By the end of 2013, growth in Latin America had begun to decelerate sharply (Munyo and Talvi 2013) as China’s growth faltered, commodity prices levelled off before starting to fall, and expectations of higher interest rates in the US slowed down capital inflows into the region. In response, in October 2013 the Central Bank of Chile cut the policy interest rate by 25 basis points (Figure 1). A year later, the policy rate had been cut by 200 basis points (from 5% to 3%). In sharp contrast, Brazil’s central bank increased the policy rate in October 2013 and by mid-2015, the cumulative increase had reached 700 basis points (Figure 1). Why would Brazil follow a procyclical monetary policy (i.e. raise policy rates in bad times), which, after all, can only aggravate the recessionary conditions, rather than embarking on a countercyclical monetary policy as Chile did (and most industrial countries would do under similar circumstances)?
Figure 1. Policy interest rate and real GDP in Brazil and Chile
This is just but the latest example of a macroeconomic policy conundrum faced by Latin America – and emerging markets more generally – for more than 40 years. As much as they would like to, Latin American countries find it extremely difficult to pursue expansionary monetary and fiscal policy in times of distress (i.e. countercyclical macroeconomic policies). When it comes to monetary policy, the problem arises because a fall in economic activity is typically accompanied by currency depreciation, which may lead to higher inflation, exacerbate capital outflows, and cause financial distress to public and private entities carrying dollar debts. Hence, faced with a sharp depreciation of the currency (as has been the case in Brazil recently), the central bank may have no choice but to raise the policy rate to defend the currency (Vegh and Vuletin 2012, 2013). In terms of fiscal policy, when access to international credit markets tightens in bad times, many emerging markets find themselves in the unenviable position of having to cut government spending and/or raise taxes in the midst of recessionary conditions (Frankel et al. 2011).
Fortunately, macroeconomic policy procyclicality in emerging markets is not an inescapable trap. In fact, over the last 15 years or so, about one-third of developing countries have managed to switch from being procyclical to countercyclical (i.e. have ‘graduated’) in terms of both monetary and fiscal policy. In the case of fiscal policy, this remarkable turnaround can be traced back to an improvement in fiscal arrangements and institutions (such as fiscal rules requiring a zero structural balance, as in Chile).1 In the case of monetary policy, in earlier work (Vegh and Vuletin 2012) we argue that as central bank independence boosts credibility and macroprudential measures reduce large currency mismatches, large depreciations become less harmful, which allows central banks to lower policy rates in bad times.
The proof, however, is in the pudding, and graduation would be of little value if it did not enable countries to act countercyclically in times of crises. To this effect, rather than focusing on average policy behaviour over the business cycle, this column looks specifically at policy responses to GDP crises in Latin America over the last 40 years, both on the fiscal and monetary fronts, to ascertain whether they have become more countercyclical over time. In other words, we want to see whether Chile’s countercyclical response to the latest slowdown in the world economy is the rule rather than the exception.2
GDP crises in Latin America
For this study, our sample of Latin American countries comprises what is commonly referred to as LAC-7 (Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela) and Uruguay. The combined GDP of these eight economies account for more than 90% of the region’s GDP.
Analysing policy responses to ‘crises’ naturally requires defining a ‘crisis’. For simplicity, we focus exclusively on the behaviour of real GDP and define a crisis as beginning in the quarter in which real GDP falls below the preceding four-quarter moving average and ending in the quarter in which real GDP reaches the pre-crisis level. Using this definition, we identify 34 crises in our eight Latin American countries.3 The average duration of the crises is 11 quarters and the average intensity (measured as the fall in the level of GDP from the start of the crisis to the trough) is 8.6%.
If we break our sample into before and after 1998, how have the frequency, duration, and intensity of crises in Latin America changed?4 Figure 2, Panel A shows that, on average, GDP crises have become less frequent after 1998.5 The average duration and intensity of crises in Latin American have also fallen in the post-1998 period (Figure 2, Panel B). At this point, however, we cannot say too much more because the fall in any of the three (frequency, duration, and intensity) could be simply due to exogenous factors. External shocks, for instance, could have become less frequent over time and of shorter duration, thus causing less economic hardship. Based on a detailed analysis, however, we argue in our paper that while the lower frequency of crises may indeed be the result of ‘good luck’ (given that crises in Latin America are mainly explained by external factors, such as commodity prices and global liquidity), the reduction in the severity and duration of crises is instead explained by a shift, on average, from a procyclical policy response to crises (i.e. contractionary monetary/fiscal policy during crises) to a countercyclical one. In other words, we find significant statistical evidence to the effect that more countercyclical fiscal/monetary policy leads to shorter and less severe GDP crises.
