III. How Resilient Is Latin America’s Expansion?

International Monetary Fund. Western Hemisphere Dept.
Published Date:
November 2007
  • ShareShare
Show Summary Details

A striking fact of Latin America’s economic history is the frequency and regularity with which growth, once under way, has suffered setbacks. Expansions have often been short lived, ending with crises or prolonged periods of stagnation. Latin American business cycles have tended to be volatile compared with those of both advanced countries and other developing countries (Aiolfi, Catão, and Timmerman, 2006). Long-run growth spells—sustained periods of growth, with only transitory interruptions—have tended to be rarer (Berg, Ostry, and Zettelmeyer, forthcoming). Finally, large output drops have happened more frequently in Latin America than in any other region except for Africa (Becker and Mauro, 2006). Previous issues of this Regional Economic Outlook argued that improvements in Latin American macroeconomic fundamentals since the beginning of this decade, coupled with reforms initiated in the 1980s and 1990s, justified hopes that the current economic expansion could be more sustained than its predecessors.

The recent global financial turbulence, and slowing growth in the United States and other industrial countries, constitute Latin America’s first real test of resilience to external shocks since 2002. How will the region’s expansion, now in its fifth year, react to these shocks? This chapter examines the potential channels of transmission of foreign shocks, and analyzes Latin America’s outlook in the context of a deteriorating external environment, using econometric techniques. It concludes that, based on current WEO projections for external growth, commodity prices, and international credit market conditions, the Latin American expansion is likely to continue, albeit at a somewhat slower pace. However, there are significant risks. These relate, in particular, to variations in the strength of fundamentals within the region, fluctuations in U.S. growth—to which some countries in the region remain highly sensitive, especially in Central America and Mexico—and the possibility that the financial turbulence may widen or deepen, with repercussions for global growth and commodity prices.

Channels of Transmission

Tighter external financing conditions have in the past played a key role in financial crises and output collapses in Latin America. During the 1990s and in the early years of this decade, sudden stops in capital inflows have triggered currency crises, which created widespread insolvencies, as public and private sector liabilities were often denominated in U.S. dollars. This in turn set off bank runs, sharp credit crunches, and, in a few cases, a collapse of the entire financial system. In some cases monetary authorities were able to avoid sharp depreciations, but this came at the price of very high interest rates, taking a significant toll on growth.

In principle, the latest turbulence in world financial markets and its repercussions on the global economy could affect Latin America through four channels: (1) real external demand, as growth in trading partners—particularly the United States—is adversely affected; (2) declines in commodity prices and the terms of trade as a result of a deceleration in world economic activity; (3) financial channels, including through a higher cost of capital and a reduction or reversal in capital flows to the region; and finally (4) a decline in private remittances to the region, as incomes and credit of Latin American workers abroad, particularly in the United States, are reduced. Furthermore, domestic policy reactions can play a role in either mitigating or exacerbating external shocks. In the past, the first three transmission channels have played important roles in precipitating crises in Latin America, while procyclical fiscal policies and monetary policies (for example, in an attempt to reduce unfinanced deficits or defend an exchange rate level) have amplified the shock. To what extent is each likely to be relevant this time? And is there a significant role for remittances as a new channel of transmission, as a recent slowdown in remittances to Central America and Mexico, coinciding with the collapse of new housing construction in the United States, may suggest?

“Current account channels”—export demand and remittances—are surely relevant; but the strength of their effect will depend on the extent to which economic activity slows in trade partner countries. In turn, this will likely depend on the prospects for the U.S. economy. A recent IMF study (the April 2007 World Economic Outlook, Chapter 4) suggests that a “midcycle slowdown” in the United States—as opposed to a full-blown recession—would not significantly affect world growth, particularly given robust growth in emerging markets. Furthermore, Latin America as a whole has become less dependent on U.S. demand. Exports to the United States have declined from 57 percent of total exports in 2000 to 47 percent in 2006—although they remain very high for Mexico (85 percent) and some Central American countries. Finally, while remittances to Latin America have until recently been on a fast-rising trend, their magnitude as a share of GDP remains fairly modest, although not in Central American and some Caribbean countries. A recent IMF study shows that remittances to Latin America have in the past not been strongly influenced by the U.S. economic cycle (Box 8). However, past patterns may not be a particularly good guide to the future. This is in part due to the role of the weaker housing market in the United States in the recent turbulence, and the importance of the U.S. housing and construction sectors as an area of employment and—increasingly—investment for immigrants from the region.

