Adler, G. and S. Sosa, 2011, “Commodity Price Cycles: The Perils of Mismanaging the Boom,” IMF Working Paper 11/283 (Washington: International Monetary Fund).
International Monetary Fund, 2008, “Who’s Driving Whom? Analyzing External and Intra-Regional Linkages in the Americas’” ed. by M. Mühleisen, S. Roache, and J. Zettelmeyer (Washington).
Kose, M., A. Rebucci, and A. Schipke, 2005, “Macroeconomic Implications of CAFTA-DR,” in Central America: Global Integration and Regional Cooperation, ed. by M. Rodlauer and A. Schipke (Washington: International Monetary Fund).
Sosa, S., 2010, “The Influence of ‘Big Brothers:’ How Important are Regional Factors for Uruguay?,” IMF Working Paper 10/60 (Washington: International Monetary Fund).
Swiston, A., 2010, “Spillovers to Central America in Light of the Crisis: What a Difference a Year Makes,” IMF Working Paper 10/35 (Washington: International Monetary Fund).
We are grateful to Nicolás Eyzaguirre, Miguel Savastano, Charles Kramer, Luis Cubeddu, Herman Kamil, Nicolas Magud, and seminar participants at the Central Bank of Colombia for their useful comments and feedback; and to Alejandro Carrión for his research assistance.
Commodity dependence is a common feature across Latin America. See Adler and Sosa (2011) for a documentation of this pattern.
Kose, Rebucci, and Schipke (2005) and Swiston (2010) find that the impact of shocks stemming from Mexico on Central American countries is negligible. IMF (2012) documents that Mexico’s trade linkages with Central America have been, and remain, very weak.
See a survey of the literature in Muhleisen, Roache, and Zettelmeyer (2008).
A few studies, however, have examined the influence of Brazilian factors on a particular country. Sosa (2010), for example, explores spillovers from Argentina and Brazil on the Uruguayan economy.
Spillovers from systemic countries in other emerging market regions—notably China—have also been subject of recent study. See, for example, IMF 2011.
The analysis focuses on exports of goods from neighboring countries to Brazil, due to data limitations regarding trade in services. For some countries (e.g., Uruguay), the latter may also be relevant, although the overall trends do not change significantly if these are also considered.
Bolivia experienced a remarkable increase in trade with Brazil during the past decade, explained almost completely by exports of gas.
A country’s vulnerability to shocks stemming from a trading partner may also depend on the composition of trade flows, in particular whether exports are mainly commodities or manufactured products since the degree of ‘re-allocability’ may be different across these categories.
We do not formally test whether the shocks are propagated through trade channels, due to lack of data on bilateral exports in real terms. However, given the limited financial linkages, our prior is that trade is the most relevant transmission mechanism. Our results are consistent with this prior.
Previous studies on external factors driving economic cycles in Latin America had focused on the EMBI. Since we are looking for a pure exogenous global financial variable, any EMBI measure would need to exclude Latin America’s EMBI. More importantly, data on the EMBI is available starting only in the mid-1990s, so using such measure would imply losing a significant number of observations.
The use of a broad commodity price index, as opposed to country specific commodity price indices, allows to gauge the indirect impact of commodity price shocks through the knock-on effect on Brazil’s output, and also to preserve degrees of freedom in the estimation.
Standard unit root tests (augmented Dickey-Fuller) show that all the variables are stationary in first differences (except the VIX, which is stationary in levels). In addition, most co-integration tests suggest that the variables in the model are not co-integrated (i.e., the null hypothesis of no co-integration cannot be rejected). Hence, it is adequate to estimate the model in first differences. The number of lags is based on the Akaike Information Criterion (AIC), which suggested two lags in most cases, and one in the rest.
Results are robust to different orderings within the group of global variables.
Results for Bolivia should be interpreted with caution, because its exposure to Brazil changed markedly over the period (Figure 3), and trade linkages mostly reflect gas exports, which are governed by long-term contracts, with minimum volumes.
The result for Peru is puzzling, given the lack of direct (trade or financial) linkages with Brazil, and may reflect the existence of other common factor driving cycles in both countries, not captured by the model.
Results also point to amplification effects of other global shocks (global demand and commodity prices).
The analysis focuses on a horizon of 8 quarters.
In Uruguay, even though a depreciation of Brazil’s real did not lead to a (statistically significant) change in the real effective exchange rate—suggesting that Uruguay tended to adjust its exchange rate in response to this shock—domestic output declined on impact, and the effect was statistically significant. This result likely reflects the average contractionary effects of a depreciation in Uruguay, given the high degree of liability dollarization and associated currency mismatches displayed during the early part of the period of analysis.