III. External Financial Linkages: What Drives Uruguayan Sovereign Spreads?
By Gustavo Adler and Stephanie Eble
1. This chapter examines the determinants of Uruguayan sovereign spreads. It analyzes whether sovereign spreads have been largely driven by economic fundamentals or by external factors. It also assesses the relationship between the spreads of Uruguay and those of other emerging market economies (EMEs)—particularly neighboring countries—and the change in such relationship since the 2002–03 crisis.
2. The results suggest that, while fundamentals explain part of the variance of sovereign spreads, external factors also play an important and—surprisingly—growing role. Moreover, external factors have become more important since the 2002-03 financial crisis, following Uruguay's loss of investment grade.1 While in the late 1990s Uruguay was largely insulated from regional and EME shocks, since the crisis Uruguayan spreads have moved closely with those of neighboring countries. Similarly, unlike the 1990s when Uruguayan spreads were at the level of-and co-moved with Chilean spreads (an investment grade economy), after the crisis Uruguay's spreads shifted to the level of-and began co-moving with other non-investment grade countries in the region. Econometric estimates point to heightened financial spillovers from other EMEs, following the downgrading of Uruguay's credit rating, suggesting a change in investors' perception of Uruguay's vulnerability to external financial conditions.
3. Today, despite important achievements, Uruguay is still perceived by investors as more vulnerable to global and regional shocks than before the crisis. Thus, despite important improvements in the last years, it will be essential to sustain sound policies to further delink Uruguay's fortunes from those of other EMEs and global conditions. Continuing to strengthen the macroeconomic framework and implementing pending structural reforms will be key in this regard. Regaining investment grade status would also likely help insulate Uruguay to a large extent from a possible turnaround in global conditions and from regional shocks.
B. Stylized Facts
4. Following the 2002–03 crisis, Uruguay's sovereign spreads have declined significantly. After peaking in the midst of the crisis, spreads sharply dropped with the completion of the debt restructuring in mid-2003. Since then, they continued a sharp downward trend, reaching pre-crisis levels by end-2005 (Figure 1). This has allowed the government to tap markets at very favorable rates and to significantly improve the profile of public debt.
5. The marked fall in sovereign spreads has been accompanied by significant improvements in country fundamentals. Strong fiscal consolidation, high output growth, and the recovery of the real exchange rate from its post-crisis lows have contributed to a sharp reduction of public debt as a share of GDP, alleviating solvency concerns. In addition, liquidity indicators have improved substantially, with reserve coverage of short-term external debt and foreign currency deposits well above pre-crisis levels, partly reflecting a sharp reduction in short-term debt (Figure 2). Furthermore, the recovery of the financial system has reduced contingent fiscal liabilities.
Figure 2.Country Fundamentals and External Factors, 1996-2006
1/ Gross international reserves over ST external debt and foreign currency deposits
2/ As percent of GDP.
6. At the same time, Uruguay has benefited from a very benign global environment. Global financial conditions have improved markedly since 2002. Long-term U.S. interest rates have declined significantly, financial market volatility—a proxy for liquidity conditions—has reached historical lows, and U.S. corporate high yield spreads have also fallen.2 Reflecting this favorable environment, emerging market sovereign spreads have followed a steady downward trend across the board, reaching historical lows in 2006.
7. During the late 1990s Uruguay's sovereign spreads were largely insulated from financial shocks in other EMEs. They remained significantly below those of neighboring countries—about 500 bps below the Latin American EMBI during 1998–2001—and were only weakly correlated with them. This is likely to have reflected Uruguay's investment grade and the general perception of Uruguay as a safe heaven for investment in the region (Figure 3). A similar pattern is found when compared to the global EMBI.
Figure 3.Uruguay, Latin American and Global EMBI spreads, 1996-2006 1/
1/ For Uruguay, the Uruguayan Bond Index (UBI) is reported; for Latin American and Global Emeging Markets the JP Morgan EMBI spread is reported.
8. In 2002–03, Uruguay's sovereign spreads spiked in the mist of the financial crisis rooted in the withdrawal of Argentine deposits from the Uruguayan banking system. A severe contraction of the economy, a depreciation of the peso, and a marked increase in public debt led to debt sustainability concerns and to sharp increases in sovereign spreads to above 2000 bps. With the voluntary debt restructuring—which did not imply any hair cut, but a small NPV reduction—some measure of market confidence was rapidly restored. Still, sovereign spreads remained significantly above pre-crisis levels, reflecting continued concerns about Uruguay's repayment capacity and the loss of investment grade rating.
