50 likes | 222 Vues
Sum up results. Higher import penetration, especially from low-income countries, results in lower manufacturing employmentOffshoring to high-income countries has a complementary effect on manufacturing employmentIncreased exports are associated with higher employment, partially offsetting employme
E N D
1. Discussion of Salem et al., Trade, Offshoring and U.S. Multinational Employment in the Manufacturing Employment in the United States, 1999-2008 Cosimo Beverelli
WTO Research Division
2. Sum up results Higher import penetration, especially from low-income countries, results in lower manufacturing employment
Offshoring to high-income countries has a complementary effect on manufacturing employment
Increased exports are associated with higher employment, partially offsetting employment losses from import penetration
3. Strength of the paper 1999-2008 time frame is robust to Krugmans (2008) critique of studies based on data from the 1980s or 90s
Autor et al. (2011) report that the low-income-country share of US imports increased from 2.9 per cent in 1990 to 5.9 per cent in 2000, and then climbed to 11.7 per cent by 2007
China accounted for 91.5% of this import growth over the period 2000-2007
Increased exposure to trade with China, and to a lesser extent with other developing economies, suggests that the labor market consequences of trade may be larger today than 20 years ago
4. Suggestions [WIP]
No theoretical foundation for impact of foreign on US employment of U.S. parent
The two are endogenously determined and it is difficult to spell direction of causality
Could check for China effect
Quantification of economic significance of results and comparision with existing studies (e.g. Ebenstein et al. 2009, Harrison and McMillan 2011)
Give more emphasis to export share result
Divide literature review in industry-level versus firm-level studies
5. Econometric issues Concerns about small sample (GMM is for large N small T, you have small N and T)
Include the Arellano-Bond test of second-order correlation for residuals in differences (but does not work well with small N)
Treatment of standard errors not specified (at the very least, use small-sample correction to VCV estimate)
You do not control for productivity growth, which you indicate as a potential channel of employment reduction
Potential endogeneity of U.S. trade exposure (import and export shares may be correlated with industry labor demand shocks)
Could use import and export shares in other developed countries as instruments (see Autor at al., 2011)