Global shocks and trade response - terms of trade, J-curve and the Marshall-Lerner condition: evidence from Brazil

Abstract

The traditional terms of trade (TOT) shock analysis may be an unrealistic experiment, especially for a small open economy where the state of the world economy endogenously determines TOT. The dynamic relation between TOT and the trade balance needs to consider the nature of the global shock that jointly moves price and income, globally and domestically. We estimate Bayesian global SVARs for Brazil using a structure that easily communicates to traditional DSGE models to evaluate the presence of the J-curve and the Marshall-Lerner condition (MLC) following innovations we claim to resemble world supply and demand shocks. We do not encounter a J-curve for aggregate trade nor for trade in fuel, and capital and consumption goods. MLC is not verified only for trade in capital goods, but even in the cases where MLC is present, we observe that TOT and the volume exported are not correlated as expected, since improvement in the volume exported happens when TOT appreciates. Our results suggest that income effect plays the most prominent role for the dynamics of the trade balance and for the validity of the MLC, dominating the TOT effect.

Keywords

• J-curve
• Marshall-Lerner condition
• Terms of trade
• Global shocks
• SVAR