also see some others being more reserved about such an impact. This study intends to investigate whether trade is enough for better economic performance, or rather, whether the impact of trade depends on the quality of local institutions. Using a panel of China's provincial data, empirical estimation results show that the better the quality of local institutions, the stronger the positive impact of trade on total factor productivity. If local institutions did not reach a certain quality level, trade expansion could have a negative marginal effect on total factor productivity.

Keywords:

• trade
• institutions
• total factor productivity
• China (Asia)

JEL Classifications:

• F10
• P33
• O43
• O53

Acknowledgements

The author thanks Professor Fan Gang for providing the NERI Marketization Index, Professor Zhang Jun for releasing the estimates of China's provincial capital stock, and an anonymous referee for helpful comments and suggestions. Financial support from the University of Macau is gratefully acknowledged (Grant RG048/05-06S/CY/FSH). The usual disclaimer applies.

Notes

Notes

1. They point out that, due to measurement weaknesses, specification errors, or methodological inadequacies, the conclusions of many existing empirical works on trade's importance are biased.

2. See Arellano and Bond (1991) for details.

3. See for example Griliches (1984).

4. The average value of ranges from 0.45 to 0.63, and the national average is 0.52. Among others, labor share in Chow (1993) is 0.40 for the period 1952–1980, in Hu and Khan (1997) is 0.453 for the period of 1979–1994, in Chow and Li (2002) is 0.41 for the period of 1952–1998, and in Zhang (2002) is 0.50 for 1952–1998.

5. See for example Kaufmann, Kraay, and Zoido-Lobaton (2002).

6. NERI: National Economic Research Institute, Beijing.

7. We recall that institutions include formal rules and informal norms such as customs, habits, traditions …

8. For those who might be interested, we also replicated regressions as in Table 1 using within-group (fixed effects) specification. The results showed that INS ln Trade was always positive and statistically significant at the 1% level. But as within-group (fixed effects) estimates suffer from the econometric problems we mentioned above, as such we are not presenting or making inference from them.

9. The findings of Ljundwall (2006), which show that during the reform period up to 2001, only in 13 (mostly coastal) out of the 27 Chinese provinces in estimation did the export-led-economic-growth hypothesis hold, tend to lend support to our results.

10. The Index for Tibet was 2.5 in 2005. We do not report Tibet in the table as Tibet is not included in our empirical estimation, due to incomplete data for other variables.

11. Zhang (2002) showed that the capital–output ratio in China during the 1990s ranged between 3.20 and 3.45.