What have I done to deserve this? Global winds and Latin American growth

Eduardo Levy Yeyati, Luciano Cohan 12 January 2012



Four years ago, when what would become the worst crisis in 80 years was just a concern for the important but encapsulated US mortgage market, academics and practitioners were debating whether the emerging world, which showed no signs of weakening as the developed world sunk into recession, had entered a new age of real (business cycle) ‘decoupling’. Twelve months later, the synchronised collapse of the global economy put to rest these aspirations of economic autonomy (see for instance Rose 2009 and Wälti 2009 on this site).

Latin America, despite showing a more robust macroeconomic resilience during current stressful events, has still had its fair share of mood swings. In mid-2008, the challenge was how to contain inflationary pressure of an overheated economy. In 2009, the slowdown was equal or deeper than in other regions in the world. The growing ties with emerging Asia (particularly, China) were not enough to disable the systemic implosion.

Figure 1. The synchronic collapse

Source: Levy Yeyati and Cohan (2011).

As we all know, the global economy recovered quickly. Yet today the developed countries are once again in crisis mode. With disruption this time round promising to be less dramatic but more persistent than the post–Lehmann Brothers fallout, concerns about global exposure and growth recoupling take on new relevance. Are we facing 2009 all over again? Or, more generally, what is the expected impact of the global cycle on the region?

With this in mind, in a recent paper (Levy Yeyati and Cohan 2011), we use a stylised econometric approach to identify the percentage of change in the growth rate of major Latin American countries attributable to a few global drivers. Because Latin American economies went through important structural changes in the 2000s (particularly regarding their foreign currency position and dependence on external finance), we restrict the analysis to the past decade.

First, we reduce the dimensionality of the a global cycle comprising a multiplicity of typically correlated variables, by building three basic indicators corresponding to the risk appetite of global investors (to capture the influence of capital flows and financial distress), the price of commodities (positively correlated with terms of trade in our commodity-exporting Latin American sample), and global growth (acknowledging the distinct effects from the G7 and Chinese growth; see Levy Yeyati 2009).1 As an illustration, Figure 2 shows the strong correlation between our index of risk appetite and two key channels of transmission of exogenous shocks: foreign portfolio flows (the change in external liabilities from the balance of payments) and the multilateral real exchange rate (the same pattern emerges if we use the nominal multilateral exchange rate).

Figure 2. The incidence of risk appetite on the financial account

Note: REER = real effective exchange rate. Source: Levy Yeyati and Cohan (2011).

Second, we estimate the common component of regional growth, again as the first principal component of the growth rates of individual Latin American countries. Given the obvious asymmetry associated with the influence of the commodities on the terms of trade in Latin American economies during our sample period (positive for South America and, much less so, for Mexico; negative for Central America), we focus on the largest economies in the region with two exceptions: we exclude Venezuela (its growth correlated poorly with the region and the world, and is largely determined by idiosyncratic policy factors) and include Uruguay (its business cycle is highly correlated with that of growth of its neighbours). Thus, our sample includes: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Uruguay.

Figure 3. The common factor in regional growth

Source: Levy Yeyati and Cohan (2011).

Third, we regress the common component in Latin American growth on the three global drivers identified above, without including any reference to local variables, and denote the fit of the regression as the Brookings Global Wind Index or GWI (see our paper for details).

The results are broadly in line with previous attempts to relate Latin American growth with global factors (Izquierdo et al 2008, Österholmand Zettelmeyer 2008), with one important difference: country risk (sovereign bond spreads), which played the stellar role of explaining business cycles in the past (Neumeyer and Perri 2005), is no longer a relevant factor behind economic contagion in the 2000s – even when global risk appetite is excluded from the regression. This does not come as a surprise: the steady process of debt deleveraging, de-dollarisation, and ‘domestication’ in the past decade (the structural change underlying our choice of sample period) significantly reduced foreign currency rollover needs and external financial dependence.

How much of individual countries´ growth variability is explained by our simple Brookings Global Wind Index? At first glance, quite a lot. In tranquil times, more than half, on average; in distressed times, above 80%.

