2018, Gyntelberg, Jacob, Hördahl, Peter, Ters, Kristyna, Urban, Jörg

The sovereign debt crisis in the euro area saw credit spreads on sovereign bonds and credit default swaps (CDS) surge for a number of member states. The rise in sovereign yields was accompanied by a significant increase in sovereign CDS market activity. This pattern raised concerns that destabilising speculation via outright short-selling of CDS (so-called ‘naked CDS’) was behind the increase in bond yields. In response, policy-makers introduced a ban on naked CDS trading. We investigate the effect of the ban on the price discovery process of sovereign credit risk, contrasting results for the post-ban period with those obtained prior to the ban. We use intraday data on sovereign CDS and bonds across a number of euro area countries. Our first main finding is that the CDS market dominates the bond market in terms of price discovery. That is, CDS premia in most cases adjust quicker to reflect new information than bond spreads. This result holds also when taking into account transaction costs. Our second main finding is that the ban on short-selling did not alter price discovery dynamics or reduce the efficiency of the market. Finally, we find that prior to the ban, CDS spreads were persistently higher than bond credit spreads, even after controlling for transaction costs. This points to the presence of market frictions that limit the ability of arbitrage forces to fully close pricing gaps between the two markets. However, these pricing discrepancies were in many cases largely eliminated following the introduction of the ban.

2017-04-26, Ters, Kristyna, Gyntelberg, Jacob, Hoerdahl, Peter, Urban, Jörg

We find evidence that in the market for euro area sovereign credit risk, arbitrageurs engage in basis trades between credit default swap (CDS) and bond markets only when the CDS-bond basis exceeds a certain threshold. This threshold effect is likely to reflect costs that arbitrageurs face when implementing trading strategies, including transaction costs and costs associated with committing balance sheet space for such trades. Using a threshold vector error correction model, we endogenously estimate these unknown trading costs for basis trades in the market for euro area sovereign debt. During the euro sovereign credit crisis, we find very high transaction costs of around 190 basis points, compared to around 80 basis points before the crisis. Our results show, that even when markets in times of stress are liquid, the basis can widen as high market volatility makes arbitrage trades riskier, leading arbitrageurs to demand a higher compensation for increased risk. Our findings help explain the persistent non-zero CDS-bond basis in euro area sovereign debt markets and its increase during the last sovereign crisis.

2013-09, Ters, Kristyna, Hoerdahl, Peter, Gyntelberg, Jacob, Urban, Jörg

The recent sovereign debt crisis in the euro area has seen credit spreads on sovereign bonds and credit default swaps (CDS) surge for a number of member states. While these events have increased interest in understanding the dynamics of sovereign spreads in bond and CDS markets, there is little agreement in the literature as to whether one of the two markets is more important than the other in terms of price discovery of sovereign credit risk. In this paper we reexamine this issue using intraday data for both market segments and employing carefully constructed cash (bond) spreads to ensure proper comparability with CDS spreads. This enables us to obtain much sharper estimates in our empirical analysis, and hence substantially clearer results with respect to price discovery. We find that the pricing of sovereign credit risk in the bond and in the CDS market converges over time, and that deviations between the two market segments do not persist for long. A key result is that the CDS market dominates the bond market in terms of price discovery in the vast majority of cases we examine. That is, CDS premia in many cases adjust more quickly to reflect new information than bonds spreads. This result holds also when taking into account transaction costs in the analysis.