Dynamic Elasticities of the Trade Balance of Brazilian Agriculture in Relation to the Exchange Rate

ABSTRACT

This work examines the elasticities of the Brazilian agricultural trade balance in a dynamic, or non-linear, perspective since emerging economies such as Brazil are quite susceptible to internal and external shocks that can produce asymmetries in these elasticities. To this end, data from January 2000 to July 2019 are used, along with a time-varying cointegration analysis. The results suggest that the elasticities of the net exports in relation to the real exchange rate and foreign income were positive and elastic in practically the entire period considered. Domestic income had a negative influence, but with inelastic impacts and less variability.

KEYWORDS:

• Agriculture
• Brazil
• dynamic elasticities
• trade balance

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ This technique is a nonlinear cointegration method which allows the Marshall-Lerner condition to be analyzed at different points throughout time, i.e., with differing conjectures, smoothly capturing the possible existing nonlinear relationships. With this technique, it is possible to analyze the relationship between variables in a dynamic perspective.

² The five largest grain exporters are the United States, Canada, Australia, Argentina, and the European Union.

³ This sample period was chosen because, in 1999, the Brazilian foreign exchange market began operating under a free-floating exchange rate regime.

⁴ The complete derivation of the BRM model can be found, for example, in Arruda, Castelar, and Martins (2019).

⁵ The estimated coefficient of the effect of the real exchange rate on net exports should be positive to confirm the validity of the Marshall-Lerner condition.

⁶ As seen in Bierens and Martins (2010), a small value of m imposes a smooth behavior for the vector β_t approaching the time-invariant case. A caveat about the selection of the order m of the Chebyshev polynomial should be made. Neto (2015) argues that, although Bierens and Martins (2010) suggest that the usual information criteria can be used, the selection procedure is not clearly described in the econometric literature related to models that use such temporal polynomials. According to Neto (2015), this choice is somewhat ad hoc, with a maximum choice of time variation according to theory, economic beliefs, or existing literature being also preferred.

⁷ Given by the ratio between the sum of exports and imports and GDP.