Box 8.Are Remittances to Latin America Influenced by the U.S. Business Cycle?

While there is evidence that remittances may help to smooth adverse home country shocks, little is known about the potential impact on remittances of shocks in the source country. Experience in the first half of 2007, when a remittance slowdown to some Latin American countries coincided with a moderating U.S. economy, suggested a procyclical link (see first figure). However, a closer analysis of the data suggests that such links remain weak.

In a recent IMF study, Roache and Gradzka (forthcoming) examine the link between remittances to 15 Latin American countries, using quarterly data between 1994 and 2007 and 19 indicators of the U.S. business cycle, using a range of methods. Simple correlations between seasonally adjusted and deflated remittances and U.S. cyclical indicators, averaged across countries and indicators, turn out to be close to zero (left panel of second figure). Similar results were obtained using more narrowly defined groups of U.S. indicators, such as employment growth, and lagging the U.S. indicators by 1 to 4 quarters (right panel of second figure). In addition to averages, the figure shows individually the correlations for Mexico as the largest recipient of remittance (in absolute terms, not in relation to GDP), and El Salvador, the country with the highest average correlation.

Regressing remittances on a range of U.S. indicators, with lags and controlling for the domestic business cycle, also failed to detect a clear and positive relationship in most cases. In a final step, one can ask whether it is possible to detect a “common cycle” among U.S. economic activity and remittance flows. To do this, a “dynamic factor analysis” was undertaken for the six countries with the most data and four U.S. business cycle indicators, including GDP, two employment measures, and housing starts (see table). The more positive the coefficients in the table, the more sensitive are remittances to the U.S. cycle. The results again suggest little or no sensitivity for most countries. Mexico may be an exception, as remittances to Mexico appear related to a broad measure of the U.S. economic cycle; indeed, recent data for Mexico continue to suggest an emerging linkage (see Roache and Gradzka, forthcoming, for details)

Remittances and U.S. Housing Starts

Sources: Haver Analytics; national authorities; IMF, International Financial Statistics ; and IMF staff calculations.

Correlations Between Sample Countries and U.S. Indicators 1/

Source: Author's calculations.

1/ The highest (lowest) datapoint represents the maximum (minimum) average correlation between a particular country's remittances and a range of U.S. economic indicators.

It is possible that stronger linkages will emerge, as the rapid pace of recorded remittance growth begins to slow. However, it is also possible that migrants may engage in a form of consumption smoothing by sending home a fixed amount each month; this has also been proposed as a possible reason for the countercyclical response of remittances to conditions in the home country (see Sayan, 2006).

Dynamic Common Factor Model: Sensitivity of Remittances 1/
Dominican RepublicEl SalvadorU.S. GDPU.S. employment in:
Hodrick-Prescott cyclical component model - estimated factor loadings
U.S. cycle-0.13-0.10-*0.29***0.32***0.24***
Source: Roache and Gradzka (forthcoming)

The coefficients are the factor loadings (or sensitivity) of each variable in the observable vector to the unobservable factors which represent the U.S. cycle and a common remittance effect. These results were based on estimations using the cyclical component of the Hodrick-Prescott filter. Significance at 1, 5, and 10 percent levels indicated by ***, **, and *, respectively. Coefficient standard errors are in brackets.

Source: Roache and Gradzka (forthcoming)

The coefficients are the factor loadings (or sensitivity) of each variable in the observable vector to the unobservable factors which represent the U.S. cycle and a common remittance effect. These results were based on estimations using the cyclical component of the Hodrick-Prescott filter. Significance at 1, 5, and 10 percent levels indicated by ***, **, and *, respectively. Coefficient standard errors are in brackets.

Similarly, a significant reduction in commodity prices would surely affect the region. Estimates in the April 2007 Regional Economic Outlook suggested that a decline in relevant commodity export prices by 5 percent in one quarter would reduce aggregate growth in six large Latin American economies (Argentina, Brazil, Chile, Colombia, Mexico and Peru) by about one-third of a percentage point after two quarters. However, the relevant prices of goods exported by Latin American countries continue to be high, with only mild declines currently projected over the medium term. A significantly larger decline is likely only in the context of a sharp downturn in world growth, examined in the next section.

