29
(because no protection seller would accept a negative premium). In order to profit from such
positive basis situation, market participants should short the bonds and sell CDS, as
previously described. However, it may be difficult or costly to borrow the underlying bonds
because the demand to borrow highly-rated and liquid bonds may exceed the supply by those
who owe such bonds that may be inhibited from legal or institutional requirements from
supplying collateral
21
. These kinds of frictions will cause the repo rate to go below LIBOR
(so called “special” repo rate) and the arbitrage may not generate a positive return even if the
basis is positive.
Other kinds of frictions may instead explain negative basis situations. In such situations the
arbitrage in Figure 10 would require to get LIBOR-funding using the long position in the
bond as collateral. However, there might be situations of turbulence in the interbank lending
market in which market participants are unwilling to provide liquidity or require a rate higher
than LIBOR (even for high-quality collateral). A similar situation may also arise if the bonds
are perceived as very low-quality collateral. Hence, in such situations the arbitrage is not
attractive since the extra cost of the repo financing may exceed the profit form the negative
basis.
Another important determinant of the basis is related to counterparty risk, because it makes
arbitrage not totally riskless.
In principle, the protection seller's counterparty risk may be quite limited, because if the
buyer defaults or misses a premium payment the obligation is extinguished (so called “default
termination” or “close-out”), though he may lose a positive market value if the credit quality
of the reference entity has improved
22
. Following a credit event, the buyer of protection is
instead exposed to the difference between the nominal and the recovery value of the defaulted
bonds, should the protection seller default following the reference entity's credit event. Thus,
the protection buyer may ask for adequate collateral (see, however, the discussion in §3.3
highlighting that the share of uncollateralized CDS trades may be high).
If we assume that protection buyers remain exposed to significant counterparty risk, the
return from the negative basis the arbitrage may not be sufficient to compensate for the
counterparty risk in the CDS transaction. Hence, other things being equal, counterparty risk
may explain a negative basis situation. Moreover, since risk premia may vary over time with
general market conditions, counterparty risk may have a differential impact on the basis
depending on market situations.
Most of the recent research confirms the arguments previously discussed that the basis is
affected by counterparty risk, imperfections that make funding or short selling impossible or
very costly (funding risk). For example, Fontana and Scheicher (2010) show that the basis for
sovereign entities is affected by the cost of shorting bonds and by other country specific and
21
See Duffie (1996) for a thorough discussion of this point and for an illustration of why this is more
likely for liquid and on-the-run US Treasuries bonds or liquid highly-rated bonds.
22
To be more precise, the ISDA Master Agreements require the parties to elect between the "First
Method" of calculating termination payments and the "Second Method". Under the First Method, in the
case of default of one of the two parties, if the market value is positive for the non-defaulting party, then
it is paid by the defaulting party to the non-defaulting party, but, if it is negative, then no payment is due
(i.e. the non-defaulting party is not required to make a termination payment to the defaulting party after
an event of default). Under the Second Method (which is the market standard), if the market value is a
positive for the non-defaulting party, then the defaulting party will pay it to the non-defaulting party, but
if it is a negative, then the non-defaulting party will make a payment to the defaulting party.