Figure 4. How much growth does the Brookings Global Wind Index explain?

Source: Levy Yeyati and Cohan (2011).

Naturally, the Global Wind Index is not a predictive tool. However, it allows us to characterise Latin American growth in alternative scenarios for our global drivers. Figure 5 illustrates: the stable scenario assumes the values as of October 2011; the bad scenario takes them back to the bottom of the crisis in March 2009; the good scenario assumes end-2010 values. The average growth rates differ as expected.

Figure 5. Growth scenarios: Looking back to paint the future

Source: Levy Yeyati and Cohan (2011).

What does all this tell us about policy? If the economic cycle during crisis times is largely the reflection of elements beyond the policymakers´ control, does it mean that country fundamentals are secondary? Not necessarily. While the effect of the global shock is immediate and massive, the cost in terms of product and the speed of recovery should depend more closely on the policy response. Moreover, in the presence of countercyclical policies, there should be a significant difference between income and consumption (or between growth and welfare): it can be shown that the impact of an output drop on poverty and equity depends negatively on the fiscal space built to compensate it (see for example Calderón and Levy Yeyati 2007).

True, the fiscal space that most governments in the Latin American region have today is rather limited (Figure 6), in part because a large share was consumed in 2009, which indicates that the policy mix will likely be fiscally more contractive and monetarily more expansive than two years ago.

Figure 6. The lower fiscal arsenal

Source: Levy Yeyati and Cohan (2011).

More critically, judging by the influence of a global slowdown that does not seem to revert anytime soon, it is safe to conclude that the Latin American growth prints of the 2000s will hardly repeat in the 2010s, regardless of the quality of local policies.

However, looking at the glass half full, we would expect the end of the bonanza to lead to a reassessment of a development model built on cheap capital and a commodity boom that, in several Latin American economies, was already hitting its own boundaries. It resulted in slow productivity growth, insufficient regional trade integration –or in other words, growing Chinese dependence – and a failure to add value to natural resource–intensive exports. It will lead to a welcome reappraisal of an oversold Latin American miracle.


Calderón, C and E Levy Yeyati (2007), “Zooming in: From aggregate volatility to income distribution”, Universidad Torcuato Di Tella Working Paper 03/2007.

Izquierdo, A, R Romero, and E Talvi (2008), "Booms and Busts in Latin America: The Role of External Factors", RES Working Papers 4569, Inter-American Development Bank, Research Department.

Levy Yeyati, Eduardo (2009), “On emerging markets decoupling and growth convergence”, VoxEU.org, 7 November.

Levy Yeyati, E and L Cohan (2011), Latin America: Innocent bystanders in a Brave New World, Washington, DC: The Brookings Institution, September.

Neumeyer, P and F Perri (2005), "Business cycles in emerging economies: the role of interest rates", Journal of Monetary Economics, 52(2):345-380.

Österholm, P and J Zettelmeyer (2008), “The Effect of External Conditions on Growth in Latin America”, IMF Staff Papers, 55(4):595-623.

Rose, Andrew K (2009), “Business cycles become less synchronised over time: Debunking “decoupling””, VoxEU.org, 1 August.

Wälti, Sébastien (2009), “The myth of decoupling”, VoxEU.org, 27 July.  

1 Risk appetite is proxied by the first principal component of two financial variables: the VIX (implied volatility from options on the S&P500 equity index) and the spread of high-yield US corporate bonds. For commodities, we use the first principal component of the annual variation in wheat, corn, soy, oil and copper. Finally, G7 growth is the simple average of the annual aggregate GDP change. We use a quarterly sample that covers the period 2000:Q1-2011:Q1.



Topics:  Development

Tags:  globalisation, Latin America, decoupling, business cycle, global crisis

Dean, School of Government, Universidad Torcuato Di Tella

Research Associate, CIPPEC; Chief Economist, Elypsis; and Professor, Universidad de Buenos Aires


CEPR Policy Research