Figure 2. Latin America: Average frequency, duration, and intensity of GDP crises
Policy responses
We begin by looking at the fiscal policy response both on the spending and taxation side. Table 1 shows, for each of the eight countries in the sample, the average correlation during crisis periods between the cyclical component of government spending and a tax index, on the one hand, and GDP, on the other, before and after 1998.6 The table is very telling, as it pinpoints two countries (Chile and Mexico) that have clearly switched from having a procyclical fiscal policy response before 1998 to a countercyclical policy response after 1998. Not coincidentally, these are countries that are often hailed in the financial press for having considerably improved their macroeconomic management over the years. Some other countries, like Brazil and Peru, have notably improved in some fiscal dimension (spending or tax). The other four countries tend to show a procyclical fiscal response after 1998.7 In particular, Argentina, Uruguay, and Venezuela all show pronounced procyclical spending responses.
Table 1. Latin America: Country cyclicality of fiscal and monetary policies during GDP crises
Table 1 also shows the monetary policy response by calculating the average correlation during crises between the cyclical component of a policy rate and/or short-term market rate and real GDP.8 The four countries that exhibit countercyclical monetary policy response after 1998 are Brazil, Chile, Colombia, and Peru. As illustrated in Vegh and Vuletin (2014a), much of the 1998 behaviour is captured by the monetary policy response to the 2008-2009 Global Crisis when policy rates were reduced sharply in Chile, Colombia, Peru, and less dramatically, in Brazil.9 As Figure 1 illustrates, however, Brazil’s monetary policy response to the current slowdown appears to be procyclical, indicating that the process of graduating is not irreversible either and countries can easily backslide into procyclical policies.
While in light of the different fiscal and monetary ‘graduation’ stories described above it is difficult to assess the overall policy stand in the region, it seems that, on average, Latin America has improved fiscal and monetary management. While in the pre-1998 period the spending policy response was clearly procyclical (with a correlation coefficient of 0.56), it fell to almost half (the correlation is 0.27) in the post-1998 period. In terms of tax policy, the shift from procyclicality to countercyclicality between pre-1998 and post-1998 is even more dramatic (from -0.27 in the pre-1998 period to 0.08 in the post-1998 period). In terms of monetary policy, the shift from procyclicality to countercyclicality between the pre-1998 and post-1998 periods is also quite evident (from -0.28 in the pre-1998 period to 0.05 in the post-1998 period).
In sum, we find that, on average, Latin America's fiscal and monetary policy responses to crises have shifted from being procyclical before 1998 to being acyclical or even countercyclical after 1998.
In this sense, therefore, we could argue that, on average, Latin America has graduated in terms of the policy response to crises. This average response, however, masks a great deal of heterogeneity across countries. On the one hand, we have countries such as Chile and Brazil that have switched from pro- to countercyclical policy responses on both the fiscal and monetary front (though Brazil seems to have reverted back to procyclical monetary policy in the current cycle). On the other hand, we have countries such as Argentina and Uruguay that have shown consistent procyclical fiscal and monetary policy responses throughout the entire sample, or a country such as Venezuela (for which we do not have data before 1998), which has been procyclical in both its monetary and fiscal policy response after 1998. In other words, while we find helpful to characterise the average behaviour for the region, we cannot overemphasise the heterogeneity across the different countries in the region.
Has countercyclical policy worked?
Have countercyclical policies been responsible for the shorter duration and lower intensity of GDP crises in Latin America in the post-1998 period? Panel A in Figure 3 shows the relationship between the cyclicality of spending policy (as captured by the correlation between the cyclical components of government spending and GDP) and the duration of crises. The relation is positive – implying that the more countercyclical is fiscal policy, the lower is the duration of the crisis – and statistically significant. This suggests that countercyclical fiscal policy has indeed helped in reducing the duration of crises in Latin America.10
Panel B shows that the same is true of the intensity of the crisis: the more countercyclical the spending policy, the lower the fall in GDP from start to trough. Panels C and D show similar findings to those of panels A and B when focusing on tax policy. While the statistical significance decreases, the relationship indeed supports the idea that the more countercyclical the tax policy, the shorter and less intense the crises. Finally, Panels E and F in Figure 3 show the relation between the cyclicality of monetary policy and the duration and intensity of crises. In both cases the relation is, as expected, negative, implying that a more countercyclical monetary policy reduces both the duration and intensity of crises.
Figure 3. Latin America: Cyclicality of fiscal and monetary policies and duration and intensity of GDP crises
Note: †, *, ** and *** indicate statistically significance at the 15%, 10%, 5%, and 1% levels, respectively.
In sum – and leaving aside for the moment potential endogeneity problems – the evidence suggests that countercyclical spending policy has contributed to lessen both the duration and intensity of crises in Latin America.
While the evidence is weaker for tax and monetary policy, there is also some evidence that it has contributed to reducing the duration and intensity of crises. In this light, we would interpret the fact that both the average duration and intensity of crises in Latin America has fallen in the post-1998 period as partly reflecting sounder macroeconomic policies in Latin America.