Mexico and U.S. Business Cycle Indicators

(Year-on-year percent change)

Source: National authorities.

Size of Remittances in Latin America, 2006
Billions of US$Percent of
GDPFDI inflows
South America
Central America
Costa Rica0.52.373.8
El Salvador3.318.1667.3
Dominican Republic2.78.7324.7
Sources: National authorities; IMF, International Financial Statistics; and IMF staff calculations.
Sources: National authorities; IMF, International Financial Statistics; and IMF staff calculations.

Financial vulnerabilities are reduced compared with a few years ago, although they have not disappeared. Currency mismatches associated with foreign currency borrowing have come down in several countries, and external positions are generally stronger (see Box 3). Fiscal positions have also improved, and structural primary balances are generally in surplus (see Chapter 4). Public sector financing requirements are generally lower, with maturing public debt in 2007 below 10 percent of GDP in most large countries. There is also little indication of direct exposure of Latin American banks to the U.S. subprime market or a clear cause for concern over domestic bank solvency at this stage (Box 9). However, both fiscal and external positions in the region are expected to deteriorate this year, and recent rapid credit growth may have led to financial sector vulnerabilities that are not yet visible.

Corporate health among larger companies has also improved. Comparing the period of 2002–06 with that of 1998–2001, the publicly listed firms in the region have generated higher earnings, become more liquid, and face lower gross rollover risks. These improvements mean that the corporate sector—at least, publicly listed firms—may be in a stronger position than before to cope with adverse external shocks. While larger and better-rated companies may be able to absorb the impact of shocks more easily, smaller and lower-rated companies could still be vulnerable to funding difficulties if credit conditions tighten. More broadly, continued corporate health will depend on macroeconomic stability to keep down spreads related to country and inflation risk.

Box 9.Will the U.S. Subprime Crisis Affect Bank Lending in Latin America?

There is little indication of direct exposure of Latin American banks to the U.S. subprime market, or clear cause for concern over bank solvency. Market-based default estimates for publicly listed financial institutions in the region have stayed low in absolute terms and below their historic values. The aggregate liquidity condition of the banking systems in the larger Latin American countries in the first part of 2007 was generally strong, providing some buffer for the subsequent turmoil in the global financial markets. There was little sign of liquidity problems in reaction to the turbulence (a brief spike in interbank rates in Argentina was addressed by central bank liquidity provisions to local banks).

The impact on bank lending through indirect channels is also expected to be limited as long as the global financial turbulence does not widen or deepen. Bank lending could be adversely affected if mark-to-market losses originating from adverse price movements in the local equity, bond, and foreign exchange markets erode profitability and net worth. However, as local markets have generally stabilized, with limited or no net losses in asset prices so far, the impact on profitability and net worth of banks is likely to be small.

Estimated Default Frequency of Financial Institutions

(In percent) 1/2/

Source: Moody's KMV.

1/ Includes publicly listed banks and other financial institutions. One-year-ahead median values.

2/ The approximate Moody's rating equivalents for these frequencies are as follows: Ba2 = 0.08%; Caa1 = 0.60%; Caa2 = 0.90%; Caa3 = 1.70%; Ca = 5.5%.

Second, the credit crunch, reflected in the higher credit default swap (CDS) spreads, and subprime losses experienced by banks in industrial countries could indirectly affect some Latin American countries, where international banks have played a role in funding local offices’ domestic operations. However, recent credit growth appears to be mainly funded by local deposit growth (Chapter 2), and Latin American banking systems’ foreign liabilities generally account for a small and declining part of their balance sheets. Finally, a persistent deterioration in broad credit conditions would likely have a more pronounced effect on smaller financial institutions, which are more reliant on wholesale funding from the local capital markets. This risk is mitigated by the ample capacity of most central banks to provide liquidity support in the current context of high reserves and generally stronger private and public balance sheets.

International Bank CDS Spreads 1/

(Basis points)

Source: Bloomberg, L.P.

1/ CDS (credit default swap) five-year senior debt in US dollars for Citigroup, and in euros for the other banks.

