No. 3 - Volumes of sovereign CDS

A financial economist, July 2015


Summary: The price charged on CDS, the spread, is usually understood by analysts and market participants as an indication of the risk perceived by markets. However, CDS can be very shallow with net CDS volumes representing typically less than 2% of government debt; moreover, CDS spreads refer to specific maturities (usually 5-year debt), which represent an even smaller fraction of the total outstanding debt. In this context, the evolution of CDS spreads should be taken with caution as they may be driven more by liquidity constraints than by movements in actual underlining risk perceptions.

This column reviews the volumes of CDS as an indicator of the risk perceived by investors given that it reflects their hedging behaviour. Contrary to what seems to be indicated by CDS spreads, net volumes tend to indicate that the risk perception has continuously declined for most countries since mid-2010 and particularly since mid-2012 (including for Spain, Greece, Ireland or Portugal). On the other hand, the risk perception, temporarily increased for the UK, France and Germany between 2010 and 2013 and declined thereafter.



Credit default swaps (CDS) were created to insured lenders against the default of their borrowers. The price charged on CDS, the spread, is usually understood by analysts and market participants as an indication of the risk perceived by markets. For instance, Chart 1 plots the CDS spreads for a selection of EU sovereign bonds. The evolution of CDS spreads seems to reflect the different episodes of markets turmoil observed throughout the crisis. E.g. the concerns about Greece in late 2009 and early 2010, the subsequent bailout of Greece, the concerns about other peripheral countries throughout 2011 and 2012 and the improvement in market sentiment thereafter.

Chart 1: CDS spreads, government bonds, 5 year senior, basis points

Source: Bloomberg.

However, CDS spreads should not be interpreted in isolation as CDS markets can be very shallow and prices can been driven more by liquidity constraints than by actual movements in underlining risk perceptions. The volumes of CDS contracts amount to only a fraction of the total outstanding sovereign debt; moreover, CDS spreads refer to specific maturities (usually 5-year debt), which represent an even smaller fraction of the total outstanding debt. In this context, the volumes of CDS can be seen as a complementary measure of the risk perceived by investors as they reflect their hedging behaviour. Therefore, this column provides an analysis of CDS volumes for a selection of EU countries.

Gross sovereign CDS volumes have maintained an increasing trend in most countries between 2009 and 2013 and have stabilised or even declined thereafter (Chart 2). Exceptions to this pattern are Greek and Portuguese sovereign CDS, for which the stabilisation in gross volumes occurred already since early 2010. The declines observed in 2013-2015 are explained by the fact that CDS contracts have a typical duration of 5 years; therefore, the CDS contracts signed in the early stages of the crisis (2008-2010), particularly on the wake of the turbulence following the failure of Lehman Brothers in September 2008, expired throughout 2013-2015.

Chart 2: Gross notional volumes of CDS, sovereign bonds, all maturities, € billion

Notes: The spikes correspond to expiring contracts not being rolled-over. Greek CDS were triggered in March 2012 following the PSI debt exchange.
Source: DTCC and own calculations.

It is also important to note the relative importance across countries. The largest (gross) volumes concentrate in Italian sovereign bonds with a peak of over €350 billion; Spanish, French and German bonds follow with a volume of around €150 billion in the peak of 2013. On the other hand, gross volumes for countries such as Greece, Portugal or Ireland were much lower (around €50-60 billion in the peak). This picture from gross volumes contrasts with the one provided by the spreads (Chart 1).

Early cancellation of CDS contracts are usually costly so that investors willing to neutralise a position may prefer to enter in a new contract in the opposite direction to the contract to be untied¹. This dynamic translates into a smooth increase in gross volumes with sharp declines when the (gross) contracts reach actual maturity (as a convention, on a quarterly basis). This is indeed observed in the data (Chart 2).

Given this feature, CDS volumes in net terms provide a better picture of the amount of hedging sought by markets participants as they correspond to the actual open positions. Data on net volumes show a different profile (Chart 3) than the one observed for gross volumes. Net volumes have declined for most countries since mid-2010 and particularly since mid-2012 (this trend is also observed for net volumes of CDS for banks, see the Annex). Therefore, investors seem to have sought less hedging of debt and therefore perceived a declining risk already since 2010. This trend has temporarily reversed in late 2014, particularly for Italy, France and Germany.

However, France and Germany (and to a certain extent also the UK) represent an exception to the general trend. Net volumes for French and German bonds kept increasing until mid-2012. This may be reflecting the burden on these two countries perceived by markets due to the bailout of countries like Greece, Ireland and Portugal. The reversal in mid-2012 coincided with the agreement for the creation of the banking union and of the European Stability Mechanism (ESM); two factors which would imply a certain mutualisation of the burden across the euro area². As it has been observed for gross volumes, net volumes of CDS contracts are significantly larger for Italy, France or Germany than for Greece, Ireland or Portugal.