9. In contrast with the 1990s, since the crisis Uruguayan spreads have remained close to, and have displayed high correlation with, neighboring countries' spreads. Since 2002 Uruguayan spreads have remained at about the level of, and closely co-moved with, the Latin America EMBI. Sovereign spread correlations across countries have generally increased, but Uruguay has been particularly affected (Figure 4). While before the crisis daily correlation with neighboring countries was low, Uruguayan cross border correlations increased after the crisis to the levels displayed by other countries in the region (Table 1).3 Notably, correlation with the Latin American EMBI and the Global EMBI have tripled. Interestingly, while the correlations between Uruguayan and other non-investment grade Latin American countries' spreads have increased, the correlation with Chile has declined.
Figure 4.Selected country spreads and Latin American EMBI Spread, pre and post Uruguayan crisis. 1/
1/ Argentina is excluded, as most of Uruguay's post-crisis period coincides with the Argentine debt restructuring process.
|Pre-crisis (Jan 1999-April 2002)|
|Post Crisis (June 2003-Dec 2006)|
10. Furthermore, there is evidence of heightened financial spillover risks following the crisis. Before the crisis, Uruguayan spreads were relatively insulated from shocks in other countries in the region, as confirmed by Granger causality tests between Uruguayan and Latin American spreads based on daily data (Table 2). With exception of Argentina—with which Uruguay had strong trade and financial links—changes in neighboring countries' spreads did not spill over to Uruguay.4 Most notably, the Brazilian and Latin American EMBI's had no significant effect on Uruguayan spreads.5After the crisis, however, financial spillovers from the region and other EMEs seem to have increased. All pairwise tests show significant causality, pointing to higher vulnerability of Uruguay to regional and global shocks to the emerging market asset class.6
|URUGUAY does not Granger Cause BRAZIL||828||1.12||0.327||896||0.90||0.407|
|BRAZIL does not Granger Cause URUGUAY||1.61||0.200||39.33||0.000 ***|
|URUGUAY does not Granger Cause CHILE||726||2.56||0.078 *||896||0.40||0.671|
|CHILE does not Granger Cause URUGUAY||1.51||0.221||7.32||0.001 ***|
|URUGUAY does not Granger Cause COLOMBIA||828||0.47||0.624||896||0.15||0.864|
|COLOMBIA does not Granger Cause URUGUAY||0.28||0.757||25.23||0.000 ***|
|URUGUAY does not Granger Cause EMBI||828||0.96||0.384||896||0.06||0.944|
|EMBI does not Granger Cause URUGUAY||3.33||0.036 *||28.23||0.000 ***|
|URUGUAY does not Granger Cause LATIN_AMERICA||828||0.93||0.396||896||0.39||0.680|
|LATIN_AMERICA does not Granger Cause URUGUAY||1.83||0.161||35.37||0.000 ***|
|URUGUAY does not Granger Cause PERU||828||0.93||0.396||896||1.40||0.247|
|PERU does not Granger Cause URUGUAY||1.21||0.300||19.45||0.000 ***|
|URUGUAY does not Granger Cause MEXICO||828||1.15||0.316||896||0.40||0.672|
|MEXICO does not Granger Cause URUGUAY||0.29||0.747||18.95||0.000 ***|
|URUGUAY does not Granger Cause ARGENTINA||828||0.05||0.947||896||0.13||0.880|
|ARGENTINA does not Granger Cause URUGUAY||6.11||0.002 ***||1.24||0.291|
|URUGUAY does not Granger Cause ARGENTINA 2/||686||1.70||0.183||375||0.40||0.670|
|ARGENTINA does not Granger Cause URUGUAY 2/||4.83||0.008 **||8.62||0.000 ***|
11. The heightened financial spillovers seem to be associated with the loss of investment grade. A simple comparison of sovereign spreads among Latin American countries reveals that during the late 1990s—when Uruguay had investment grade—spreads were significantly lower than those of most neighboring countries and close to the ones of Chile—the other investment grade economy in the region. Furthermore, Uruguayan spreads co-moved with those of Chile during that period. Since the 2002–03 crisis (and the associated loss of investment grade), however, Uruguayan spreads have moved close to those of other non-investment grade economies—with which high correlations are observed (Figures 5 and 6).7 Since then, Uruguay's credit rating has been recovering, but it is still significantly below investment grade.8
Figure 5.S&P Credit rating for selected Latin American countries, 1994-2007 Figure 6.Uruguay and Selected Latin American country spreads, pre and post crisis.
C. What Drives Sovereign Spreads?
12. A Vector Error Correction model (VECM) is estimated to quantify the contribution of external and domestic factors on Uruguay's sovereign spreads. Following the methodology previously applied by Arora & Cerisola (2000) and Larzabal, Valdez and Laporta (2001) among others, the paper studies the determinants of individual country spreads. A VEC specification provides an adequate framework to disentangle short-term from long-term effects, while allowing for country-specific structural breaks.