Asset Share of International Banks by Host Country, 2006

(In percent of total domestic banking system assets)

Sources: National authorities; and IMF staff estimates.

Latin American Banks: Foreign Liabilities

(In percent of total liabilities) 1/

Sources: IMF, International Financial Statistics ; and the Bank of Mexico.

1/ The 2007 figures for Panama and Jamaica are as of 2007Q1 and 2006Q4 respectively.

Note: This box was prepared by Jingqing Chai.

Banking Systems: Net Open Position in Foreign Exchange

(In percent of regulatory capital) 1/

Sources: National authorities; and IMF staff estimates.

1/ Distribution of net open position in foreign exchange as a share of capital, based on a sample of 9 countries.

In the past, domestic vulnerabilities have often amplified crises not only by exacerbating capital outflows but also by hampering the ability of domestic policy to buffer external shocks. High debt and deficits and binding public borrowing constraints often led to procyclical fiscal policy. Lack of monetary policy credibility and “fear of floating”—usually a consequence of currency mismatches in balance sheets—meant that monetary policy was also typically tightened when the external environment weakened. Today, in contrast, the prospect for effective domestic policies—particularly, monetary policy—has improved with a shift to more flexible exchange rates and relatively well-contained inflation. With currency mismatches less of a concern, several central banks allowed exchange rates to depreciate in response to the rise in risk premiums after August 2007. And with better anchored inflation expectations, the concern that exchange rate depreciations might be passed through to domestic prices does not appear to have influenced monetary policy decisions following the August turbulence.

Private Dollarized Credit

(In percent of total private credit, end year) 1/

Sources: National authorities; and IMF staff estimates.

1/ Unweighted averages for 12 countries. For 2007, rates are based on latest available data.

It is sometimes argued that the favorable external environment for Latin America in the last few years—particularly high commodity prices and strong U.S. and global growth—is mainly responsible for improved fundamentals, and so these could unravel once external conditions deteriorate. While growth in Latin America is indeed sensitive to the external environment, this line of argument is too simplistic. First, Latin America has benefited from strong policies and genuine institutional improvements in recent years as well as favorable external conditions. For example, monetary policy frameworks are more credible and sophisticated, structural fiscal balances have improved even in countries that are not major commodity exporters (see Chapter 4), and many countries have implemented financial sector reforms (see Chapter 5). Second, even if Latin America’s improved economic performance were purely a result of good luck, the resulting lower public debt, higher net foreign assets, and better composition of public and private balance sheets have put the region in a better position to withstand foreign shocks than in the past. External investors so far appear to agree with this view, as evidenced in the modest increase in sovereign spreads in most Latin American countries, particularly compared with the much sharper rise in high-yield corporate bond spreads in the United States since August.5

Short-Term Public Sector Debt at Remaining Maturity (Debt Exposed to Rollover Risk) 1/

(In percent of GDP)

Source: IMF staff calculations based on national authorities' data.

1/ Short-term debt (end of previous year) plus maturing medium- and long-term debt (current year). 2007 data are projections.

Latin America: Developments in Corporate Fundamentals

Source: Corporate Vulnerabilities Utility.

1/ Interest coverage ratio equals earnings before interest and tax (EBIT) divided by interest expense.

U.S. High-Yield Bond Spreads and EMBI Spreads

(In basis points)

Source: Bloomberg, L.P.

This said, significant risks remain. First, fundamentals vary significantly across the region. Currency mismatch problems associated with dollar liabilities continue in some countries and, as described in Chapter 2, fast-rising government expenditures are shrinking fiscal surpluses, weakening current accounts, and contributing to inflationary pressures in several countries. Furthermore, the external crisis may widen or deepen. One risk arises from the possibility that the crisis may engulf a major emerging market country, leading to a larger reassessment of risks for the emerging market asset class than has been the case so far. Another arises from the possibility of a recession in the United States, which could significantly slow economic growth in other regions, lead to a further increase in risk premiums worldwide and weaken commodity prices. The next section employs econometric tools to study the likely effects of the August turbulence both under a relatively benign baseline for growth outside the region, and under more adverse conditions.

