قیمت گذاری روزانه بازار نوظهور CDS قبل و در طول بحران مالی جهانی

In this paper, we study the determinants of daily spreads for emerging market sovereign credit default swaps (CDSs) over the period April 2002–December 2011. Using GARCH models, we find, first, that daily CDS spreads for emerging market sovereigns are more related to global and regional risk premia than to country-specific risk factors. This result is particularly evident during the second subsample (August 2007–December 2011), where neither macroeconomic variables nor country ratings significantly explain CDS spread changes. Second, measures of US bond, equity, and CDX High Yield returns, as well as emerging market credit returns, are the most dominant drivers of CDS spread changes. Finally, our analysis suggests that CDS spreads are more strongly influenced by international spillover effects during periods of market stress than during normal times.

Walter Wriston, then Chairman of Citibank, once famously remarked that “countries don’t go bust” (Guill, 2009). Yet, by October 1983, and only a few months after that statement was made, 27 countries owing about $240 billion in debt had rescheduled these obligations or were in the process of doing so—in what is now commonly known as the “LDC debt crisis.” Indeed, countries do go bust, in the sense of refusing or being unable to meet their financial obligations. Recent developments surrounding the restructuring of Greek government debt are a stark reminder of this fact, even though such events are relatively rare. Historical data reveal a number of important patterns: sovereign defaults often occur in waves and tend to be heavily concentrated in periods of extreme stress, with the largest wave of defaults occurring during the Great Depression and World War II. Historically, the majority of defaults involve countries’ external debt; defaults involving domestic debt are less common. Both emerging market and industrialized country issuers experience default, but the former are more prone to it than the latter. Specifically, based on the frequency with which a country has moved into default, emerging market borrowers are about 10 times more likely to default than are their peers from developed markets. 1 Thus, country risk is an important factor in the pricing of sovereign debt, especially—though not only—for emerging market borrowers. Assuming rational investors, we would expect that credit spreads on sovereign debt instruments reflect such risks. Surprisingly, the evidence on the importance of country-specific risks in the pricing of sovereign debt is rather mixed. Longstaff et al. (2011), for example, show—using monthly data—that returns on sovereign credit default swaps (CDSs), a common measure of credit risk, are substantially more correlated across countries than are corresponding stock index returns. They find that these spreads are more related to US stock and high-yield credit markets, proxies of global risk premia, and international liquidity patterns, than they are to local financial measures. Thus, the country-specific risk premium—after adjusting for global and, in particular, US risk factors—appears to be almost negligible. This suggests a potentially important role for international and, in particular, US financial variables in the determination of non-US sovereign CDS spreads—and thus a channel for spillovers into those countries’ funding costs in international debt markets. Our approach is related to literature on the determinants of corporate credit spreads and the pricing of individual firms’ CDS, including work by Collin-Dufresne et al., 2001, Campbell and Taksler, 2003 and Ericsson et al., 2009. Longstaff and Rajan, 2008, Bhansali et al., 2008 and Fender and Scheicher, 2009 apply similar methodologies to multi-name CDS contracts (i.e. contracts based on portfolios of underlying credits). A standard finding of these studies is that broad factors, such as measures of risk appetite and market liquidity, play an important role in the determination of observed CDS spreads. Fontana and Scheicher (2010) study the relative pricing of euro area sovereign CDS and underlying government bonds and find that repricing of sovereign credit risk in the CDS market seems mainly due to common factors. We explore these spillovers by building on extant literature on the impact of US financial variables on foreign asset returns. Studies of policy spillovers typically focus on equity and bond markets; however, our analysis employs daily CDS spread data for 12 emerging market borrowers.2 Sovereign CDS are traded in relatively liquid markets and provide a direct indicator of the credit risk premium demanded by investors. As such, CDS premia are close proxies for the excess funding costs of sovereign borrowers relative to benchmark US Treasury yields. They also often serve as a lower-bound measure for the wholesale funding costs of banks and corporate issuers from the same countries. Although we take Longstaff et al.’s (2011) analysis as a starting point, our methodology differs from theirs in a number of ways. First, we employ daily instead of end-month data to measure spillover effects on sovereign CDS returns. Low-frequency data tend to exhibit higher correlations and, hence, an empirical analysis using monthly financial market data could potentially overestimate the importance of spillover effects. In contrast, daily data measure the direct impact of US financial markets on CDS spreads, thereby also incorporating the high degree of volatility typical for financial market data (see Fig. A.2 in Appendix A). Second, we quantify the economic relevance of such spillover effects, as we provide point estimates and their significances instead of presenting only t-statistics. Third, in addition to financial factors, we also incorporate several macroeconomic variables, such as economic growth, debt/GDP levels, fiscal deficits, net foreign assets, and country ratings. As a consequence, we can directly and systematically compare the effects of international financial variables with domestic financial variables and macroeconomic factors. Furthermore, we also control for target rate movements by the European Central Bank (ECB) and the Federal Reserve (Fed). Fourth, we split our sample to examine potential differences in the reaction of CDS spreads to various factors before and during the recent financial crisis. 