Energy use in U.S. manufacturing and increasing imports from China: Empirical analysis

ABSTRACT

This study numerically estimates the effect of increased Chinese import penetration using U.S. manufacturing-level panel data from 1997 to 2005. It also decomposes this effect into factor substitution effects and output scale effects. Since import penetration rates may have endogeneity problems, we use China’s share of world trade to quantify Chinese import penetration. The results show that increased imports from China increase fuel and electricity consumption: the marginal effect of Chinese import penetration is small but statistically significant, ranging from about 0.05% to 0.08%. Interestingly, the decomposition reveals opposite effects across energy types: the factor substitution effect for fuel is positive, while the factor substitution effect for electricity is negative, despite the impact of Chinese imports. The effects on fuel and electricity are both significantly positive, while the effect on electricity is negative. These results suggest the need for different policy responses across industries to the effects of trade, with implications for the environmental problems caused by energy use.

KEYWORDS:

• Energy use
• Chinese import penetration
• factor substitution effects
• instrumental variable

JEL CLASSIFICATION:

• F10
• F18
• F64
• Q40

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The United States is an energy-abundant country, and the price of energy is cheaper than in other countries. Therefore, the United States is relatively specialized in energy-intensive industries compared to China. On the other hand, China has cheap labour and a comparative advantage in labour-intensive industries.

² This survey had also been collected in previous periods. However, this survey was classified by SIC 4-digit classification instead of the NAICS 6-digit classification before 1997. This prevents having longer panel data from earlier years.

³ This specification is similar to Antweiler, Copeland, and Taylor (2001) and Cole (2006), but it is different in that industrial -specific characteristics are considered.

⁴ All specifications included year and industry-fixed effects to control for industry-specific macroshocks and time-invariant unobserved shocks.