13. The data set comprises monthly information for 1996–2006. Following the literature on determinants of sovereign spreads,9 we choose public debt, the fiscal balance, and external debt (all as share of GDP), reserve coverage (as share of short-term debt and foreign currency deposits), and the real effective exchange rate to account for country fundamentals. These fundamentals reflect the economy's repayment capacity and its vulnerability to external shocks. To capture external factors, we use U.S. interest rates, terms of trade, the high yield spread index, a market volatility measure (VIX) and EMBI spreads. As a measure of sovereign risk, we use the Uruguayan Bond Index, instead of the EMBI due to its longer time span.10 Standard unit root tests (Augmented Dickey-Fuller) confirm that all variables are nonstationary, and support the notion of first-order integration (Table 4).
|In Level||First Difference|
14. A co-integrating relation is found among the Uruguayan spreads, public debt, reserve coverage, TOT, the VIX, the HY and the Latin American EMBI (Table 5).11 From all variables considered that represent country fundamentals, only the public debt-to-GDP ratio and the reserve coverage are significant in the long-run relationship.12 They both display the expected signs. A 1 percentage point increase in the public debt-to-GDP ratio increases the spread by about 17 bps, while a 1 percentage point increase in reserve coverage reduces the spread by about 28 bps (Equation 1). Global factors, such as the ToT and the VIX enter in the long-run relationship with the Uruguayan spread, and with the right sign.13 It is interesting to note that, in this specification without structural break, the effect of the Latin American EMBI spread on the Uruguayan spread is not significantly different from zero. As it is shown next, this reflects the presence of a structural break.
|Variance Decomposition 1/|
|Period||UBI PDGDP||RES||TOT||VIX||HY||EMBI_LA||EMBI_LA *(1-InvGrade)|
15. If a structural break is allowed, results confirm that the influence of other EMEs on Uruguay has changed following the loss of investment grade. Although the Latin American EMBI is not statistically significant by itself when included in the initial specification, evidence of financial spillover is found if a structural break associated with Uruguay's downgrading to speculative grade is allowed (Equation 2). Moreover, estimates show that before the crisis (and after controlling for other global financial factors) the effect of regional shocks on Uruguayan spreads was negative, suggesting that Uruguay was perceived as a safe heaven within the region.14 After the loss of investment grade, however, the effect of regional shocks on the Uruguayan spread is positive. Similar, and even stronger, results are found if the Global EMBI is used (Equation 3). While this may suggest that spillovers may stem from the emerging market asset class as a whole, it should be noted that Latin American countries (mainly Brazil and Mexico, and previously Argentina) have a significant weight in the global index.
16. While fundamentals explain part of the variance of spreads, global factors have played an important role. Variance decomposition analysis identifies some 20 percent of the variance explained by country fundamentals and about 60 percent by external factors.15 Among country fundamentals, public debt is most important in explaining Uruguay's spreads, and among external factors the VIX and the EMBI LA have the largest contributions (Table 5).
D. Conclusions and Policy Implications
17. While country fundamentals have explained part of the variation in sovereign spreads, external factors have played an important role. External factors have become more important since the crisis, particularly after Uruguay was downgraded to speculative grade in 2002. In the late 1990s, Uruguay was largely insulated from regional and EME financial shocks, with sovereign spreads at the level of Chile's and significantly below those of non-investment grade neighbors. Since the crisis, however, Uruguayan spreads have been decoupled from those of Chile, standing at the level of—and co-moving with those of non-investment grade neighboring countries. Furthermore, after controlling for country fundamentals and global factors, econometric estimates point to heightened financial spillovers from other EMEs, following the downgrading of Uruguay's credit rating. This suggests that, following the crisis and the loss of investment grade, there has been a change in investors' perception of Uruguay's vulnerability to shocks in other EMEs.
18. These results suggest that, while Uruguay has made remarkable strides over the last years, it is still perceived as more vulnerable to global and regional shocks than before the crisis. While financial and real (trade) linkages with neighboring countries have declined in recent years, Uruguay is now more vulnerable to what happens in global markets, reflecting higher perceived riskiness. In this context, continuing to strengthen the macroeconomic framework, diversifying trade destinations, further improving the public debt structure, and implementing pending structural reforms are key towards regaining investment grade status and, thus, insulating Uruguay from a possible turnaround in global conditions and from regional shocks.
Arora, V., andM.Cerisola,2000, “How Does U.S. Monetary Policy Influence Economic Conditions in Emerging Markets,” IMF Working Paper No. 00/48, (Washington: International Monetary Fund).
