تاثیر پیش فرض آرژانتین بر روی نوسانات اقدامات مشترک در بازار اوراق قرضه نوظهور

This paper investigates the impact of the Argentine default of December 2001 on daily spreads on sovereign bonds issued by 10 emerging countries located in Latin America and Asia. It addresses the problem of shift contagion examining the dynamics of variances and covariances obtained with conditional procedures. The intra-regional analysis detects signs of contagion in Latin America, where the default has long-lasting effects. In Asia, changes in spread covariation due to Argentine and other extra area shocks tend to be less persistent. The conditional covariances between Asia and Latin America reflect this state of affairs: they are positive and large only during the turbulent months following September 11 2001 and corroborate the hypothesis of a temporary contagious reaction of the Asiatic spreads to the Argentine default. The blurred vision of the idiosyncratic nature of emerging market bonds is short-lived, and so is the ensuing herding behaviour. A policy of diversification with assets from different geographical areas may thus reduce portfolio risk, in terms of overall variance, in spite of cross-area contagion.

The volume of bonds issued by developing countries had risen substantially from 1990 to 1995, Brady bonds being progressively superseded by direct sovereign and private bond issues actively traded in secondary markets.1 Recent financial crises in emerging countries, ranging from Mexico to Russia and Argentina, altered this promising scenario. Large capital outflows were associated with a rising degree of interest rate volatility and higher cost of borrowing. Indeed, an overall increase in volatility also tended to involve assets issued in markets that had little in common with the market originally affected. This paper investigates the impact of the Argentine default of December 2001 on daily spreads on sovereign debt issuance from 10 emerging countries located in Latin America and in Asia. Spreads are measured as the interest rate differential between the annualized yield on a dollar denominated emerging country sovereign bond and the annualized yield of a U.S. treasury bond of the same maturity. We use daily data from 3 January 2000 to 15 April 2002. The analysis focuses on spread volatility co-movements which may be affected by contagion phenomena triggered by a shock, here identified as the Argentine default crisis. Financial contagion is a notoriously elusive concept and requires close preliminary definition. It may simply be seen as “shift contagion”, i.e. an increase in asset price correlation linkages in the aftermath of a financial crisis, symptomatic of the crisis-contingent nature of the volatility transmission mechanisms (Forbes and Rigobon, 2000). An element of surprise is often introduced, due either to the geographical location of the markets involved, far from the epicentre of the turmoil, or to the dimension of the shifts. The latter exceed what is to be expected according to a financial and/or economic rationale (Mody and Taylor, 2003 and Bekaert et al., 2004). Indeed, the literature on contagion suggests that price co-movement explanations based on fundamentals and economic and financial linkages miss out some relevant explanatory factors, such as multiple equilibria (Masson, 1998), endogenous liquidity shocks (Valdés, 1997 and Kaminsky and Schmukler, 1999) and asymmetric dissemination of information (Calvo and Mendoza, 1995 and Calvo and Mendoza, 2000). A related issue, also investigated in this paper and set out in Mauro et al. (2002), among others, is whether spreads are influenced by global events and tend accordingly to co-move or, rather, are influenced by idiosyncratic shocks only. Our analysis follows Forbes and Rigobon insofar as it associates contagion with an increase in spread volatility linkages among markets brought about by a financial crisis. It differs from their approach, however, in two significant aspects: (i) it discards correlation coefficients and focuses on the behaviour of (conditional) variances and covariances which, according to a stylized regularity identified by Corsetti et al. (2001), tend to rise simultaneously during crises; (ii) it relies on an assessment of cross-area price volatility segmentation. From our analysis it turns out that the tight segmentation between the Latin American and Asiatic financial areas is weakened, only temporarily, by the Argentine crisis and provides useful information on contagion dynamics. Contrary to common wisdom that emerging market bonds tend to be a single asset class, we find evidence of strong segmentation between geographical areas; only under exceptional circumstances do shocks spill over across regions. This analysis improves upon previous work in the following aspects. – International comparisons are facilitated by the use of homogeneous secondary yield spreads on bonds denominated in U.S. dollars. We thus avoid the biases associated with the use of Brady bonds and primary market yields, discussed in the literature. – A flexible two-step estimation procedure discards idiosyncratic impulses and focuses on spread volatility co-movements that are due to common factors. – Estimation of large conditional variance–covariance matrices of spread differences provides detailed information on their daily volatility dynamics. At the same time, it avoids some of the shortcomings affecting the shift contagion analysis of Forbes and Rigobon, such as the arbitrary choice of the dating of the crisis, the exogeneity restriction on the contagious idiosyncratic shock and the low power of parameter stability testing with subsamples of uneven length. The paper is structured thus: Section 2 outlines the Orthogonal-GARCH estimation procedure; Section 3 describes our data and reports some preliminary findings on contagion; Section 4 analyses spread difference conditional covariation across the Argentine default and discusses within and inter area volatility segmentation; concluding remarks are offered in Section 5. 2. The two-step Orthogonal GARCH procedure Correlation analysis does not produce an exhaustive representation of the complex interconnections between spreads. Accurate analysis of the latter is, however, of great relevance for portfolio investment decisions. In this section we use the Orthogonal GARCH procedure set out in Alexander and Leigh (1997) and Alexander (2001) to deduce the behaviour of the conditional second moments of a large set of spread differences from the second moments of a small number of principal components.

This paper presents an analysis of how the second moments of emerging market sovereign bond spreads co-move over time, possibly because of financial contagion. In line with recent empirical evidence we find that volatility linkages are mostly of a regional nature. Only in exceptional circumstances, e.g. September 11 or the Argentine default, do we detect significant cross-area interlinkages. Correlation stability tests based on the approach by Forbes and Rigobon find, over the same data set, no evidence of shift contagion in the aftermath of the Argentine crisis. This result is probably due to the violation of the heteroskedasticity identifying restriction of the contagious Argentine shock. The huge global impulse of September 11 affects most emerging market spread volatilities and introduces an omitted variable bias while the large shifts of the Brazilian and Venezuelan spreads point to an additional endogeneity bias in the corresponding correlations with the Argentine spread. Our empirical investigation addresses the issue of contagion examining the dynamics of conditional variances and covariances obtained with the Orthogonal GARCH procedure. The intra-regional analysis detects signs of contagion in Latin America, where—in most cases—the increase in volatility brought about by the default has long-lasting effects. In Asia, shifts in spread covariation brought about by Argentine and other extra area shocks tend to be less persistent. The conditional covariances, at inter-regional level, reflect this state of affairs. They are positive and large only during the turbulent months which follow September 11 2001 and corroborate the hypothesis of a temporary contagious reaction of the Asiatic spreads to the Argentine default. The evidence of segmentation across geographical areas can reflect the presence of legal and regulatory barriers on capital flows and the impact of a different ability in processing and/or acquiring information by foreign investors. The relevance of these factors on the emerging equity markets has already been documented in Bae et al. (2004) and Bailey et al. (2004). A specific analysis of the effects of institutional barriers on emerging bond markets provides an interesting topic for future research.