Scenario Analysis

This section is based on an econometric framework developed in Österholm and Zettelmeyer (2007) for analyzing the relationship between external factors and growth in Latin America, estimated using data from 1994 until the second quarter of 2007.6 The model focuses on six variables: an export-weighted index of world growth outside Latin America, in which the United States has a current weight of about 0.55; the U.S. short-term interest rate; the U.S. high-yield bond index as a proxy for conditions in U.S. and global credit markets; an index of net-export-weighted commodity prices relevant to Latin America; an index of aggregate growth in six large Latin American countries (Argentina, Brazil, Chile, Colombia, Mexico, and Peru); and the JPMorgan Latin America Emerging Market Bond Index (Latin EMBI). The model can generate both “unconditional forecasts,” in which all these variables are forecast after conditioning only on past data, and “conditional forecasts,” in which the future paths of some variables are taken as given and forecasts are derived for the remaining variables in the model. This provides an intuitive and convenient way to study Latin American growth under alternative scenarios.7

The baseline scenario is based on the projections for world growth and financial variables presented in the October 2007 World Economic Outlook. The WEO provides quarterly forecast paths for four of the six variables in the model: world growth, Latin American growth, U.S. short-term interest rates, and commodity prices. Since the objective here is to forecast Latin American growth, the WEO forecast of this variable is ignored. WEO forecasts for world growth (transformed into an export-weighted index), short-term interest rates, and commodity prices are used as “conditioning paths.” For the high-yield corporate bond spread, a gradual decline is assumed, albeit to levels above those preceding the August turbulence. For the Latin EMBI, no assumption is made except that the third-quarter value in 2007 (the first quarter within the forecasting period) is set equal to its actual value in that quarter. Hence, the baseline forecast of Latin American growth is conditioned on (1) the actual “shock” to credit markets observed in the third quarter of 2007; and (2) paths for world output, commodity prices, and U.S. interest rates, corresponding to the WEO’s baseline forecast. As discussed in Chapter 1, the latter envisages fairly robust world growth in 2008, and only minor declines in commodity prices.

LA6 Growth: Baseline Forecast

(In percent)

Sources: WEO; and IMF staff calculations.

Conditioning Paths for Baseline and Downward Scenarios

Trade-Weighted External Demand

(Percent change with respect to the same quarter of the previous year)

Sources: WEO; and IMF staff calculations.

The model predicts that in this baseline scenario, Latin American growth would slow only modestly in 2008, to 4.3 percent, from 5.0 percent in 2007. The model was estimated over a sample period that includes the Latin American crises of the 1990s and, as such, has a tendency to show large reactions of growth to deteriorations of the external environment—and in particular to higher financial costs. Nonetheless, the baseline forecast broadly bears out the relatively optimistic view of regional growth laid out in the preceding section’s analysis of potential transmission channels for external shocks.8 However, the main reason is that the external baseline on which this forecast is conditioned itself remains relatively favorable. It embodies a slowdown in U.S. growth but without a recession. Moreover, the U.S. slowdown is assumed to be broadly offset by growth in the rest of the world, so that export-weighted external growth for Latin America continues essentially unchanged over the forecast period, not far below average external growth during 2004–06. Commodity prices are assumed to remain relatively high, in line with WEO projections, and corporate credit markets to return gradually to tranquil conditions, albeit with some permanent repricing of risk. Finally, the initial jump of the EMBI in the third quarter of 2007 on which the forecast is conditioned is small relative to the increases seen in previous crisis episodes. Through this channel, the simulation captures some of the improvements in Latin American fundamentals that are too recent to be reflected in the structure of the model itself.

The question remains how Latin America would react if external conditions were to deteriorate more than envisaged in the WEO baseline. An alternative scenario was constructed around the assumption that the United States suffers a recession, with negative growth in the first and second quarters of 2008, followed by a slow recovery.9 Given the high weight of the United States in the world growth index weighted by Latin America’s exports, this translates into a significant slowing of external demand growth, from 3 percent in 2007 (on an export-weighted basis) to 2.1 in 2008, about 1 percentage point below baseline.10 In line with the assumption that the United States slides into recession, a significant further increase of corporate spreads is assumed, to about 700 basis points, together with a reduction in U.S. short-term interest rates, as the Federal Reserve reacts to the recession. No assumptions for commodity prices or the Latin EMBI are made, that is, the model is allowed to predict these variables along with aggregate Latin American growth.