Eble, S.,2006, “Uruguay's Growth History”, IMF Staff Country Report No. 06/427(Washington: International Monetary Fund).
Forbes, K., andR.Rigobon,1999, “No Contagion, only Interdependence: Measuring Stock Market Co-Movements,” NBER Working Paper 7267(Cambridge, Massachusetts: National Bureau of Economic Research).
Garcia Herrero, A., andA.Ortiz,2005, “The Role of Global Risk Aversion in Explaining Latin American Sovereign Spreads,” Documento de Trabajo No. 0505(Madrid: Banco de España).
Gelos, R. G. andR.Sahay,2001, “Financial Market Spillovers in Transition Economies,” Economics of Transition, Vol. 9(January)2001, pp. 53–86.
González Rozada, M., andE.Levy Yeyati,2005, “Global Factors and Emerging Market Spreads” Documento de Trabajo 07/2005(Buenos Aires: Universidad Torcuato Di Tella).
JaqueS., andA. RojasO.,2003, “Determinants of the Chilean sovereign Spread: Is it Purely Fundamentals?” Money Affairs, Vol. XVI, N.2., July–December2003, CEMLA.
Kamin, S., andKarstenvon Kleist,1999, “The Evolution and Determinants of Emerging Market Credit Spreads in the 1990s,” Federal Reserve Board of Governors, International Finance Discussion Papers No.653.
Kanagasabapathy, K., andRajanGoyal,2002, “Yield Spread as a Leading Indicator of Real Economic Activity: an Empirical Exercise on the Indian Economy,” IMF Working Paper 02/91(Washington: International Monetary Fund).
Larzabal, M.,M.Valdes, andS.Laporta,2001, “Spread Soberano: Evidencia Empírica del Caso Uruguayo,” Departamento de Estudios de República AFAP S.A.
Mauro, P.,S.Nathan, andY.Yishay,2000, “Emerging Market Spreads: Then Versus Now,” IMF Working Paper 00/190(Washington: International Monetary Fund).
In the remainder of the paper, credit ratings refer to foreign currency long-term debt.
Financial market volatility is proxied by the Chicago Board Options Exchange Volatility Index, a popular measure of the implied volatility of S&P 500 index options, which aggregates market expectations of volatility over the next 30-day period. To measure the average spread of U.S. speculative grade corporate bonds (a proxy for investor risk appetite) the Merry Lynch high yield index is used.
Following Forbes and Rigobon (1999) and Gelos and Sahay (2001), a formal test of increase in correlation, applying the correction for changes in variance, was conducted, confirming the findings. As the variance has fallen in the post crisis period, the adjustment accentuates the increase in correlations.
See Eble (2006).
It is also interesting to note the impact of Uruguayan shocks on Chilean spreads (at 10 percent significance level) in the pre-crisis period, and the following reversion in the direction of causality after the crisis.
Since the table displays pairwise tests, results should be interpreted with caution. Spillover from small countries in the region is likely to reflect aggregate shocks to either all emerging markets or to the region.
Increased correlation and financial spillovers from non-investment grade countries may reflect a change in the set of investors for Uruguayan debt instruments, due to the fact that some institutional investors (often with buy-and-hold strategies) are not allowed to hold instruments with speculative ratings.
In June 2007, S&P introduced a new methodology for rating sovereign debt issuers according to the expected recovery rate in the event of a default. Interestingly, Uruguay's recovery rating (2='substantial recovery') is higher than all graded countries in the region. A higher recovery rating combined with a lower overall rating—vis-à-vis neighboring countries—suggests that, despite a good reputation for debt repayment, Uruguay is perceived to be vulnerable to external shocks.
For most recent work in this topic see Garcia Herrero and Ortiz (2005), Gonzalez Rozada and Levy Yeyati (2005), Jaque et al (2005), Makin and Karsten von Kleist (1999), Mauro et. al (2000), and Min (1998).
The UBI is built from spreads of fixed-rate, dollar-denominated Euronotes and Global Bonds issued by the Uruguayan government. UBI spreads are highly correlated with the Uruguayan EMBI spread produced by JP Morgan, since the introduction of the later in 1998.
Both Trace and Maximum Eigenvalue tests confirm the existence of a unique co-integrating relation.
External debt is excluded as it is highly collinear with the public debt-to-GDP ratio.
Unlike previous work that has stressed the effect of US interest rates on EMC spreads, this link is not found for Uruguay.
These results are consistent with previous work by Larzabal, Valdes and Laporta (2001) covering the pre-crisis period who find that changes in the EMBI had a negative impact on Uruguayan spreads.
Based on Equation 2.