Causation
Needless to say, since correlations do not imply causation, we needed to worry about two possible sources of endogeneity in our analysis. The first was that expected changes in fiscal/monetary policy could cause the crises themselves (i.e. in anticipation of fiscal problems, market agents may withdraw funds and provoke a crisis). The second was that the duration/intensity of the crises could drive the policy ‘responses’ (i.e. in a severe crisis, tax revenues may fall precipitously and lead to a fiscal contraction, which could be misconstrued as a procyclical fiscal policy response).
To address the first potential source of endogeneity, we constructed a detailed narrative that shows that most of the GDP crises in our sample were driven by external factors (oil shocks in the 1970s, increases in global interest rates in 1979-81, Asian 1997 and Russian 1998 crises, and 2008 Global Crisis). In other words, by and large, the GDP crises in our sample were not caused by domestic fiscal/monetary policies. To address the second possible source of endogeneity, we computed ‘readiness’ indices (i.e. measures of fiscal and monetary ‘space’ available to policymakers to undertake countercyclical measures). Since, by construction, such indices are unrelated to subsequent events, we could use them as ‘instruments’ for monetary and fiscal policy responses. Our results indicate that, indeed, the more countercyclical the fiscal and monetary policies were, the shorter and less intense were the GDP crises.
Policy conclusions
While, on average, our study concludes that Latin America has graduated in terms of monetary and fiscal responses to crises (that is, countercyclical policy responses to crises are more common in the last 15 years than before), it is fair to say that this average masks a great deal of heterogeneity across countries. While countries like Chile and, to a large extent, Brazil and Mexico have made important strides in escaping the procyclical trap of the past, other countries such as Argentina, Uruguay, and Venezuela continue to be procyclical. Further, the fact that in the current slowdown Brazil has reverted back to raising interest rates shows how ephemeral policy gains may be in this area. Clearly, Latin American countries must continue to build sound and credible fiscal and monetary institutions that enable them to not only graduate in terms of policy response to crises but also consolidate those gains as time goes by.
References
Frankel, J (2011), "A Solution to Fiscal Procyclicality: The Structural Budget Institutions Pioneered by Chile," Journal Economia Chilena, Central Bank of Chile, Vol. 14, pp. 39–78.
Frankel, J, C A Vegh, and G Vuletin (2011), “Fiscal policy in developing countries: Escape from procyclicality”, VoxEU.org, 23 June.
Munyo I and E Talvi (2013), “Are the golden years for Latin America over?”, VoxEU.org, 7 November.
Vegh, C A and G Vuletin (2012), “Graduation from monetary policy procyclicality”, VoxEU.org, 22 August.
Vegh, C A and G Vuletin (2013), “Tax-policy procyclicality”, VoxEU.org, 1 October.
Vegh, C A and G Vuletin (2014a), "The Road to Redemption: Policy Response to Crises in Latin America," IMF Economic Review, Vol. 62, pp. 526–68.
Vegh, C A and G Vuletin (2014b), “The social impact of fiscal policy responses to crises,’ VoxEU.org, 12 June.
Endnotes
[1] See Frankel (2011) for a detailed discussion of the Chilean case.
[2] The analysis follows Vegh and Vuletin (2014a).
[3] The beginning of our sample period varies from country to country, depending on data availability, with Argentina starting in 1970.Q1 and Venezuela in 1998.Q1. All samples end in 2013.1 and thus do not include the current slowdown which, within our framework, could only be properly analyzed after it ends.
[4] While admittedly arbitrary, the choice of 1998 seems a natural one because (i) through formal regressions we can detect a shift in fiscal policy in the late 1990’s; (ii) 1998 is a year without any crises and thus provides a clean break; and (iii) we needed to have a reasonable long window for the “after” period.
[5] Frequency is measured as number of quarters in which at least one country is in crisis over the total number of quarters in the sample.
[6] The cyclical components have been estimated using the Hodrick-Prescott filter. For government spending, a positive (negative) correlation indicates procyclical (countercyclical) spending policy, whereas for the tax index, a negative (positive) correlation indicates procyclical (countercyclical) tax policy. For details on the tax index, see Vegh and Vuletin (2013).
[7] We should note that Colombia did not have crises before 1998 and we do not have data for Venezuela before 1998.
[8] In this case a positive (negative) correlation indicates a countercyclical (procyclical) policy response.
[9]Chile is the most prominent case, with the central bank lowering the monetary policy rate by 775 basis points from 8.25% in December 2008 to 0.5% in July 2009.
[10] In Vegh and Vuletin (2014b), we show that countercyclical fiscal policy has also improved the behavior of social indicators, such as the poverty rate, income inequality, unemployment, and domestic conflict.