LA6 GDP Growth: Downside Scenario (U.S. Recession and Credit Crunch)

(In percent)

Sources: WEO; and IMF staff calculations.

The main result is that a scenario of this type would hit Latin American growth very significantly, although it would fall short of causing a full-blown recession across the region. Specifically, the model forecasts year-on-year quarterly growth falling from 5¼ percent in the second quarter of 2007 (the last observed value) to about 4.5 percent in late 2007, then, much more steeply, to a low of 1.8 percent in the third quarter of 2008. This implies annual average growth of about 2½ percent in 2008, around 1.8 percentage points below baseline. In addition to the assumed 1 percentage point fall in external growth, this is due to the sharp tightening of external credit markets—leading to a forecast rise in the Latin EMBI to around 600 basis points in late 2007 and early 2008—as well as a steep decline in commodity prices in this scenario. According to the model, commodity prices would fall—reflecting lower global growth—by almost 20 percent between late 2007 and mid-2008, before beginning to recover.11


A period of institutional reforms and stronger policy frameworks has left the LAC region better prepared for times of global turbulence. More flexible exchange rates have become the first line of defense against abrupt changes in capital flows, while a greater commitment to low inflation—including with the adoption of inflation-target frameworks by many countries in the region—has brought predictability and transparency to policies. Fiscal policies have also improved, leading to primary fiscal surpluses and lower debt-to-GDP ratios in the region. At the same time, governments have been able to shift their financing strategy away from bonds linked to short-term interest rates and the exchange rate toward fixed-rate and inflation-indexed paper, thus reducing the sensitivity of the public debt to temporary changes in financial conditions.

Nevertheless, important policy challenges remain, particularly if the external environment deteriorates significantly. Downside risks to U.S. growth have increased recently, raising the likelihood of a larger external demand shock to Latin America. A significant tightening in corporate credit, spurred for example by more market turbulence, could spill over to local markets in the region and cut short the ongoing credit expansion, hurting investment and growth. With improved public and private sector balance sheets and more credible policies, many of the large economies in the region are now in a position—in some cases, for the first time in decades—to respond to such a shock through countercyclical monetary policy. Indeed, a few countries may have created sufficient fiscal space to respond with countercyclical fiscal policy, should this become necessary, or at least to maintain essential spending rather than implement cuts as in earlier crises.

Even under the baseline scenario, countries face significant challenges. As discussed in more detail in the next chapter, fiscal authorities will need to reduce the pace of current expenditure growth to prevent structural fiscal surpluses from quickly turning into deficits. Monetary authorities should remain vigilant, as a mild reduction in growth in 2008 may not offset inflationary pressures from food-price increases and supply constraints. Although uncertainty about the future path of inflation has increased, improved monetary policy frameworks in many countries in the region should help them to rise to this challenge.

5Traditionally, the EMBI and Latin EMBI spreads have reacted about one-for-one to shocks to the U.S. high-yield bond corporate spread (González Rozada and Levy Yeyati, 2006; Österholm and Zettelmeyer, 2007).
6See also related work by Izquierdo, Romero, and Talvi (forthcoming).
7Technically, the model consists of a “Bayesian Vector-Autoregression.” (See Österholm and Zettelmeyer, 2007, and April 2007Regional Economic Outlook for details.) The only difference between the models discussed there and the model employed in this section is that this version (1) was re-estimated on a sample including the first two quarters of 2007, and (2) uses an export-weighted index of world GDP, to better capture the impact of expected divergences between U.S. growth and economic activity elsewhere in the world. (The model discussed in the April 2007 Regional Economic Outlook used the IMF’s GDP in PPP terms as an index of world growth).
8The model-based baseline forecast is almost identical to the forecast for Latin American growth shown in Chapter 2, which was generated independently, built up from country-level forecasts by IMF country teams.
9On an average annual basis, the United States is assumed to grow by 0.8 percent in this scenario, compared with projected annual growth of 1.9 in the baseline scenario.
10This reduction also reflects some assumed slowing of the world economy outside the United States in the event of a U.S. recession, in line with the model of Bayoumi and Swiston (2007), and consistent with the risks to world growth analyzed in the October 2007 WEO.
11Note that this sharp fall and incipient recovery is not an assumption but rather a model forecast conditioned on the assumed paths of world growth and financial variables.

    Other Resources Citing This Publication