3 Finally, we use a panel framework to estimate the influence of the factors more efficiently and to illustrate the most relevant drivers of emerging market CDS spreads at a glance in one model. We investigate the determinants of daily spreads for emerging market sovereign CDS spreads by addressing three closely related research questions. First, are there common factors that cause daily sovereign CDS spread changes across emerging financial markets? Second, what is the impact of (1) domestic financial and macroeconomic variables and country ratings compared to (2) US and international financial variables on daily sovereign CDS spread changes? Third, are there noticeable differences in the reaction of CDS spreads before and during the recent financial crisis? Our findings suggest that (1) common factors play a role for daily sovereign CDS spread changes across emerging financial markets, (2) daily CDS spreads for emerging market sovereigns are more related to global and regional risk premia than to country-specific risk factors. This result is particularly evident during the second subsample (August 2007–December 2011), where neither macroeconomics variables nor country ratings significantly explain CDS changes. Finally, (3) the amplified reaction to international financial variables in the second subsample suggests that CDS spreads are more strongly influenced by spillover effects during the financial crisis. The remainder of this paper is organized as follows. Section 2 presents a brief introduction to the mechanics of sovereign CDS. Section 3 describes the CDS data and our sample selection process and introduces the explanatory variables used in the empirical analysis. Section 4 presents the econometric methodology, illustrates the results, and reports a number of robustness tests. Section 5 concludes. 2. Sovereign credit default swaps (CDSs) Sovereign CDS are financial contracts offering insurance against losses from credit events on outstanding debt issued by sovereign entities. Standard contracts have two legs. The protection buyer pays a premium (the premium leg), expressed in basis points per notional amount of the contract, in exchange for a contingent payment (the contingent leg) if any of the contractually pre-specified credit events occur. Settlement on these contracts is typically by physical delivery of admissible bonds in return for payment of the original face value.4 As such, CDS for both sovereign and corporate reference entities have five distinct contractual features: (1) the debt issuer (reference entity), (2) a set of reference obligations, (3) the contract term (e.g., 5 years), (4) a notional principal amount, and (5) a list of events triggering protection payments (Markit, 2008b). The Standard International Swaps and Derivatives Association (ISDA) defines six different credit events, some or all of which may be selected for individual CDS contracts: (1) bankruptcy of the reference entity, (2) failure to pay (the reference entity fails to make interest or principal payments when due; a grace period and materiality threshold may apply), (3) debt restructuring (the configuration of debt obligations is changed in a way unfavorable to the creditor; e.g. maturity extension, coupon or par amount reduction, postponement in coupon dates, or change in currency), (4) obligation default, (5) obligation acceleration, and (6) repudiation/moratorium. The range of restructuring events included in the CDS contract depends on the selected restructuring clause. In our sample, the most common clause in sovereign CDS is the so-called complete (or cum-) restructuring (CR) clause, which allows for any form of restructuring and delivery of any bond of maturity up to 30 years.5 This stands in contrast to CDS for corporate issuers, which tend to limit the range of qualifying events, as well as the allowable maturity of deliverable obligations.6 Pricing of such contracts results in a CDS premium (spread) equating the present value of both payment legs over the (expected) lifetime of the deal. Holding the annual probability of default (conditional on earlier non-default) constant over time, pricing can be interpreted in terms of a constant hazard rate (Duffie and Singleton, 2003). Making further assumptions, such as the absence of counterparty default risk and continuous premium payments, the CDS spread at origination (i.e. with market value of zero) can be shown to equal (1 − ρ)λ, where ρ is the recovery rate and (1 − ρ) denotes loss given default (LGD). The hazard rate λ corresponds to a risk-neutral loss probability that reflects the risk preferences of investors. Using actual probabilities of default (PD) instead, this yields (1 − ρ)PD + RP for the annual CDS spread, where a risk premium (RP) accounts for the difference between λ and PD, which is typically positive. In other words, observed CDS spreads tend to represent a combination of expected loss (EL = (1 − ρ)PD) and an extra premium to compensate investors for risks in addition to EL. Note that full-scale default is only one type of event that sovereign CDS insure against. In particular, relevant standard documentation tends to be based only on restructuring, repudiation/moratorium, and failure to pay (allowing for pre-defined grace