سرریز نقدینگی در اوراق قرضه و بازار مبادله عدم پرداخت بدهی های مستقل اعتباری: تجزیه و تحلیل بحران بدهی منطقه یورو

At the end of 2009, countries in the Eurozone (euro area) began to experience a sudden divergence of bond yields as the market perception of sovereign default risk increased. The theory of complete markets suggests that sovereign debt and credit default swap (CDS) credit spreads should track each other closely. In addition, liquidity risk should be priced into both instruments in such a way that buying exposure to the same default risk is identically priced. We use a time-varying vector autoregression framework to establish the credit and liquidity spread interactions over the 2009–2010 crisis period. We find substantial variation in the patterns of the transmission effect between maturities and across countries. Our major result is that, for several countries, including Greece, Ireland and Portugal the liquidity of the sovereign CDS market has a substantial time varying influence on sovereign bond credit spreads. This evidence is of particular importance in the current policy context.

This paper develops an empirical model of the co-evolution of credit and liquidity spreads in sovereign bond and CDS markets. We apply this model to study the liquidity and credit interactions for Eurozone countries over the January 1, 2007 to October 1, 2010 period. During November 2009, shortly after the election of a new Greek government, the Greek public sector deficit was revised from 6% to 12.7% of GDP. This event initiated a sovereign debt crisis that has resulted in large financial interventions in Greece, Ireland and Portugal. The resultant fluctuations in financial markets and sudden impact on fiscal policy in the affected states is often referred to as the ‘Eurozone sovereign debt crisis’. The deterioration in government finances, after the credit crisis of 2008, led to a sudden loss of confidence in sovereign debt markets, which revealed itself in the form of widening credit spreads between several euro area members and the Eurozone's largest economy, Germany. The response by Eurozone member states and international bodies such as the International Monetary Fund (IMF) and the European Union (EU) has been a suite of interventions led by the European Central Bank (ECB).1 Throughout 2010 the focus of the crisis has been on Greece, Ireland and Portugal where there is concern about the rising cost of financing government debt and the possibility of default on a sovereign debt issuance. On May 2, 2010, the Eurozone countries and the IMF agreed to a EUR 110 billion loan for Greece, conditional on the implementation of a package of severe austerity measures. On May 9, 2010, Europe's Finance Ministers approved the creation of the European Financial Stability Facility (EFSF) aimed at preserving financial stability in Europe by providing financial assistance to Eurozone states in economic difficulty. The objective of the EFSF is to collect funds and provide loans in conjunction with the IMF to cover the financing needs of Eurozone Member States in difficulty, subject to strict policy conditionality. The Greek bailout was followed by a EUR 85 billion rescue package for Ireland in November 2010, and a EUR 78 billion bailout for Portugal in May 2011. During the crisis, several commentators expressed concern that manipulation of the CDS market by speculative investors was playing a crucial role in exacerbating the liquidity dry up in the market for Greek, Irish, Portuguese and Spanish sovereign debt. In particular,‘naked’ CDS positions were blamed for driving bond yields on these countries’ sovereign debt higher during the first half of 2010. In this context, Greece was suggested to have been a victim of short-term speculative short selling practices on its national debt and naked shorting practices in the CDS market. In this paper, we investigate the potential spillover effects between the credit and liquidity spreads in the Eurozone sovereign bond market and the sovereign credit default swap (CDS) market during the 2010 European sovereign debt crisis. We define two credit spreads, denoted by the suffix CS. First, the difference in the required discount rate on Eurozone members’ benchmark sovereign debt issuance against a benchmark (in this case equivalent German benchmark bonds). Second, the difference between the CDS spread on benchmark sovereign debt against the CDS spread on equivalent German sovereign debt. In equilibrium, these should co-move almost precisely. We then use the bid–ask spread on the bond and CDS markets to proxy for market depth and transactions costs and hence market liquidity and enter these four variables into a vector model. This is denoted by the suffix LS for liquidity spread. For brevity we utilise the commonly understood equity/foreign exchange parlance bid–ask spread, to refer to the difference between the yield (spread) bid and the yield (spread) ask for the sovereign bond (CDS) market. The remainder of the paper is organised as follows. Section 2 provides background information on the Eurozone crisis and reviews the related literature on the pricing and liquidity of the bond and CDS market. Section 3 describes our empirical methodology and Section 4 the data pre-processing used in the study. Section 5 contains our analysis of the results whilst Section 6 offers our concluding remarks.

We have outlined a an empirical model of cross liquidity and price discovery in parallel markets using a time varying vector autoregression. We have used this model to elucidate the impact of cross liquidity effects on the Eurozone sovereign debt and CDS spreads during the Eurozone sovereign debt crisis. We document three noteworthy findings. First, we show that explosive trends did appear during the sovereign crisis and that the CDS market does appear to have been a driver in most cases. Second, we find a positive and significant lagged transmission from the liquidity spread of the CDS market (proxied by the bid–ask spread) to the credit spread in the bond market. Finally, we observe several variance breaks in the time varying models indicating that the noise structure inundating the VAR model has changed markedly over the 2007–2010 period. These findings indicate that: (a) static models would be uninformative and indeed biased when estimated over such a dataset and (b) that the information structure of the sovereign credit market has undergone substantial and dramatic adjustments throughout the crisis period. The implications of our results for policy makers and practitioners are significant. For instance, it is clear that, in the absence of a coordinated EU action, the explosive trend would have resulted in a complete market failure for the trade in sovereign debt instruments for several Eurozone countries. The time evolution of the coefficients points to a multitude of structural breaks in the slope coefficients, illustrating the difficulty that the band market participants have faced in correctly pricing these instruments. For instance, time invariant VAR models would not be an appropriate choice for asset pricing in this context, nor would standard factor models of the yield curve. Future research should focus on an analysis of potential spillover effects between countries, rather than simply the CDS versus the bond market for individual countries. We believe that the commonality in the time evolution of the coefficients for the various countries in the sample points towards this type of contagion effect.