Having said that, the volumes of CDS contracts (e.g. between €5 and €20 billion for the large countries) represent a very limited proportion of the overall volume of government debt outstanding (e.g. over €2,000 billion in countries like Germany, France, the UK and Italy). See below for further details.

Chart 3: Net notional volumes of CDS, sovereign bonds, all maturities, € billion

Notes: Greek CDS were triggered in March 2012 following the PSI debt exchange.
Source: DTCC and own calculations.

The number of outstanding contracts (Chart 4) has followed a similar evolution to the one observed for gross volumes: an increase in the number of contract is observed across countries between 2009 and 2013 and a stabilisation and decline thereafter. As it was noted for the series of gross volumes, the evolution of CDS contracts present some "cliffs" corresponding to the (quarterly) expiration of contracts.

Chart 4: CDS contracts, all maturities, number of contracts

Notes: The spikes correspond to expiring contracts not being rolled-over. Greek CDS were triggered in March 2012 following the PSI debt exchange.
Source: DTCC and own calculations.

Both in terms of volume and number of contracts, the large majority of both buyers and sellers of euro area sovereign CDSs were entities from outside the euro area (mainly from the US). Although the euro area counterparts have entered the CDS markets, they remain a relatively small player (Chart 5).

Chart 5: Geographical distribution of buyers (B) and sellers (S), Euro area sovereigns
Gross volumes, € billion                                         Contracts, thousands

Notes: The spikes correspond to expiring contracts not being rolled-over. Greek CDS were triggered in March 2012 following the PSI debt exchange.
Source: DTCC and own calculations.

For most countries, CDS contracts have an average size between €5 and €20 million (Chart 6), but tend to be larger in larger countries with higher levels of debt in absolute terms (e.g. Italy, Germany, France or the US) than in smaller countries (e.g. Greece, Estonia, Lithuania, Latvia or Croatia). As a general rule, the average size of contracts have declined between 2008 and 2013 and increased again between 2013 and 2015. This evolution may also be signalling movements in risk perceptions.

Chart 6: Average size of CDS contracts , gross volumes, € million

Notes: Data as of December of each year; for 2015 data as of July 2015.
Source: DTCC and own calculations.

In general, a very limited proportion of debt is covered by CDS contracts (Chart 7). By 2015, net CDS volumes represented less than 2% of government debt and gross CDS volumes less than 20% of government debt in the majority of countries. Across the board, a decline in both gross and net volumes is observed over time due to the double effect of declining CDS volumes (numerator effect) and increasing debt levels (denominator effect).

A few exceptions can be observed in some countries (e.g. Bulgaria, Latvia, Estonia, Lithuania or Romania), where CDS volumes may have represented a significant proportion of overall debt. All these cases correspond with small countries with rather illiquid sovereign markets and currencies, which may explain a relatively larger use of CDSs for hedging than in other deeper markets. However, given the size of the overall debt of these countries, they do not seem to pose a systemic risk. Furthermore, the proportion of CDS volumes with respect to debt has significantly declined over time including in these countries (mainly due to a significant increase in the issuance of debt, without a corresponding increase in CDS contracts).

Chart 7: Ratio of CDS volumes to government debt , percentage
Gross CDS volumes to government debt

Net CDS volumes to government debt

Notes: Data as of December of each year; for 2015 data as of July 2015.
Source: DTCC, ECB and own calculations.

Annex.- Additional charts: data for main euro area banks

Chart A1: CDS volumes , bank bonds, all maturities, € billion
Gross notional volumes


Net notional volumes

Source: DTCC and own calculations.

Chart A2: CDS contracts, bank bonds, all maturities, number of contracts

Source: DTCC and own calculations.

Chart A3: Average size of CDS contracts, bank bonds, gross volumes, € million

Notes: Data as of December of each year; for 2015 data as of July 2015.
Source: DTCC and own calculations.




Notes:
¹To understand the difference between gross and net volumes, one could think of a contract for renting an apartment. Imaging that you rent an apartment for 3 years for €1,000 a month, with a guaranteed of €2,000 deposited upfront. If, after one year, you would need to move out (e.g. because your family has increased or because you need to move to another city) and you decide to cancel your contract, you would lose your €2,000 deposit. Alternatively, you can find somebody else to sublet the apartment. The new tenant will pay you the rent that you will pass on to the landlord. After some time, this new person may be in a similar situation and so on… If this situation is repeated say four times, the "gross" rent for the apartment would be €5,000 a month; however, it corresponds to a single apartment with €1,000 a month of actual "net" rent.
²It also corresponds to the ECB president Mario Draghi's commitment to undertake "whatever it takes" to preserve the unity of the euro. However, we consider that the agreement on the ESM and the BU were more relevant.