periods). Thus, the bankruptcy credit event usually is not covered in sovereign CDS. Instead, the contingent CDS payment might be triggered, for example, when the legal code governing the issuance of sovereign debt is amended7 or when interest or principal payments (subject to minimum threshold amounts) on individual obligations are made with (even relatively short) delays, which would give the protection buyer the right to deliver (subject to any deliverability requirements) any discounted bonds at face value—an event that would tend to generate relatively high recovery rates. As a result, and abstracting from other factors, such as liquidity premia, observed increases in sovereign CDS spreads may reflect rising probabilities of “technical default”—along with the transition risk of a sovereign rating downgrade—as much as genuine concerns about principal losses on outstanding debt. One advantage of working with sovereign CDS data is that these contracts are among the most actively traded instruments in global credit markets. Depository Trust and Clearing Corporation (DTCC) data for end-December 2009 indicate that, among the top-100 reference names (by US-dollar-equivalent gross notional amounts outstanding) in the CDS market, 19 were sovereign entities (of which 11 were emerging market sovereigns), including all the top-6 names. In volume terms, at $1.46 trillion, these sovereigns accounted for almost 30% of the aggregate notional amount of the top-100 reference names taken together. By the first quarter of 2012, this ratio had risen to 40%, with gross notional amounts outstanding for the 25 sovereign issuers among the top-100 at $2.24 trillion. Detailed volume data for the full period covered in this study (see below) are unavailable, but other studies suggest that the overall market for sovereign CDS was active enough even during these earlier years to support our analysis (see Longstaff et al., 2011).

Credit default swaps, especially those involving sovereign issuers, are a hotly debated topic in both public and private circles, not least because of recent events involving Greece. Yet, even though these instruments have been around since the early 1990s, academic studies of CDS markets and of price determination in those markets are rare—at least compared to the level of interest in those instruments by those outside the academic community. To fill some of the gaps in the current literature, this paper investigates the determinants of daily spreads for emerging market sovereign CDS, focusing in particular on the effects of financial market spillovers for CDS pricing. Specifically, we use GARCH models estimated over the periods April 2002–July 2007 and August 2007–December 2011 to establish the impact of a broad range of potential influences, particularly that of international financial variables. Our study provides answers to three closely related questions regarding the importance of spillovers from broader financial markets for the determination of sovereign CDS spreads. First, our results suggest that daily CDS spreads for emerging market sovereigns are more related to global and regional risk premia than to country-specific risk factors, although idiosyncratic risk can play an important role for individual emerging market borrowers. This is in line with earlier findings that diversification opportunities from spreading fixed income investments across countries and regions might be more limited than expected (and more so for fixed income instruments than for equity market investments; see Longstaff et al., 2011). In particular, measures of US bond, equity, and CDX High Yield returns, as well as emerging market credit returns, are the most dominant drivers of CDS spreads. Second, we detect noteworthy differences in the determination of CDS spreads over time. During the early years of our sample, emerging market CDS spreads react to country-specific factors. This pattern changes dramatically during the financial crisis, as country-specific factors no longer play a statistically significant role. Finally, during the financial crisis, external (international) factors are more important for the determination of CDS spreads than are internal (domestic) factors, both in terms of statistical significance and economic importance. The impact of these variables increases by a factor of between 2.5 and 10 in the second subsample, thus supporting the conclusion that spillover effects play a much larger role during crisis periods than during more tranquil times. Overall, these findings support the possibility of significant international spillovers from developments in international and, particularly, US financial markets to emerging market sovereign borrowers. More specifically, as CDS premia are close proxies for the excess funding costs of sovereign borrowers relative to benchmark US Treasury yields, movements in US and other major financial markets can exert a meaningful influence on the funding costs of sovereign issuers, banks, and corporate borrowers from emerging market economies. By extension, policy measures taken in the major advanced economies can feed directly into the funding costs of their emerging market peers, at least temporarily. Finally, our finding that monetary policy decisions and their communication (see the robustness tests described in Section 4) can influence sovereign borrowers raises important issues in the context of current discussions about exit strategies from historically low, crisis-induced monetary policy rates. Put differently, monetary-policy-induced changes to these spreads, as a result of both policy actions and their communication, can have a potentially significant impact on the borrowing costs in emerging market economies. Country authorities may want to take these effects into account when devising their own policies.